September 8, 2022 17:1 JOIN S0219265921460075 page 1

Journal of Interconnection Networks

Vol. 22, Supp. 4 (2022) 2146007 (22 pages)
© World Scientific Publishing Company
DOI: 10.1142/S0219265921460075

Intelligent Building with Multi-Energy System

by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

Planning Method Considering Energy

Supply Reliability

Guanghua Qiao
College of Civil Engineering and
Architecture, Zhengzhou Shengda University
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

Zhengzhou 451191, P. R. China

[email protected]

Received 14 July 2021

Accepted 13 September 2021

Energy consumption is important to consume less power, reducing toxic fumes released
by plants, preserving natural resources, and protecting ecosystems against damage. The
challenging characteristics in energy supply include lack of renewable energy adoption,
and policy and energy management are considered essential factors. An artificial intel-
ligent building with a multi-energy planning method (AIBMEM) has been proposed to
design multi-energy systems to achieve the best policy and energy management techniques.
The intelligent construction problem with multi-energy is framed as a predictive energy
model to minimize the overall utilization of energy levels. The normal distribution with
the artificial intelligent model is introduced to solve the problem of renewable energy.
The experimental results based on reliability, effectiveness, preservation, energy consump-
tion, and control systems show that the suggested model is better than existing models,
producing good performance analysis results.

Keywords: Energy; system; multi-energy system; intelligent building; reliability.

1. Introduction
It is estimated that the building sector uses around 31% of the total global energy
consumption. This rate is constantly increasing with the changes in the global cli-
mate, environmental factors, urbanization, etc., [1]. This high level of energy con-
sumption by the building sector poses various challenges to the world’s sustainable
development. Thereby, eco-friendly smart buildings with minimal energy consump-
tion and maximization of renewable resources have become a trending research
topic [2].
The amount of energy consumed in buildings is determined based on various
factors [3]. The commonly influencing factors include the ambient temperature level
in the atmosphere, the type of devices available inside the building, the type of
appliances used, the activities performed inside the building, etc., [4]. Among all

2146007-1
September 8, 2022 17:1 JOIN S0219265921460075 page 2

G. Qiao

these factors, ambient temperature plays a prominent role in determining energy

consumption. This significance is because if the temperature is very low, devices like
heaters are used. Similarly, if the temperature is very high, devices like air coolers
are popularly used [5].
These devices consume more amount of energy. In the United States, energy
consumed by the cooling and heating devices amounts to around 50% of the total
energy consumption of the buildings [6]. It has been found that a rise in ambient
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

temperature is directly proportional to the increase in power consumption [7].

Human activities like urbanization, deforestation, etc., have a more significant
impact on the rise in ambient temperature, leading to an increase in energy con-
sumption [8]. To avoid this issue, renewable resources like solar energy, wind energy,
and tidal energy are getting popular in recent years [9]. These sources of energy are
eco-friendly to the environment and also cheap of cost. Further, to decrease energy
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

wastage, many smart devices are incorporated in smart, intelligent buildings [10].
These devices sense the atmospheric parameters and help automate the home
appliances to minimize the wastage of energy [11]. Energy modeling programs are
used to simulate energy usage to make energy predictions [12]. These predictions help
to automate the devices at home in an unmanned manner. For instance, cooling and
heating devices can be automatically switched off in humans’ absence [13]. This pro-
cess can be done using sensors that can detect the presence of humans. In this way,
energy wastage can be minimized to a large extent [14]. The simulation programs
are designed to represent the physical parameters [15] virtually. Modeling techniques
such as deep learning, artificial intelligence, machine learning can be used for cre-
ating programs that can be used for prediction and classification purposes. These
classifications can aid in creating real-time automatic smart buildings [16, 17]. The
world’s population growing and developing countries becoming more industrialized,
humanity’s thirst for energy has reached new heights. More than half of the energy
we use is derived from fossil fuels found deep within the Earth’s crust. We have
sucked up more than 135 billion tones of crude oil since commercial oil drilling began
in the 1850s to power our cars, fuel our power plants, and heat our homes. Every
day, that number rises. A new artificial intelligence-based model was constructed
based on training data collected from numerous buildings in the proposed scheme.
The energy estimation was done using the trained model. The estimated energy was
verified and employed in the conservation and efficient use of energy resources.
Excessive energy consumption in the building sector due to changes in ambi-
ent temperature is a major issue faced in recent years. To avoid this issue, in this
research, we propose a new model called Artificial intelligent building multi-energy
planning method (AIBMEM) for reducing the amount of energy consumption and
improving the utilization of renewable resources. Section 2 represents the related
works, and Sec. 3 gives the paper contributions. The proposed work is discussed in
Sec. 4, and the results are evaluated in Sec. 5. The conclusion, along with the future
scope, is presented in Sec. 6. Finally, the references are provided.

2146007-2
September 8, 2022 17:1 JOIN S0219265921460075 page 3

Intelligent Building with Multi-Energy System Planning Method

Based on training data collected from several buildings in the proposed scheme, a
new artificial intelligence-based model is being developed. The trained model is used
to estimate the energy. The estimated energy is validated and put to good use in
terms of energy conservation and efficiency. In addition, the suggested methodology
collects real-time data using renewable energy and a mix of IoT sensors.

2. Related Work
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

This research performed a comprehensive literature study on building planning and

management under energy efficiency significance. Among the many, this section
reports the most significant works that lead to the research formulation, as follows:
Liu et al. [18] proposed a system for energy planning based on the usage of renew-
able devices. These devices were used in the high residential buildings. Two types of
storage devices, namely the battery and hydrogen vehicle device, were integrated to
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

achieve energy planning. This scheme used multi-objective optimization. The draw-
back of this scheme was the usage of four different decision-making strategies for
energy planning.
Ezbakhe et al. [19] designed a system in which decision analysis was performed
to sustain energy resources. The renewable energy planning was done based on
the uncertainty principle. The main advantage of this scheme was the utilization
of technological and technical aspects in planning. However, this scheme failed to
address the issues faced due to cost considerations.
Zhang et al. [20] used an evolutionary neural network for energy planning based
on decomposition theory. Energy forecasting was done using six different decomposi-
tion networks. The performance of the system was improved based on the humidity
ratio and climatic information. This scheme was not capable of forecasting the load
efficiency accurately with precision.
Ahmadi et al. [21] proposed a model for long-term energy planning based on the
pollution reduction criterion. A new optimization model was proposed to achieve a
sustainable environmental goal. The main pillar of this model was the usage of an
aggregated algorithm for energy estimation. However, this scheme failed to forecast
the future energy requirements based on the past datasets.
Ahmad et al. [22] presented a model for energy management based on the
deep regression model. This scheme was used in demand planning based using
real-time load data. Tree-based ensemble models were used for the classification of
energy requirements. The forecasting was done based on two different climatic zone
datasets. This system failed to address the non-linear characteristics of the climatic
load data.
Stermieri et al. [23] created a building simulation model for the long-term plan-
ning of energy requirements using planning strategies. The main objective of this
framework was to improve the energy-saving task in urban regions. A new simula-
tion model was proposed based on the energy demand. The only drawback of this
scheme was the usage of post-retrofit energy demand in the energy calculation.

2146007-3
September 8, 2022 17:1 JOIN S0219265921460075 page 4

G. Qiao

Moghadam et al. [24] presented a system for energy retrofitting based on building
stock data. This scheme was built to support urban energy planning. The spatial
energy support system was used in this framework based on the multi-support cri-
terion. Here, building stocks were integrated to support complex challenges. The
main outcome of this research was the decision support system to support decision-
making.
Zhong et al. [25] proposed a model for predicting energy consumption using
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

a vector field support system. This system was used in the creation of prediction
models based on linearity theory. Dataset from Chinese buildings was used in this
modeling. The generalization capability was the main advantage of this framework.
However, the computational complexity was the main drawback of this scheme.
Pasichnyi et al. [26] presented a strategic planning model based on energy
retrofitting. Aggregated projections were used in the social investment planning.
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

Energy-efficient windows were used for the creation of retrofitting packages. The
total energy demand was estimated for the large-scale analysis in this research.
Ahmed BilalAwan et al. [27] proposed a detailed comparative analysis for the
two solar technologies namely: PV and dish stirling engine. The two technologies
are analyzed based on several performance indicator factors such as capacity factor,
solar to electric efficiency, energy output, levelized electricity cost and net present
value.
Iskander Tlili et al. [28] presented a mathematical model is realistic to assess
a comparative study of nanofluid and hybrid nanofluid flow between two eccentric
pipes. Study of nanofluid has been developed recurrently over the earlier era.
IskanderTlili [29] proposed in this paper, the hydromagnetic dissipative
Newtonian/non-Newtonian fluid flow over a stretched surface is examined numeri-
cally. Brownian moment, Joule heating, thermophoresis effects are considered.
Aldabesh et al. [30] presented, the asymmetric analysis for the Casson nano-
material has been analyzed due to two parallel stretchable disks. The novel ther-
mophoresis and Brownian aspects of nanofluid are studied by using Buongiorno
nanofluid model.
The above literature survey indicates the drawbacks in the domain of energy
management in intelligent buildings. To address these issues, in this research, a
new model based on artificial intelligence is proposed to enable multi-energy system
planning based on supply reliability. The advantage of using artificial intelligence
is the high reliability offered by these models in predicting energy requirements in
intelligent building sectors.

3. Contribution of the Paper

The contributions of this paper are as follows

• A novel technique called Artificial intelligent building multi-energy planning

method (AIBMEM) is proposed for energy planning in intelligent buildings.

2146007-4
September 8, 2022 17:1 JOIN S0219265921460075 page 5

Intelligent Building with Multi-Energy System Planning Method

• The various design criteria for intelligent building are presented.

• Layers of artificial intelligent modeling in smart buildings are presented.
• A new energy prediction model is proposed for predicting energy usage.
• The proposed model is validated using metrics like reliability rate, effectiveness
ratio, preservation rate, energy consumption rate, energy efficiency, prediction
rate, energy management ratio, and power conservation rate.
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

4. Proposed Methodology
4.1. Design criteria of intelligent building
The design criteria of intelligent buildings are based on various intelligent parameters
that are adopted in smart cities. These parameters are used for enhancing the user’s
experience based on modern technological advancements. A holistic approach to
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

both modeling and optimization is required to efficiently manage a multi-vector

district. Major factors of system planning methods are management and executive
time, commitment towards the work, estimation, research and development, review.
Figure 1 shows the various design criteria considered while constructing intelli-
gent buildings. Heating, ventilation, and air conditioning account for 35% of total
building energy consumption; lighting accounts for 11%; major appliances such
as water heating, refrigerators and freezers account for 18%, and the remaining
36% is spent on other items such as electronics. In multi energy system temper-
ature rising issue can avoid with following plans like reduce energy consumption
through building design and layout, Plant shade trees, Minimize artificial lighting
and Make use of skylight windows. Also use Air Compressors and Install Motion sen-
sitive light switches. These include automation systems, renewable resources, smart
sensors, connectivity, machine learning, policy management, energy management,

Fig. 1. Design criteria of intelligent building.

Figure 1.Design criteria of intelligent building
2146007-5
Figure 1 shows the various design criteria considered while constructing intelligent buildings. Heating, ventilation, and
air conditioning account for 35% of total building energy consumption; lighting accounts for 11%; major appliances such
as water heating, refrigerators and freezers account for 18%, and the remaining 36% is spent on other items such as
electronics. In multi energy system temperature rising issue can avoid with following plans like reduce energy
consumption through building design and layout, Plant shade trees, Minimize artificial lighting and Make use of skylight
windows. Also use Air Compressors and Install Motion sensitive light switches. These include automation systems,
September 8, 2022 17:1 JOIN S0219265921460075 page 6

G. Qiao

energy prediction, artificial intelligence, eco-friendly, smart material, and reliability.

Automation is important in the intelligent building to decrease the human load.
Renewable resources are adopted to create an eco-friendly and healthy environment.
Smart sensors are used for sensing various physical parameters that can be used in
automation. Connectivity is an important parameter as it helps to integrate with
IoT devices. Machine learning algorithms are used for various classification and
prediction purposes during automation. Smart homes conserve energy by tracking
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

which rooms are used and when they are used. The system will keep track of the
rooms you are in and ensure that the temperature in those rooms is warm enough
when you are there. Your home will no longer heat itself inefficiently while you are
away. Building energy models serve as a simplified and efficient model for predicting
future energy consumption in buildings. As tools to aid in the management of data
and information related to energy systems develop, the use of emerging technolo-
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

gies such as cloud computing, Internet of Things, and Big Data is increasing. This
enables for more flexibility, scalability of solutions, energy efficiency improvement,
and energy device control.
Policy and energy management have been done for regulating the amount of
energy usage in smart buildings. Artificial intelligence algorithms are used for
improving the user experience through advanced prediction systems. Smart mate-
rials are used for improving the quality of smart living. Reliability is an important
criterion as it helps to enhance the user experience. Among various criteria discussed
above, the usage of artificial intelligence plays a vital role and is discussed in the
following section.

4.2. Layers of artificial intelligence modeling in smart buildings

The layers of artificial intelligence modeling in smart buildings comprise three main
layers. These layers include the bottommost sensing layer, middle communication
layer, and the topmost control layer. For energy conversion and storage, the util-
isation of different energy carriers and distributed energy supplies is highlighted.
Two key factors underpin the visionary concept. The so-called Energy Hub concept
describes converters and storage devices. The second unique approach is an Energy
Interconnector device, which allows for the simultaneous conveyance of electrical,
chemical, and thermal energy in a single subsurface device. Energy Hubs, Energy
Interconnectors, and traditional elements can then be used to design the entire
energy system.
Figure 2 shows layers of artificial intelligence modeling in smart buildings. The
bottom sensing layer comprises various sensors like RFID, camera, wearable devices,
ambient sensors, etc. These sensors are used for the acquisition of physical param-
eters available in the buildings. These sensors include both indoor and outdoor
sensors. The camera sensor can be used for monitoring the surveillance. Wearable
devices are used for applications like fall detection of older adults. The Ambient sen-
sors can be used for home automation. The next layer comprises the communication

2146007-6
September 8, 2022 17:1 JOIN S0219265921460075 page 7

The layers of artificial intelligence modeling in smart buildings comprise three main layers. These layers include the
bottommost sensing layer, middle communication layer, and the topmost control layer. For energy conversion and
storage, the utilisation of different energy carriers and distributed energy supplies is highlighted. Two key factors
underpin the visionary concept. The so-called Energy Hub concept describes converters and storage devices. The second
unique approach is an Energy Interconnector device, which allows for the simultaneous conveyance of electrical,
chemical, and thermal energy in a single subsurface device. Energy Hubs, Energy Interconnectors, and traditional
Intelligent Building with Multi-Energy System Planning Method
elements can then be used to design the entire energy system.
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

Fig. 2. Figure 2. Layers

Layers of artificial
of artificial intelligence modeling
intelligence modeling in smart buildings
in smart buildings.
Figure 2 shows layers of artificial intelligence modeling in smart buildings. The bottom sensing layer comprises various
layer. Inlike
sensors this layer,
RFID, camera,the datadevices,
wearable acquired
ambient bysensors,
the sensors are transferred
etc. These sensors are used for theto the cloud
acquisition using
of physical
various communication
parameters mediums.
available in the buildings. This
These sensors communication
include medium
both indoor and outdoor sensors.includes
The camera Bluetooth,
sensor can be
Zigbee,
used for Wi-Fi,
monitoringand LTE. This
the surveillance. data
Wearable communication
devices to the
are used for applications likecloud is done
fall detection foradults.
of older storage
The
Ambient
and sensorsprocessing.
further can be used for home
Theautomation.
processed The next
datalayercan
comprises the communication
be used in achieving layer. Inautomation
this layer, the
data acquired
through by the sensors
various machineare transferred
learning to the cloud using various
algorithms. Thecommunication
control layer mediums. This communication
comprises energy
medium includes Bluetooth, Zigbee, Wi-Fi, and LTE. This data communication
management systems, predictive algorithms, cloud computing, and machine learning to the cloud is done for storage and
further processing. The processed data can be used in achieving automation through various machine learning
algorithms.
algorithms. The control layer comprises energy management systems, predictive algorithms, cloud computing, and
machine learning algorithms.
4.3. Energy prediction model to improve energy efficiency
4.3. Energy prediction model to improve energy efficiency
Energy prediction models play an essential role in the intelligent building as they
Energy prediction models play an essential role in the intelligent building as they help in energy conservation. This model
help in energy conservation. This model helps to reduce the cost of living and often
helps to reduce the cost of living and often helps to save energy for future usage.
helps to save energy for future usage.
Figure 3 presents the energy prediction model that can be used for improving
energy efficiency. This model comprises of sensor layer, energy prediction layer, and
controller unit. The sensor layer involves both indoor and outdoor sensors. The
indoor sensors include sensors used inside the building like fan, light, humidity, heat,
and smoke sensors. The outdoor sensors include a camera, vehicular sensors, door
lock sensors, etc. The values collected by these indoor and outdoor sensors are com-
municated to the cloud. The energy prediction layer comprises of data acquisition
unit, modeling unit, and the energy prediction unit. The acquisition unit collects the
data from indoor and outdoor sensors. The modeling unit is used to create a machine
learning or artificial intelligence model, and the energy prediction unit is used to pre-
dict the energy usage level. Based on the output of the prediction unit, the control
layer is used for controlling various devices. The control devices comprise switches
and relay units. Because they rely less on fossil fuels, energy efficient buildings emit
fewer greenhouse gases. Emissions are lowest in buildings that use predominantly

2146007-7
September 8, 2022 17:1 JOIN S0219265921460075 page 8

G. Qiao
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

Fig. 3. 3.
Figure Energy
Energyprediction model to
prediction model toimprove
improveenergy
energy efficiency.
efficiency

Figure 3 presents the energy prediction model that can be used for improving energy efficiency. This model comprises of
clean energy, such as hydroelectricity. Energy storage allocation should be possible
sensor layer, energy prediction layer, and controller unit. The sensor layer involves both indoor and outdoor sensors. The
at all points within a network, including transmission, distribution, and demand
indoor sensors include sensors used inside the building like fan, light, humidity, heat, and smoke sensors. The outdoor
side, in a proper model for future energy system planning. Identifying important
sensors include a camera, vehicular sensors, door lock sensors, etc. The values collected by these indoor and outdoor
time series information for use in energy system models, including demand, solar,
sensors
andare wind.
communicated to the cloud. The energy prediction layer comprises of data acquisition unit, modeling unit,
and the energy prediction unit. The acquisition unit collects the data from indoor and outdoor sensors. The modeling unit
is used to create a machine learning or artificial intelligence model, and the energy prediction unit is used to predict the
4.4. The architecture of the proposed Artificial Intelligent Building
energy usage level. Based on the output of the prediction unit, the control layer is used for controlling various devices.
Multi-Energy Planning Method (AIBMEM)
The control devices comprise switches and relay units. Because they rely less on fossil fuels, energy efficient buildings
emit The
fewer proposed
greenhouse AIBMEM model
gases. Emissions arecomprises trainingthat
lowest in buildings andusetesting units. The
predominantly clean training is as
energy, such
done based
hydroelectricity. on the
Energy training
storage building
allocation dataset
should be possiblethat
at allispoints
collected
within from multiple
a network, buildings.
including transmission,
Energy management is a method of regulating and lowering a building’s energy use,
distribution, and demand side, in a proper model for future energy system planning. Identifying important time series
allowing owners and operators to: Reduce costs - energy accounts for 25% of an
information for use in energy system models, including demand, solar, and wind.
office building’s total running expenditures. To achieve internal sustainability goals
and regulatory requirements, reduce carbon emissions. Testing is done using the test
building data.
4.4. The architecture of the proposed Artificial Intelligent Building Multi-Energy Planning Method (AIBMEM)
Figure 4 shows the architecture of the proposed Artificial intelligent Building
The proposed AIBMEM Planning
Multi-Energy model comprises training
Method and testing units.
(AIBMEM). In The
thistraining
model,is done based on
initially, thedata
the training building
from
the
dataset thattraining
is collectedbuilding dataset
from multiple is collected.
buildings. Features
Energy management are extracted
is a method from
of regulating and this dataset.
lowering a building's
The extracted features are used in parameter optimization. Using the optimized
energy use, allowing owners and operators to: Reduce costs - energy accounts for 25% of an office building's total running
parameters, the artificial intelligence model is generated. This AI model is used in
the testing step. During testing, features are extracted from the test data. Using the
AI-trained model and the extracted features, the prediction is made. The predicted

2146007-8
September 8, 2022 17:1 JOIN S0219265921460075 page 9

expenditures. To achieve internal sustainability goals and regulatory requirements, reduce carbon emissions. Testing is
done using the test building data. Intelligent Building with Multi-Energy System Planning Method
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

Fig. 4. The architecture of the proposed Artificial intelligent Building Multi-Energy Planning
Figure 4. The architecture
Method of the proposed Artificial intelligent Building Multi-Energy Planning Method (AIBMEM)
(AIBMEM).

Figure 4 shows the architecture of the proposed Artificial intelligent Building Multi-Energy Planning Method (AIBMEM).
In this model,energy value
initially, is used
the data fromin
theenergy optimization.
training The
building dataset proposedFeatures
is collected. AIBMEM methodfrom
are extracted clearly
this dataset.
achieves the maximum average efficiency. Other models, such as the sparse represen-
The extracted features are used in parameter optimization. Using the optimized parameters, the artificial intelligence
tation model (SRM), the k-nearest neighbour model (KNNM), the support vector
model is generated.
machine This AI model
model is used in
(SVMM), thethelogistic
testing regression
step. Duringmodel
testing,(LRM),
featuresthe
are extracted
non-linear from the test data.
regres-
Using the AI-trained
sion model model(NRM),
and the extracted
and the features,
Nave Bayes the prediction is made. The
model (NBM), predicted
achieve energy
average value
rates is used in
per-
centage.The
energy optimization. The suggested
proposed AIBMEMAIBMEMmethodstrategy achieves
clearly achieves the the
maximumbest value
average since it employs
efficiency. Other models,
the energy management model. An AI (artificial intelligence) model is a programme
such as the sparse representation model (SRM), the k-nearest neighbour model (KNNM), the support vector machine
that has been trained to recognise patterns using a set of data (called the training
model (SVMM), set).theTologistic
reason regression
over andmodel
learn(LRM), the non-linear
from this regression
data, AI models model a(NRM),
employ variety and
of the Nave Bayes model
methods,
(NBM), achievewithaverage rates percentage.
the overarching goalThe
ofsuggested
addressing AIBMEM
businessstrategy achieves The
challenges. the best value since
increasing it employs the
adop-
tion of computational modeling frameworks in these domains has
energy management model. An AI (artificial intelligence) model is a programme that has been trained to recognise been facilitated
by the use of modern technology in building management systems, such as sensors,
patterns using a set of data (called the training set). To reason over and learn from this data, AI models employ a variety
network connection, cloud computing, wireless transmission, information, and smart
of methods, with the overarching
devices. goal of addressing
The statistical methodologies business challenges.
used to the The increasing
modeling adoption ofproduce
frameworks computational
modeling frameworks
reliable and in these domains
precise has been facilitated
projections of consumer by theenergy
use of modern
demand, technology
which are in building
classedmanagement
as
systems, suchAI-based
as sensors,engineering and hybrid
network connection, approaches.
cloud computing, The impact
wireless of major
transmission, elements
information, and such
smart devices.
as building attributes and environmental circumstances is taken into account in a
The statistical methodologies used to the modeling frameworks produce reliable and precise projections of consumer
four-step process: (1) data collection, (2) data pre-processing, (3) model training,
energy demand,andwhich are classed
(4) model as AI-based
testing engineering
are all steps in theand hybrid
model approaches.
training The impact of major elements such as
process.
building attributes and environmental
Because circumstances
of its relevance is taken into
in decreasing energyaccount
waste,in a the
four-step process:
Building (1) data
Energy collection, (2)
Man-
agement
data pre-processing, (3) System (BEMS)
model training, andhas become
(4) model a hot
testing are topic
all steps in inrecent years.
the model However,
training process. due to
issues such as low forecast accuracy, one of BEMS’ applications, energy consumption
prediction, has remained static. The proposed AI algorithm is discussed below.
Because of Algorithm
its relevance1 inuses input energy
decreasing data waste,
like Produced
the Buildingenergy PE = [p11System
Energy Management , . . . , p(BEMS)
xy , . . . has
pXY ], Renewable energy RE = [r11 , . . . , rxy , . . . rXY ], Multi-energy function
become a hot topic in recent years. However, due to issues such as low forecast accuracy, one of BEMS' applications,
energy consumption prediction, has remained static. The2146007-9
proposed AI algorithm is discussed below.
September 8, 2022 17:1 JOIN S0219265921460075 page 10

G. Qiao

Algorithm 1. Optimized Adaptive Energy Efficient Strategy (OEES) algorithm

Input:
Produced energy PE = [p11 , . . . , pxy , . . . pXY ]; 1 ≤ x ≤ X, 1 ≤ y ≤ Y
Renewable energy RE = [r11 , . . . , rxy , . . . rXY ]; 1 ≤ x ≤ X, 1 ≤ y ≤ Y
Multi-energy function ME = [m11 , . . . , mxy , . . . mXY ]; 1 ≤ x ≤ X, 1 ≤ y ≤ Y
Toxic fumes TF = [t11 , . . . , txy , . . . tXY ]; 1 ≤ x ≤ X, 1 ≤ y ≤ Y
Intelligent function IF = [i11 , . . . , ixy , . . . iXY ]; 1 ≤ x ≤ X, 1 ≤ y ≤ Y
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

Required power RP = [rp11 , . . . , rpxy , . . . rpXY ]; 1 ≤ x ≤ X, 1 ≤ y ≤ Y

Usable power UP = [u11 , . . . , uxy , . . . uXY ]; 1 ≤ x ≤ X, 1 ≤ y ≤ Y
Predicted power PP = [pp11 , . . . , ppxy , . . . ppXY ]; 1 ≤ x ≤ X, 1 ≤ y ≤ Y
Output:
Energy minimization function EMF
Steps:
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

Initialize x variable x ← 0
Initialize y variable y ← 0
Set Maximum X range X ← 10000
Set Maximum Y range Y ← 10000
Set Maximum EC(f ) range ECmax (f ) ← 50000
for each x = 1 : X
for each y = 1 : Y
Initialize the value of RR(f )
Initialize the value of ER(f )
Initialize the value of P R(f )
Initialize the value of EC(f )
Calculate the values of RR(f ), ER(f ), P R(f ) and EC(f )
end for
end for
for each x = 1 : X
for each y = 1 : Y
if EC(f ) > ECmax (f )
Initialize the value of CR(f )
Initialize the value of EE(f )
Initialize the value of P RA(f )
Initialize the value of EM (f )
Initialize the value of P C(f )
Calculate the values of CR(f ), EE(f ), P RA(f ), EM (f ) and P C(f )
end if
end for
end for
Calculate Energy MinimizationFunction EM F
Return EMF

2146007-10
September 8, 2022 17:1 JOIN S0219265921460075 page 11

Intelligent Building with Multi-Energy System Planning Method

ME = [m11 , . . . , mxy , . . . mXY ], Toxic fumes TF = [t11 , . . . , txy , . . . tXY ], Intelligent

function IF = [i11 , . . . , ixy , . . . iXY ], Required power RP = [rp11 , . . . , rpxy , . . . rpXY ],
Usable power UP = [u11 , . . . , uxy , . . . uXY ], and Predicted power PP =
[pp11 , . . . , ppxy , . . . ppXY ]. In this algorithm, the energy consumption value is used
as an upper bound. For each two-dimensional value, the value of reliability rate,
effectiveness ratio, preservation rate, and energy consumption rate are calculated.
If the amount of energy consumption exceeds the predefined threshold, then the
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

control rate, prediction rate, energy management ratio, and power conservation rate
are computed.
The reliability rate is computed as
X  R 
X rxy
RR(f ) = pxy + + ∂mxy − txy . (1)
x∗y
x=1
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

In Eq. (1), RR(f ) is the reliability rate, pxy is the produced energy variable, rxy is
the renewable energy variable, mxy is the multi-energy function variable, txy is the
toxic fume variable, ∂mxy is the rate of change of multi-energy function, and x ∗ y
is the product of two-dimensional variables. The reliability rate gives the rate at
which the proposed scheme is reliable with effective energy consumption.
Effectiveness rate is computed as

X Y
s 2
Y rxy X mxy RR(f )
ER(f ) = RR(f ) ∗ ∗ ∗ . (2)
mxy txy ixy
x=1 y=1 2

In Eq. (2), ER(f ) is the effectiveness rate, RR(f ) is the reliability rate, rxy is the
renewable energy variable, mxy is the multi-energy function variable, txy is the
mxy
toxic fume variable, ixy is the intelligent function, txy gives the division of multi-
q
RR(f )
energy function and the toxic fume function, and ixy gives the root of the
renewable energy variable and the intelligent function. The effectiveness rate is the
function that shows the effective rate at which the proposed system makes the energy
prediction.
Preservation rate is given by
v
u Y X
uX X p
P R(f ) = t exp(ER(f ) ± mxy + txy + ixy − rpxy ). (3)
y=1 x=1

In Eq. (3), P R(f ) is the preservation rate, ER(f ) is the effectiveness rate, mxy is the
multi-energy function variable, txy is the toxic fume variable, ixy is the intelligent
√
function, rpxy gives the requested power function, mxy + txy gives the root of the
sum of multi-energy function variable and the toxic fume variable. The preservation
rate gives the rate at which the renewable resources are preserved in the proposed
intelligent building scheme.

2146007-11
September 8, 2022 17:1 JOIN S0219265921460075 page 12

G. Qiao

Energy consumption ratio is defined as

X X
Y
X ∂txy + ∂ixy
EC(f ) = ± P R(f ) ∗ . (4)
∂rpxy + ∂uxy
x=1 y=1

In Eq. (4), EC(f ) is the energy consumption ratio, P R(f ) is the preservation rate,
txy is the toxic fume variable, ixy is the intelligent function, rpxy gives the requested
power function, uxy is the usable power function, ∂txy is the rate of change of
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

toxic fume variable, and ∂ixy is the rate of change of an intelligent function. The
energy consumption ratio gives the ratio at which the proposed scheme consumes
the available energy.
Control rate is computed as
X
1 1 X
CR(f ) = ∗ ∗ EC(f ) ∗ [ixy ∗ rpxy ∗ uxy ∗ ppxy ]. (5)
x−1 y−1
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

x=1
In Eq. (5), CR(f ) is the control rate, EC(f ) is the energy consumption ratio, ixy is
the intelligent function, rpxy gives the requested power function, uxy is the usable
power function, ppxy is the predicted power function and ixy ∗ rpxy ∗ uxy ∗ ppxy
gives the product of the intelligent function, requested power function, usable power
function, and the predicted power variable.
The control rate gives the rate at which the proposed model is capable of con-
trolling the energy parameters. Figure 5 gives the path diagram of control rate
computation. It is obvious that the control rate is directly proportional to the energy
consumption ratio.
Energy efficiency is given by
v
uX Y  
ixy * rpxy * u xy * ppxy gives the product uX X θ ∗ (CR(f ) − rpxy − uxy )
EE(f ) = t intelligent function, requested power function,
of the
. usable power function, and
(6)
(CR(f ) − ppxy − pxy )
the predicted power variable. x=1 y=1

Fig. 5. 5. Path
Figure Path diagram
diagramofofcontrol
controlrate.
rate

The control rate gives the rate at which the proposed 2146007-12
model is capable of controlling the energy parameters. Figure 5
gives the path diagram of control rate computation. It is obvious that the control rate is directly proportional to the energy
consumption ratio.

Energy efficiency is given by

X Y  *(CR( f )  rpxy  u xy )
September 8, 2022 17:1 JOIN S0219265921460075 page 13

Intelligent Building with Multi-Energy System Planning Method

As seen in Eq. (6), EE(f ) is the energy efficiency, CR(f ) is the control rate, θ is
the angular change in the control rate, rpxy gives the requested power function,
uxy is the usable power function, ppxy is the predicted power function, and pxy
is the produced energy variable. The energy efficiency gives the overall efficiency
of the proposed artificial intelligence-based model for utilizing the available energy
efficiently.
Prediction rate is computed as
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

H H
EE(f ) ∗ uxy ∗ ppxy
P RA(f ) = arg min H H . (7)
xy pxy ∗ rxy
In Eq. (7), P RA(f ) is the prediction rate, EE(f ) is the energy efficiency, uxy is the
usable power function, ppxy is the predicted power function,
H pxy is the produced
energy variable, rxy is theH renewable energy variable, uxy gives the total integral
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

usable power value, and ppxy is the total integral predicted power function. The
prediction rate is the rate at which the system is capable of predicting the energy
requirement effectively.
Energy management ratio is calculated as
X
X |ppxy − pxy |
EM (f ) = P RA(f ) − . (8)
|rxy − mxy |
x=1

As seen in Eq. (8), EM (f ) is the energy management ratio, P RA(f ) is the prediction
rate, ppxy is the predicted power function, pxy is the produced energy variable, rxy
is the renewable energy variable, and mxy is the multi-energy function variable. The
energy management ratio gives the ratio of the energy management level to the
available energy level.
Power conservation rate is given by
P C(f ) = kRR(f ) + ER(f ) + P R(f ) − EC(f )k22 . (9)
As seen in Eq. (9), P C(f ) is the power conservation rate, RR(f ) is the reliability
rate, P R(f ) is the preservation rate, ER(f ) is the effectiveness rate, and EC(f )
is the energy consumption ratio. The power conservation rate is computed as the
l–2 norm of the sum of power conservation rate, reliability rate, and effectiveness
rate subtracted by the energy consumption ratio. This value gives the overall energy
conserved in the proposed AIBMEM scheme.
Energy Minimization FunctionEM F can be computed using the following
equation
H H H 2
( P C(f ) + CR(f ) + EE(f ))
EM F arg min H H subject to f ≥ 0. (10)
f P RA(f ) − EM (f ) 2

In Eq. (10), EM F is the Energy Minimization Function, CR(f ) is the control rate,
EC(f ) is the energy consumption ratio, P C(f ) is the power conservation rate,
P RA(f ) is the prediction rate, EM (f ) is the energy management ratio, and EE(f )

2146007-13
consumption ratio, PC ( f ) is the power conservation rate, PRA( f ) is the prediction rate, EM ( f ) is the energy
September 8, 2022 17:1 JOIN S0219265921460075 page 14
management ratio,and EE ( f ) is the energy efficiency. The Energy Minimization Function is the optimization function at

which the energy consumption in the intelligent building can be minimized. Different sorts of data are generated through
quantitative and qualitative methods. Qualitative data is expressed in words, while quantitative data is expressed in
numbers. Mixed methods evaluations have become more popular as a result of this. Different sorts of data are generated
through quantitative and qualitative methods.
G. Qiao
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

Fig. 6. Path diagram of energy minimization function.

is the energy efficiency. The Energy Minimization Function is the optimization func-
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

tion at which the energy consumption in the intelligent building can be minimized.
Different sorts of data are generated through quantitative and qualitative meth-
ods. Qualitative data is expressed in words, while quantitative data is expressed in
numbers. Mixed methods evaluations have become more popular as a result of this.
Different sorts of data are generated through quantitative and qualitative methods.
Figure 6 shows the path diagram of the energy minimization function. In this dia-
gram, EM F is the Energy Minimization Function, CR(f ) is the control rate, EC(f )
is the energy consumption ratio, P C(f ) is the power conservation rate, P RA(f )
is the prediction rate, EM (f ) is the energy management ratio, and EE(f ) is the
energy efficiency. The Energy Minimization Function is the optimization function at
which the energy consumption in an intelligent building can be minimized. Preser-
vation of capital is a conservative investment approach that focuses on preserving
wealth and avoiding portfolio losses. The safest short-term securities, such as Trea-
sury bills and certificates of deposit, must be invested in this approach. The energy
use index, which is defined as energy use per unit conditioned floor space, is a widely
used tool for assessing building energy performance. Energy consumption is a com-
mon metric for assessing a building’s environmental impact. Recent research has
emphasized the importance of both operational and embodied energy in buildings
over their lifetime. Life-cycle energy analysis is a method for determining how much
energy a building will need during its lifetime.
In summary, this section reports the architecture of proposed Artificial Intelligent
Building Multi-Energy Planning Method (AIBMEM) designing methodologies, AI
algorithm and implementation method also described.

5. Results and Discussion

In this section, we have performed the performance analysis of the proposed scheme.
The proposed AIBMEM scheme was compared using various models like sparse
representation model (SRM), k-nearest neighbor model (KNNM), support vector

2146007-14
September 8, 2022 17:1 JOIN S0219265921460075 page 15

Intelligent Building with Multi-Energy System Planning Method

machine model (SVMM), logistic regression model (LRM), non-linear regression

model (NRM), and Naı̈ve Bayes model (NBM).

5.1. Performance analysis

For quantitative evaluation, we have employed metrics like reliability rate, effective-
ness ratio, preservation rate, energy consumption rate, energy efficiency, prediction
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

rate, energy management ratio, and power conservation rate.

Table 1 shows the comparison of energy efficiency for different test building
datasets. It is evident that the proposed AIBMEM scheme attains the highest aver-
age efficiency of 91.59%. Other schemes like sparse representation model (SRM),
k-nearest neighbor model (KNNM), support vector machine model (SVMM), logis-
tic regression model (LRM), non-linear regression model (NRM), and Naı̈ve Bayes
model (NBM) attains an average rate of 56.63%, 55.41%, 52.11%, 53.09%, 57.5%,
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

and 58.07% respectively. The highest value is attained by the proposed AIBMEM
scheme as it involves using the energy management model.
Table 2 shows the comparison of prediction rate for variation in the testing
dataset. It is observed that the overall prediction rate for sparse representation model
(SRM) is 46.39%, k-nearest neighbor model (KNNM) is 52.72%, support vector
machine model (SVMM) is 52.95%, logistic regression model (LRM) is 52.15%, non-
linear regression model (NRM) is 52.34%, and Naı̈ve Bayes model (NBM) is 46.51%.
The proposed AIBMEM model has a prediction rate of 80.93%. This high value
greater than 80% is due to the application of artificial intelligence in the modeling
of training data.
Figure 7 shows the variation of the reliability rate. The average reliability of the
proposed scheme is as high as 84.5%. This value is exceedingly high compared to
all other schemes as shown in Fig. 7. This extreme peak value is achieved since the
proposed AIBMEM model is based on minimizing energy usage and maximization
of usage of renewable resources.

Table 1. Energy efficiency (%).

Energy efficiency (%)

Building
test data SRM KNNM SVMM LRM NRM NBM AIBMEM

1 41.9 44.16 55.6 50.78 45.91 46.48 98.81

2 52.14 60.89 50.43 62.09 64.65 63.72 89.77
3 53.45 42.81 44.5 51.84 52.9 68.48 96.81
4 50.97 55.76 57.58 60.5 66.64 49.83 91.17
5 62.91 55.91 47.86 61.12 51.73 60.14 82.94
6 58.84 65.84 41.33 53.27 63.08 53.16 83.79
7 63.16 54.55 62.65 40.58 51.9 65.01 87.73
8 67.99 51.8 47.28 49.92 64.26 63.07 94.23
9 69.19 60.14 53.27 52.73 62.65 45.01 95.69
10 45.76 62.24 60.64 48.11 51.32 65.86 95.01

2146007-15
 September 8, 2022 17:1 JOIN S0219265921460075 page 16

G. Qiao

Table 2. Prediction rate (%).

Prediction rate (%)

Building
test data SRM KNNM SVMM LRM NRM NBM AIBMEM

1 46.37 41.1 57.31 56.91 59.65 44.19 75.96

2 50.68 57.02 52.25 54.78 55.38 51.05 74.25
3 41.79 51.21 59.8 51.72 51.63 52.6 75.48
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

4 42.23 58.6 50.55 44.93 58.57 40.64 71.23

5 42.72 53.94 49.59 53.33 51.6 52.3 88.42
6 53.57 51.66 56.03 41.67 40.33 47.25 78.17
7 49.9 56.31 44.55 52.52 42.41 40.99 85.62
8 43.79 57.58 49.96 53.22 57.26 49.79 90.16
9 49.9 59.78 58.02 54.6 49.69 43.85 84.47
10 42.95 40.01 51.49 57.82 56.9 42.46 85.54
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

SRM KNNM SVMM LRM NRM NBM AIBMEM

100
90
80
Reliability rate (%)

70
60
50
40
30
20
10
0
1 2 3 4 5 6 7 8 9 10
Building test data

Fig. 7. Variation of reliability rate.

Figure 7. Variation of reliability rate

Figure 8ofshows
Figure 7 shows the variation the comparison
the reliability of the effectiveness
rate. The average ratio.
reliability of the The proposed
proposed scheme isscheme
as high as 84.5%. This
achieves the highest effectiveness ratio of 0.864. The next highest effectiveness ratio
value is exceedingly high compared to all other schemes as shown in figure 7. This extreme peak value is achieved since
rate is attained by LRM with a value of 0.425. SVMM achieves the third-highest
the proposed AIBMEM model
rate with is based
a rate on minimizing
of 0.419. energy
the lowest usage andratio
effectiveness maximization of by
is attained usage
SRMof renewable
with a resources.
value of 0.39. Thus, we infer that the proposed AIBMEM has the best performance
in terms of effectiveness.
SRM KNNM
The SVMM
highest LRM
effectiveness
NRM
ratio
NBM
is achieved
AIBMEM
using highly
reliable smart sensors and IoT devices.
Figure 9 1shows the variation of the preservation rate. The average preservation
0.9
rate of the proposed scheme is as high as 80.75%. This value is exceedingly high
0.8
Effectiveness ratio

compared to 0.7all other schemes, as shown in Fig. 7. This high value is achieved since
0.6
2146007-16
0.5
0.4
0.3
0.2
0.1
0
1 2 3 4 5 6 7 8 9 10
Building test data
 Building test data
September 8, 2022 17:1 JOIN S0219265921460075 page 17

Figure 7. Variation of reliability rate

Figure 7 shows the variation of the reliability rate. The average reliability of the proposed scheme is as high as 84.5%. This
value is exceedingly high compared to all other schemes as shown in figure 7. This extreme peak value is achieved since
the proposed AIBMEM model is based on minimizing energy
Intelligent usagewith
Building andMulti-Energy
maximization of usage
System of renewable
Planning Method resources.

SRM KNNM SVMM LRM NRM NBM AIBMEM

1
0.9
0.8
Effectiveness ratio

0.7
0.6
0.5
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

0.4
0.3
0.2
0.1
0
1 2 3 4 5 6 7 8 9 10
Building test data
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

Fig. 8. Variation of effectiveness ratio.

Figure 8. Variation of effectiveness ratio

Figure 8 shows the comparison of

SRM
the effectiveness
KNNM
ratio. The
SVMM LRM
proposed
NRM
scheme
NBM
achieves
AIBMEM
the highest effectiveness ratio of
0.864. The next highest effectiveness
100
ratio rate is attained by LRM with a value of 0.425. SVMM achieves the third-highest
rate with a rate of 0.419. the90
lowest effectiveness ratio is attained by SRM with a value of 0.39. Thus, we infer that the
Preservation rate (%)

80
proposed AIBMEM has the best performance in terms of effectiveness. The highest effectiveness ratio is achieved using
70
highly reliable smart sensors 60
and IoT devices.
50
40
30
20
10
0
1 2 3 4 5 6 7 8 9 10
Building test data

Fig. 9. Variation of preservation rate.

Figure 9. Variation of preservation rate

Figure 9 shows the

thevariation
proposed of AIBMEM
the preservation
modelrate. The average
is designed preservation
to support rate of thebuildings
the intelligent proposedbased
scheme is as high as
on the real-time data collected using IoT sensor devices.
80.75%. This value is exceedingly high compared to all other schemes, as shown in figure 7. This high value is achieved
Figure 10 shows the comparison of the energy consumption ratio. The suggested
since the proposed AIBMEM
system model is designed
incorporates to support the
precise prediction intelligent
with maximum buildings based
control rateonand
the real-time
highest data collected
using IoT sensor dependability,
devices. the best energy consumption value is achieved. The proposed scheme
achieves the least energy consumption ratio of 0.467. The highest least energy con-
sumption ratio rate is attained by LRM with a value of 0.752. The next highest rate
is achieved by SVMM with a rate of 0.769. Thus, we infer that the proposed AIB-
MEM has the best performance in terms of energy consumption. The best energy
SRM KNNM SVMM LRM NRM NBM AIBMEM

0.9 2146007-17
0.8
consumption ratio

0.7
0.6
0.5
0.4
 Figure 9. Variation of preservation rate

September 8, 2022 17:1 JOIN S0219265921460075 page 18

Figure 9 shows the variation of the preservation rate. The average preservation rate of the proposed scheme is as high as
80.75%. This value is exceedingly high compared to all other schemes, as shown in figure 7. This high value is achieved
since the proposed AIBMEM model is designed to support the intelligent buildings based on the real-time data collected
using IoT sensor devices.

G. Qiao

SRM KNNM SVMM LRM NRM NBM AIBMEM

0.9
0.8
Energy consumption ratio

0.7
0.6
0.5
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

0.4
0.3
0.2
0.1
0
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

1 2 3 4 5 6 7 8 9 10
Building test data

Fig. 10. Variation of energy consumption ratio.

Figure 10. Variation of energy consumption ratio

Table 3. Comparison of energy management ratio.

Figure 10 shows the comparison of the energy consumption ratio. The suggested system incorporates precise prediction
with maximum control rate and highest dependability,Energy
Building
management ratio
the best energy consumption value is achieved. The proposed
test data SRM KNNM SVMM LRM NRM NBM AIBMEM
scheme achieves the least energy consumption ratio of 0.467. The highest least energy consumption ratio rate is attained
1 68.9 64.12 74.58 67.97 76.85 62.36 81
by LRM with a value 2of 0.752. 62.47
The next highest
78.96 rate is achieved 62.67
74.76 by SVMM71.18
with a rate of 0.769. Thus,
79.77 81.78 we infer that the
proposed AIBMEM has 3 the best 69.81 61.64 in terms
performance 61.26 60.61
of energy 77.09 The 70.8
consumption. best energy83.59
consumption value is
4 77.06 62.11 77.21 78.79 66.96 74.14 80.56
5 77.48 62.84 78.69 66.02 68.92 79.99 85.22
6 65.4 63.33 79.69 65.91 61.08 65.75 83.36
7 64.17 72.42 77.18 66.66 63.54 68.29 81.75
8 71.3 71.47 75.71 69.34 73.26 69.3 82.09
9 72.81 61.04 70.27 72.97 66.61 75.28 89.06
10 68.34 78.63 63.55 60.5 77.97 76.37 86.76

consumption value is attained since the proposed system has incorporated accurate
prediction with maximum control rate and highest reliability.
Table 3 shows the comparison of energy management rates. The proposed scheme
achieves the highest energy management ratio of 83.47%. SVMM attains the next
highest energy management rate with a value of 73.29%. The third-highest rate is
achieved by NBM, with a rate of 72.20%. LRM attains the lowest energy manage-
ment rate with a value of 67.14%. Thus, we infer that the proposed AIBMEM has
the best performance. The highest energy management rate is achieved with the
help of the accurate prediction capability of the system.
Table 4 shows the comparison of power conservation rates for different test build-
ing datasets. It is evident that the proposed AIBMEM scheme attains the highest

2146007-18
 September 8, 2022 17:1 JOIN S0219265921460075 page 19

Intelligent Building with Multi-Energy System Planning Method

Table 4. Comparison of power conservation rate (%).

Power conservation rate (%)

Building
test data SRM KNNM SVMM LRM NRM NBM AIBMEM

1 49.37 44.25 46.15 55.23 45.88 45.76 71.28

2 58.25 57.9 49.12 52.62 54.93 53.85 75.5
3 42.08 41.42 42.03 41.79 40.2 51.13 74.85
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

4 54.91 44.85 59.91 41.61 40.96 47.93 78.91

5 54.73 41.07 46.64 55.55 53.36 41.23 77.99
6 51.24 48.83 45.94 58.11 52.07 55.61 77.35
7 43.68 40.26 41.24 50.68 50.52 46.75 70.51
8 51.95 57.95 45.96 42.18 54.6 52.16 70.72
9 46 43.93 40.92 56.52 54.15 54.83 70.88
10 42.68 41.86 50.11 46.76 55.63 42.09 77.99
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

9 46 43.93 40.92 56.52 54.15 54.83 70.88

average power conservation rate of 74.59%. Other schemes like sparse representation
10
model (SRM),42.68 k-nearest
41.86neighbor
50.11 46.76
model (KNNM), 55.63 vector 42.09
support 77.99
machine model
(SVMM), logistic regression model (LRM), non-linear regression model (NRM),
and Naı̈ve Bayes model (NBM) attains average power conservation rate of 49.48%,
46.23%, 46.81%, 50.19%, 50.23%, and 49.13% respectively. The proposed AIBMEM
Table 4 shows the comparison of power conservation rates for different test building datasets. It is evident that the
scheme attains the highest value due to reusable power and intelligent energy iden-
proposed AIBMEM schemefunction.
tification attains the highest average power conservation rate of 74.59%. Other schemes like sparse
representation model (SRM),
Figure 11 k-nearest
shows the neighbor model of(KNNM),
comparison control support
rates forvector machine
different test model (SVMM), logistic
building
datasets. It is evident that the proposed AIBMEM scheme attains the highest control
regression model (LRM), non-linear regression model (NRM), and Naïve Bayes model (NBM) attains average power
rate of 87.57%. Other schemes like sparse representation model (SRM), k-nearest
conservation rateneighbor
of 49.48%, 46.23%,
model 46.81%, support
(KNNM), 50.19%, 50.23%, and 49.13%
vector machine respectively.
model (SVMM),The proposed
logistic AIBMEM scheme
regres-
attains the highest value
sion due to
model reusablenon-linear
(LRM), power andregression
intelligent model
energy identification
(NRM), and function.
Naı̈ve Bayes model

SRM KNNM SVMM LRM NRM NBM AIBMEM

100
90
80
Control rate (%)

70
60
50
40
30
20
10
0
1 2 3 4 5 6 7 8 9 10
Building test data

Fig. 11. Variation of control rate.

Figure 11. Variation of control rate
2146007-19
Figure 11 shows the comparison of control rates for different test building datasets. It is evident that the proposed
AIBMEM scheme attains the highest control rate of 87.57%. Other schemes like sparse representation model (SRM), k-
nearest neighbor model (KNNM), support vector machine model (SVMM), logistic regression model (LRM), non-linear
regression model (NRM), and Naïve Bayes model (NBM) attains an average rate of 60.24%, 62.41%, 67.32%, 56.82%,
57.52%, and 61.45% respectively. The proposed AIBMEM scheme attains the highest value since it has the best-controlling
September 8, 2022 17:1 JOIN S0219265921460075 page 20

G. Qiao

(NBM) attains an average rate of 60.24%, 62.41%, 67.32%, 56.82%, 57.52%, and
61.45% respectively. The proposed AIBMEM scheme attains the highest value since
it has the best-controlling capability with artificial intelligence modeling.
In summary, this section reports the experimental analysis, and the comparison
was made using other models like sparse representation model (SRM), k-nearest
neighbor model (KNNM), support vector machine model (SVMM), logistic regres-
sion model (LRM), non-linear regression model (NRM), and Naı̈ve Bayes model
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

(NBM). Based on the results observed, the proposed AIBMEM model achieved the
highest 91.59% energy efficiency rate, 80.93% prediction, 84.5% reliability, 0.864
effectiveness, and 80.75% preservation rate.

6. Conclusion and Future Work

This research presented a new Artificial intelligent building multi-energy planning
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

method (AIBMEM) for effective energy management in intelligent buildings. In the

proposed scheme, a new artificial intelligence-based model was developed based on
the training data acquired from multiple buildings. The trained model was used
in the estimation of energy. The energy estimated was validated and was used in
the conservation and effective utilization of energy resources. Further, the proposed
model adopted renewable resources using the combination of IoT sensors for real-
time data acquisition. The proposed model was validated using metrics like reliability
rate, effectiveness ratio, preservation rate, energy consumption rate, energy effi-
ciency, prediction rate, energy management ratio, and power conservation rate. The
comparison was made using other models like sparse representation model (SRM),
k-nearest neighbor model (KNNM), support vector machine model (SVMM), logis-
tic regression model (LRM), non-linear regression model (NRM), and Naı̈ve Bayes
model (NBM). The proposed AIBMEM model achieved the highest 91.59% energy
efficiency rate, 80.93% prediction, 84.5% reliability, 0.864 effectiveness, and 80.75%
preservation rate.

References
1. J. Gao, H. Wang and H. Shen, Smartly handling renewable energy instability in sup-
porting a cloud datacenter, in 2020 IEEE Int. Parallel and Distributed Processing Sym-
posium (IPDPS ) (IEEE, 2020), pp. 769–778.
2. M. Kaur, G. Kaur, P. K. Sharma, A. Jolfaei and D. Singh, Binary cuckoo search
metaheuristic-based supercomputing framework for human behavior analysis in smart
home, The Journal of Supercomputing 76(4) (2020) 2479–2502.
3. S. S. L. Preeth, R. Dhanalakshmi and P. M. Shakeel, An intelligent approach for energy
efficient trajectory design for mobile sink based IoT supported wireless sensor networks,
Peer-to-Peer Networking and Applications 13(6) (2020) 2011–2022.
4. M. Shu, S. Wu, T. Wu, Z. Qiao, N. Wang, F. Xu and B. A. Muthu, Efficient energy
consumption system using heuristic renewable demand energy optimization in smart
city, Computational Intelligence (2020).

2146007-20
September 8, 2022 17:1 JOIN S0219265921460075 page 21

Intelligent Building with Multi-Energy System Planning Method

5. N. T. Nguyen, M. C. Leu and X. F. Liu, RTEthernet: Real-time communication for

manufacturing cyberphysical systems, Transactions on Emerging Telecommunications
Technologies 29(7) (2018) e3433.
6. P. Gomathi, S. Baskar and P. M. Shakeel, Concurrent service access and management
framework for usercentric future internet of things in smart cities, Complex and Intel-
ligent Systems (2020), https://doi.org/10.1007/s40747-020-00160-5.
7. M. Gheisari, H. E. Najafabadi, J. A. Alzubi, J. Gao, G. Wang, A. A. Abbasi and A.
Castiglione, OBPP: An ontology-based framework for privacy-preserving in IoT-based
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

smart city, Future Generation Computer Systems 123 (2021) 1–13.

8. K. Wang, J. Wu, X. Zheng, A. Jolfaei, J. Li and D. Yu, Leveraging energy function
virtualization with game theory for fault-tolerant smart grid, IEEE Transactions on
Industrial Informatics (2020).
9. D. Prerna, R. Tekchandani, N. Kumar and S. Tanwar, An energy-efficient cache local-
ization technique for D2D communication in IoT environment, IEEE Internet of Things
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

Journal (2020).
10. A. Shankar and N. Jaisankar, Dynamicity of the scout bee phase for an artificial bee
colony for optimized cluster head and network parameters for energy efficient sensor
routing, Simulation 94(9) (2018) 835–847.
11. M. Elhoseny, R. S. Rajan, M. Hammoudeh, K. Shankar and O. Aldabbas, Swarm
intelligence–based energy efficient clustering with multihop routing protocol for sus-
tainable wireless sensor networks, International Journal of Distributed Sensor Networks
16(9) (2020), 1550147720949133.
12. A. Alanezi, B. Abd-El-Atty, H. Kolivand, A. El-Latif, A. Ahmed, A. El-Rahiem and
S. H. Khalifa, Securing digital images through simple permutation-substitution mech-
anism in cloud-based smart city environment, Security and Communication Networks
(2021).
13. X. Lin, X. Sun, G. Manogaran and B. S. Rawal, Advanced energy consumption system
for smart farm based on reactive energy utilization technologies, Environmental Impact
Assessment Review 86 (2021) 106496.
14. V. C. Nguyen and P. Kostarakis, The impact of green systems and signals on the health
of green residences’ habitants, Annals of General Psychiatry 17(1) (2018) A12.
15. N. Gunasekaran and Y. H. Joo, Nie–Tan fuzzy method of fault-tolerant wind energy
conversion systems via sampled-data control, IET Control Theory Applications 14(11)
(2020) 1516–1523.
16. X. Zhang, G. Manogaran and B. Muthu, IoT enabled integrated system for green energy
into smart cities, Sustainable Energy Technologies and Assessments 46 (2021) 101208.
17. B. H. Liu, N. T. Nguyen, V. T. Pham and Y. X. Lin, Novel methods for energy charging
and data collection in wireless rechargeable sensor networks, International Journal of
Communication Systems 30(5) (2017) e3050.
18. J. Liu, S. Cao, X. Chen, H. Yang and J. Peng, Energy planning of renewable applications
in high-rise residential buildings integrating battery and hydrogen vehicle storage, Appl.
Energy 281 (2020) 116038, doi: 10.1016/j.apenergy.2020.116038.
19. F. Ezbakhe and A. Pérez-Foguet, Decision analysis for sustainable development: The
case of renewable energy planning under uncertainty, Eur. J. Oper. Res. 291(2) (2021)
601–613, doi: 10.1016/j.ejor.2020.02.037.

2146007-21
September 8, 2022 17:1 JOIN S0219265921460075 page 22

G. Qiao

20. D. Zhang, T. Ahmad and W. A. Shah, Efficient energy planning with decomposition-
based evolutionary neural networks, IEEE Access 8 (2020) 134880–134897, doi:
10.1109/ACCESS.2020.3010782.
21. S. Ahmadi, A. H. Fakehi, A. Vakili and M. Moeini-Aghtaie, An optimization model
for the long-term energy planning based on useful energy, economic and environmental
pollution reduction in residential sector: A case of Iran, J. Build. Eng. 30 (2020) 101247,
doi: 10.1016/j.jobe.2020.101247.
22. T. Ahmad and D. Zhang, Novel deep regression and stump tree-based ensemble models
by FUDAN UNIVERSITY on 12/15/24. Re-use and distribution is strictly not permitted, except for Open Access articles.

for real-time load demand planning and management, IEEE Access 8 (2020) 48030–
48048, doi: 10.1109/ACCESS.2020.2978937.
23. L. Stermieri, C. Delmastro, C. Becchio and S. P. Corgnati, Linking dynamic build-
ing simulation with long-term energy system planning to improve buildings urban
energy planning strategies, Smart Cities 3(4) (2020) 1242–1265, doi: 10.3390/smartci-
ties3040061.
J. Inter. Net. 2022.22. Downloaded from www.worldscientific.com

24. S. Torabi Moghadam and P. Lombardi, An interactive multi-criteria spatial deci-

sion support system for energy retrofitting of building stocks using CommuntiyVIZ
to support urban energy planning, Build. Environ. 163 (2019) 106233, doi: 10.1016/
j.buildenv.2019.106233.
25. H. Zhong, J. Wang, H. Jia, Y. Mu and S. Lv, Vector field-based support vector regression
for building energy consumption prediction, Appl. Energy 242 (2019) 403-414, doi:
10.1016/j.apenergy.2019.03.078.
26. O. Pasichnyi, F. Levihn, H. Shahrokni, J. Wallin and O. Kordas, Data-driven strategic
planning of building energy retrofitting: The case of Stockholm, J. Clean. Prod. 233
(2019) 546–560, doi: 10.1016/j.jclepro.2019.05.373.
27. A. B. Awan, M. Zubair, Z. A. Memon, N. Ghalleb and I. Tlili, Comparative analysis
of dish Stirling engine and photovoltaic technologies: Energy and economic perspective,
Sustainable Energy Technologies and Assessments 44 (2021) 101028.
28. I. Tlili, Impact of thermal conductivity on the thermophysical properties and rheological
behavior of nanofluid and hybrid nanofluid, Mathematical Sciences (2021) 1–9.
29. I. Tlili, S. P. Samrat and N. Sandeep, A computational frame work on magnetohy-
drodynamic dissipative flow over a stretched region with cross diffusion: Simultaneous
solutions, Alexandria Engineering Journal 60(3) (2021) 3143–3152.
30. A. Aldabesh, M. Hussain, N. Khan, A. Riahi, S. U. Khan and I. Tlili, Thermal variable
conductivity features in Buongiorno nanofluid model between parallel stretching disks:
Improving energy system efficiency, Case Studies in Thermal Engineering 23 (2021)
100820.

2146007-22