Power-Voltage and Load Control of a Multi-Port

Converter Using Fuzzy Logic

Christopher Ranisch1,2 , Christian Weiner1 and Heinz Peter Berg2
1
University of Applied Sciences Darmstadt 2 Brandenburg University of Technology
Contact: [email protected]

Abstract—The shift towards a sustainable and efficient power for galvanic isolation of all source-load pairs. Thus a separate
generation is essential for limiting global warming. Renewable transformer for each source/load port is required and energy
but volatile energies must be complemented with hybrid power flow between different ports is not possible [10]. A single
plants (HPPs) to ensure grid stability and security of supply.
Grid connection of HPPs is established by multi-port converters transformer ensures galvanic isolation between the input and
(MPCs). A vital component of MPCs is the energy management output stage in partially isolated MPCs. Different windings
system (EMS) which controls the power flow and maximizes the of the transformer are used for connecting all the in- and
efficiency. This paper discusses the design of an EMS for power- output ports and thus energy flow between all ports is possible.
voltage control and load control of a grid-tied MPC. Connected SISO converters are connected to an intermediate voltage
to the MPC is a HPP, which combines solide oxide fuel cells
(SOFCs), a micro gas turbine (MGT) and a battery storage. The (capacitor) or current (inductor) circuit in non-isolated MPCs.
proposed EMS uses fuzzy logic to regulate the power flow within This type of MPC allows for energy flow between all ports.
MPC. System boundaries are taken into account by the EMS. A comprehensive overview of different MPC topologies for
Index Terms—fuzzy inference system, energy management MISO renewable energy systems and HPPs with multiple uni-
system, multi-port converter, hybrid power plant and bi-directional ports is given in [8], [10].
The EMS of a MPC is vital as it determines the overall
I. I NTRODUCTION efficiency and dynamic response of the system. In recent
Renewable energies like wind, solar and hydro power are an years various approaches for an EMS ranging from hybrid
integral part in shifting from a fossil fuel based to a sustainable algorithms to artificial intelligence have been proposed [11].
and efficient electrical power generation [1]. Conventional base In hybrid EMSs a flowchart determines the operating mode of
load power plants (i.e. coal) are substituted by emission-free the energy system based on the current system parameters and
energy sources to mitigate the induced effects of greenhouse local PID controllers compensate the error [12]. The design
gases. As the share of volatile renewable energies increases and implementation of this type of EMS strongly relies on the
the grid is subjected to high fluctuations in power generation topology of the renewable energy system or HPP. Integrating
and local imbalances [2]. This may result in an instability or additional electrical loads or sources or using the EMS for a
outages of the power grid. In order to satisfy the grid demand different energy system is complex and requires a redesign of
and compensate periods of time with too little or no renewable the flowchart. To overcome this limitation much research in the
energy HPPs have been investigated [3]. A HPP combines area of artificial intelligence based EMSs has been conducted.
multiple sustainable energy sources and storage systems for Methods like fuzzy inference systems (FISs), reinforcement
electrical power generation [4], [5]. A combined SOFC-MGT learning (RL) or artificial neural networks (ANNs) have been
and battery storage system is discussed in [6]. This type of studied [13], [14]. The authors of [15] propose a FIS for
HPP can be used with various sustainable fuels like hydrogen controlling the battery power for DC bus regulation and a
or ammonia while reaching an electrical efficiency of 70 %. switchable load for periods with excessive available energy.
Also, the rated electrical power of this system can be scaled While charging or discharging the battery the EMS keeps the
for different mobile or stationary applications and a seamless state of charge (SOC) of the battery within nominal range. In
integration into the existing gas network is possible. [16] different RL algorithms for an agent of a microgrid are
The SOFC-MGT-battery HPP topology discussed in [6] implemented and analyzed. The main tasks of the RL-based
represents a multi-input-single-output (MISO) energy system EMS are the prioritization of all participants and routing of the
with variable uni- and bidirectional AC and DC ports. Standard energy flow. Participants in the discussed scenario are flexible
amplitude or frequency converters are designed and optimized energy sources, storage systems, the grid and controlled loads.
for single-input-single-output (SISO) energy systems. Hence An ANN-EMS controls the power flow in an AC/DC microgrid
these converter topologies cannot be used for MISO energy with distributed power generation and batteries as discussed
systems [7]. To facilitate the grid integration and control the in [17]. The ANN-EMS runs the microgrid in different modes
energy flow within HPPs various topologies of so called MPCs depending on the power generation by the renewable energy
have been studied [8]. Based on the isolation characteristics sources and SOCs of the batteries. ANNs and RL algorithms
these can be classified into three different types: Isolated strongly depend on data of the system for offline training,
MPCs use passive or more recently solid-state transformers [9] validation and forecast of the EMS. Approaches based on FISs
require a set of if-than rules which describe the causality of
the energy system [18], [19]. A physical system model of the SOFC DC
MPC and energy system is neither required for ANN- nor for String 2 DC
DC
fuzzy-based EMSs [18], [20]. Grid
In the following an EMS using fuzzy logic for a HPP with SOFC AC
DC
SOFCs, an MGT and a battery storage is presented. The EMS String 1 DC
handles multiple control tasks of the MPC:
• regulation of the DC bus voltage to guarantee a stable Battery DC
system and and keep the MPC within its defined operating Storage DC
limits,
• satisfying the grid demand by charging or discharing AC
the battery and minimize the power error between grid MGT
DC
demand and the power generation and
• adjusting the operating point of the SOFCs to minimize
the charging or the discharing battery power and extend HPP Non-isolated MPC Grid
the battery lifetime.
Fig. 1. Topology of the non-isolated MPC with SISO power converters and
Further the EMS considers system boundaries (i.e. battery a common DC bus.
SOC). This paper is organized as follows: Section II deals with
the topology of the non-isolated MPC and HPP. Both systems
and the local current controllers for each source and/or load are source inverter (VSI) connects the generator to the DC bus
descriebed briefly. In section III the structure of the fuzzy EMS of the MPC. The phase currents are regulated by a predictive
with three FISs, the in- and output signals and membership deadbeat controller (PDBC) [22] and a PI controller maintains
functions are presented. A criterion for detecting steady-state nominal speed of the MGT. For the modeling of the battery
operation of the MPC is derived. The overall system with local a Thevenin model [23] with internal resistance and first order
current controllers and EMS is introduced. Simulation results system is used. The charging and discharging power is set by
of the EMS for different grid and system scenarios are shown a half-bridge DC/DC converter with PI current control. Grid
and discussed in section IV. In section V a conclusion is drawn connection of the MPC is established by a VSI with line filter.
and section V gives a short outlook for further research. A PI current controller with active damping [24] regulates the
active and the reactive power fed to the grid. The DC link
II. T OPOLOGY O F T HE H YBRID P OWER P LANT A ND T HE voltage of the MPC is 800 V. A detailed explanation of the
M ULTI -P ORT C ONVERTER modeling and control algorithm of each energy and/or storage
In this paper the design of an EMS for an HPP with multiple system and the SISO converters is given in [25]. In Table I
energy and storage systems is discussed. The topology of the and II the parameters of the HPP, the converter topologies and
HPP is a combined SOFC-MGT system with an additional the control algorithms of the MPC are listed. For the control
battery storage as proposed by the authors of [6]. In case of algorithms the references are given.
a energy deficit or surplus the battery storage acts as a buffer
and either delivers power or is charged. A common DC link
capacitor connects all energy or storage systems of the HPP TABLE I
by means of SISO converters. Fig. 1 shows the topology of the E LECTRICAL PARAMETERS OF THE HPP
non-isolated MPC with SISO power converters and a common Subsystem Parameter Value
DC bus. Since all energy and storage systems are connected SOFC strings
Maximum power per string 10.2 kW
to an intermediate DC circuit this topology allows for energy Open circuit voltage per string 453.5 V
Nominal battery capacity 1440 Ah
flow between ports. Depending on the characteristics of each Battery storage
Rated battery voltage 426.3 V
energy system bi-directional energy flow is also possible. Rated mechanical speed 50 krpm
MGT
The SOFC subsystem of the HPP consists of two strings Rated electrical power 6.5 kW
Rated grid frequency 50 Hz
where one string is a series connection of ns = 12 stacks. In Grid
Rated voltage amplitude 325 V
each stack nc = 30 cells are connected in series to increase
the terminal voltage. The SOFCs are modeled using a current-
controlled voltage source first order system and experimental TABLE II
data. Each SOFC string is connected to a boost converter and C ONVERTER TOPOLOGIES AND CONTROL ALGORITHMS OF THE MPC
a maximum power point tracker (MPPT) with adaptive step Subsystem Converter topology Control algorithm
size calculation [21] maximizes the power output. The MGT SOFC strings Boost converter MPPT [21]
is connected to a three-phase generator, which converts the Battery storage Half-bridge converter PI controller
mechanical into electrical output power. During the start-up of PDBC [22] (current)
MGT Two-level VSI
PI controller (speed)
the HPP the generator accelerates the MGT to nominal speed Grid Two-level VSI PI controller [24]
and draws power from the grid or battery. A two-level voltage
 III. F UZZY BASED E NERGY M ANAGEMENT S YSTEM 1.2

LN MN SN ZE SP MP LP
A hierachical structure with top-level EMS and low-level
1
local current controllers for the MPC is implemented. Based
on the current state of the MPC the EMS calculates the set

Degree of membership (-)

0.8
points for routing the energy flow and adjusting the operating
point of the SOFC strings. This structure allows for a simple
0.6
integration of additional energy or storage systems into the
MPC [25]. The EMS is split into two independent control
loops. Fig. 2 shows the EMS with all in- and outputs and 0.4

the power-voltage control and the load control. To maintain

interpretability and adaptability the EMS uses fuzzy logic [18]. 0.2

Further, the design of the EMS is simplified compared to

conventional approaches like PID control or decision trees 0
−40 −30 −20 −10 0 10 20 30 40
and no data for the training and the validation of the EMS Active power command (kW)

is required.
Fig. 3. Membership functions of FIS 1 for the active power set point for the
A. Power-Voltage Control grid-tied inverter.

The power-voltage control handles the regulation of the DC

bus voltage udc (k) and calculates the battery power p∗b (k) to (LN), medium negative (MN), small negative (SN), zero (ZE),
satisfy the grid demand. This control loop contains two FISs small positive (SP), medium positive (MP) and large positive
for both control tasks. To reduce the computational complexity (LP). For MN, MP, SN, SP and ZE a triangular function shape
and the total number of rules this control is designed using a and for LN, LP a linear z-, s-shaped function is chosen. The
hierarchical tree structure rather than a single FIS [26]. In range for eu (k) is set to ±10 % of the nominal DC bus voltage
addition, the interpretability of the control is maintained and and for ∆eu (k) to ±1.2 V between two samples. The set point
further extensions (energy sources) or modifications (system command of FIS 1 for the grid-tied inverter p∗ (k) is limited
boundaries) can be added [27]. to ±45 kW. Fig. 3 shows the membership functions of FIS 1
FIS 1 (voltage control) is implemented as a PD controller of the set point p∗ (k) output for the grid tied inverter.
and the inputs are the proportional The error between the power generation by the HPP and the
grid demand p∗g (k) can be described by (3). Based on the sign
eu (k) = u∗dc (k) − udc (k) (1)
and the actual value of ep (k) FIS 2 decides to either discharge
and the derivative voltage error or charge the battery to satisfy the grid demand p∗g (k). By this
the power error ep (k) is driven to zero.
∆eu (k) = eu (k) − eu (k − 1) (2)
ep (k) = p∗g (k) − p∗ (k) (3)
with the current eu (k) and the previous eu (k−1) sample of the
voltage error. In (1) u∗dc (k) is the desired and udc (k) the actual Further inputs of FIS 2 are the power error change rate ∆ep (k)
DC bus voltage. Using these input signals FIS 1 calculates the between two consecutive samples of the power error ep (k) and
set point p∗ (k) for the grid-tied inverter, which is converted the battery SOC soc(k). Equation (4) shows the calculation of
into a reference current. For the in- and the outputs of FIS 1 ∆ep (k). The input soc(k) is required to operate the battery in
seven membership functions have been defined: large negative nominal SOC range (20 % to 80 % SOC) and prevent deep
dis- or overcharge [15]. The implemented structure of FIS 2
is a PD controller (cf. (3) and (4)).
∆ep (k) = ep (k) − ep (k − 1) (4)
-
Power-Voltage FIS 2 uses the same membership functions for the power error
Control ep (k) and power error change rate ∆ep (k) as FIS 1. The SOC
is described by three membership functions: LN from 0 % to
30 % SOC, normal (NO) from 15 % to 85 % SOC and LP from
70 % to 100 % SOC. Output of FIS 2 is the battery power set
- point p∗b (k) which is converted into a reference current for the
Load Control local current controller. The range for the inputs and output of
FIS 2 is scaled to the electrical parameters (cf. Table I) of the
battery. Based on the defined in- and outputs of the power-
Energy Management System voltage control the rule bases for FIS 1 and 2 including all
possible rules are defined. The power-voltage control responds
Fig. 2. Topology of the EMS with power-voltage control and load control. to any transient event (i.e. change in grid demand or power
generation). This control loop is always active and ensures a
stable operating state of the MPC by regulating the DC bus
voltage and battery power.
B. Load Control
In steady-state operation of the MPC the load control adjusts
the operating point of the SOFC strings to reduce the required
battery power. The electrical output power pfc (k) of the SOFC
strings is controlled by changing the mass flow ṁfc (k) through Fig. 4. Signal flow diagram for detecting steady-state operation of the MPC.
each string [28], [29]. Equation 5 shows the relation between
power generated by the SOFCs pfc (k) and the mass flow
ṁfc (k) through the strings [29].
- -
pfc (k) ∼ ṁfc (k) (5)
FIS 1
One FIS is used for the load control. FIS 3 computes based
on the proportional and derivative error inputs the rate of -
change command ∆ṁ∗fc (k). The inputs of FIS 3 are -
el (k) = pg (k) − ppp (k), (6)
FIS 2
∆el (k) = el (k) − el (k − 1). (7)
Equation (6) describes the load error el (k) between the grid
power pg (k) and the generation by the HPP ppp (k). Additional
input is the mass flow ṁfc (k) which is normalized in the range - -
from 0 % to 100 %. At 100 % mass flow ṁfc (k) the SOFC
strings output their maximum power (cf. Table I). FIS 3 uses FIS 3
the membership functions defined in section III-A for the el (k)
and the ∆el (k) inputs and for the ∆ṁ∗fc (k) output. Fig. 5. Topology of the fuzzy EMS for power-voltage and load control of
the MPC.
The load control is only active in steady-state operation of
the MPC. Two criteria are analyzed for detecting the steady-
state operation: Whenever a transient event occurs the output of the integrator
1. The absolute value of the grid power gradient ∆pg (k) is I(z) for criterion 1 must be reset. The reset is either triggered
within a tolerance band δ1 for a certain period of time δ2 if the grid demand gradient ∆p∗g (k) is unequal to 0 or if the
and absolute value of el (k) is less or equal to δ3 . In (11) the logical
2. the absolute value of the load error el (k) is greater than statement for the reset of the integrator I(z) is given.
a threshold value δ3 .
∆p∗g (k) ̸= 0 ∨ |el (k)| ≤ δ3 (11)
Both criteria must be fulfilled to enable FIS 3 for load control
of the MPC. The signal flow diagram for detecting steady-state FIS 3 for the load control is enabled or disabled when the
operation of the MPC is shown in Fig. 4. state of the signal s(k) changes. The load control operates
The grid power gradient ∆pg (k) is calculated using the at a smaller sample frequency than the power-voltage control
discrete-time derivative and the maximum step size µ(k) of the local MPPTs for the
z−1 SOFC strings is limited during load control (cf. (12)). This
D(z) = k · (8)
ts · z prevents unwanted oscillations between the load control and
and the absolute value |∆pg (k)| is compared with the tolerance the MPPTs.
band δ1 . A discrete-time integration of the logical output of
(
µmax , if |µ(k)| > µmax and s(k) = 1
this operation is performed to evalute the duration threshold δ2 µ(k) = (12)
µ(k), else
of criterion 1. The forward Euler method [30] is implemented
as integration method: Fig. 5 shows EMS with FIS 1 and 2 for power-voltage control
ts and FIS 3 for load control of the MPC.
I(z) = k · . (9)
z−1 IV. S IMULATION O F T HE E NERGY M ANAGEMENT S YSTEM
In (8) and (9) ts is the sample time and k is a scaling factor
Simulations of the proposed EMS are carried out for two
which is set to 1. For criterion 2 the absolute value of el (k)
different grid scenarios and system states of the HPP. For each
is compared with the load error threshold δ3 . Equation (10)
scenario the initial SOC of the battery is set and a synthetic
shows the comparision for the evaluation of criterion 2.
grid demand profile is defined. The SOFC strings and MGT
|el (k)| > δ3 (10) are in idle state. During the start-up of the HPP the MGT must
 is finished both MPPTs are enabled and the initial mass flow
25 1
through the SOFC strings is set to 100 %. The load control is
20 0.9 disabled and enabled when the criteria 1 and 2 for steady-state
15 0.8 operation are fulfilled. For all energy and storage systems the
10 0.7
active sign convention is applied. The grid is modelled as an
Load control enabled ideal three-phase AC voltage source.
Active power (kW)

5 0.6
In scenario A the initial SOC of the battery is set to 60 %

Voltage (kV)
0 0.5
and charging/discharging of the battery with maximum power
−5 0.4 is possible. The simulation results (power signals and DC link
−10 Demand Grid 0.3 voltage) for scenario A are shown in Fig. 6. For the start-up of
SOFC strings MGT
−15 Battery DC link voltage 0.2
the MGT (blue) the EMS draws power from the battery (green)
and hence the grid demand (yellow) is zero. After 1 s the
−20 0.1
MGT reaches its nominal speed and is switched to generator
−25
0 2 4 6 8 10 12 14 16 18 20
0 operation. Both MPPTs are enabled and the SOFC strings (red)
Time (s) output their maximum power (cf. Table I) after 6.4 s. From
(a) SOFC, MGT, battery, grid power and demand and DC link voltage. 10.2 s to 18.7 s the load control of the EMS is enabled and the
operating point (mass flow) of the SOFC strings is adjusted to
match the grid demand. The load control of the EMS regulates
40
the battery power to zero to increase the lifetime of the battery.
30 After 18.7 s the load error el (k) is within the threshold δ2 and
Load control enabled criterion 2 is violated. Hence, the load control is disabled. The
20
DC link voltage (brown) does not exceed the defined limits of
10 the voltage error eu (k) and the MPC operates in a stable state.
For transient changes in the grid demand the DC link voltage
Current (A)

0
udc (k) shows a slight overshoot as shown in the marked area of
−10 Fig. 6. The mass flow (blue), battery SOC (red), SOFC string
(orange) and battery current (green) are shown in Fig. 6. Due
−20
to the limited simulation time the battery SOC varies only
−30 SOFC string current
slowly. The mass flow through the SOFC strings is adjusted
Battery current to 83 % of the nominal mass flow and the battery current
−40
0 2 4 6 8 10 12 14 16 18 20 is regulated to approx. 0 A. During load control the SOFC
Time (s) string currents are gradually reduced to prevent large power
(b) SOFC string and battery current. gradients.
At the start of grid scenario B the MPC is in idle state
and the grid power output is zero. The run-up of the MGT is
100
completed and the generator outputs 5 kW. An initial battery
90
SOC of 25 % is set and the discharging power is limited. The
80 MPPTs of the SOFC strings are enabled at 1.5 s. The power
Load control enabled
70 signals and DC link voltage for grid scenario B are shown in
Mass flow and SOC (%)

60
Fig. 6. Initially the battery is charged with 5 kW and a load
jump to 4 kW grid demand is initiated at 0.4 s. The marked
50
area in Fig. 7 shows the dynamic response to load changes of
40
the power-voltage control. From 2.2 s to 3.5 s load control is
30 enabled (shaded in grey). As the grid demand changes to a
20 set point of -19 kW the load control of the EMS is disabled
10 SOFC string mass flow
and grid demand is matched by the battery (3.5 s to 7.5 s). At
Battery SOC 9.2 s the lod control is again enabled and the mass flow and
0
0 2 4 6 8 10 12 14 16 18 20 battery current are reduced (cf. Fig. 7). From 18.5 s to 21.5
Time (s) s the grid demand is reduced to a new set point of -10.2 kW
(c) SOFC mass flow and battery SOC. and load control is disabled (see Fig. 7). The DC link voltage
Fig. 6. Simulation results of the fuzzy-based EMS for grid scenario a. of the MPC stays within the defined tolerance band ±10 % of
the nominal DC bus voltage.

be accelerated to nominal speed within 1 s before generating V. C ONCLUSION A ND O UTLOOK

power. The required power for the start-up is either drawn from A novel EMS using fuzzy logic for a non-isolated MPC has
the battery storage or the grid. When the start-up of the MGT been introduced in this work. Connected to the MPC is a HPP
 in two modes for handing any transient event and adjusting the
25 1
power generation to match the demand (section III). In section
20 0.9 IV the EMS is validated in simulations for two scenarios. As
15 0.8 opposed to competing approaches (i.e. RL or flowcharts) the
10
Load control enabled
0.7
proposed EMS offers the following advantages:
• no large database for the training and the validation of
Active power (kW)

5 0.6

Voltage (kV)
the EMS is required,
0 0.5
• a physical model of the energy system is not needed for
−5 0.4
the design of the EMS and
−10 Demand Grid 0.3 • extensions (i.e. additional sources or storage systems) can
SOFC strings MGT
−15 Battery DC link voltage 0.2 easily be incorporated into the EMS.
−20 0.1 An experimental evaulation of the proposed fuzzy-based
−25 0
EMS in a test-bench is required to analyze the system under
0 2 4 6 8 10 12 14 16 18 20 22 realistic boundary conditions. Aspects, like the architecture of
Time (s)
the control platform (i.e. distributed microcontrollers), the data
(a) SOFC, MGT, battery, grid power and demand and DC link voltage. transfer and scheduling of each control task of the EMS are
unresolved challenges.
40 The presented EMS for HPPs and hybrid renewable energy
35
systems lays the foundation for further research. Especially
the membership functions (shape and range) and the rule
30
Load control enabled base (number of rules) can be optimized using methods like
25
genetic algorithms [18], [20]. An optimized rule base of each
20 FIS further reduces the computational load and increases the
Current (A)

15 interpretability [31]. Further, the influence of different grid

10
faults and grid conditions on the EMS and possible instabilities
must be investigated. The following grid faults and conditions
5
must be conducted to further research:
0
• dynamic loads (i.e. load shedding or oscillations),
SOFC string current
−5
Battery current • grid faults (voltage drop, phase drop or short circuit) and
−10 • grid conditions (i.e. low inertia or unbalanced phase
0 2 4 6 8 10 12 14 16 18 20
Time (s) voltages).
(b) SOFC string and battery current. Furthermore, an additional energy storage system and load
sharing minimize the thermal and discharge stress put on the
battery [32]. The techniques for load sharing proposed by the
100
authors of [33], [34] show a great potential for this application.
90
These can easily be integrated into the implemented topology
80 of the fuzzy-based EMS.
70
Mass flow and SOC (%)

Load control enabled ACKNOWLEDGMENT

60

50
The research project is supported by the Johannes Hübner
Stiftung Gießen, Germany. The authors express their thanks.
40

30 R EFERENCES
20 [1] M. J. B. Kabeyi and O. A. Olanrewaju, “Managing sustainability
10 SOFC string mass flow in electricity generation,” in 2020 IEEE International Conference on
Battery SOC Industrial Engineering and Engineering Management (IEEM). IEEE,
0 2020, pp. 530–536.
0 2 4 6 8 10 12 14 16 18 20
[2] J. Oyekale, M. Petrollese, V. Tola, and G. Cau, “Impacts of renewable
Time (s)
energy resources on effectiveness of grid-integrated systems: Succinct
(c) SOFC mass flow and battery SOC. review of current challenges and potential solution strategies,” Energies,
vol. 13, no. 18, p. 4856, 2020.
Fig. 7. Simulation results of the fuzzy-based EMS for grid scenario b. [3] O. Farhat, M. Khaled, J. Faraj, F. Hachem, R. Taher, and C. Castelain,
“A short recent review on hybrid energy systems: Critical analysis and
recommendations,” Energy Reports, vol. 8, pp. 792–802, 2022.
with SOFCs, a MGT and a battery. The designed EMS handles [4] J. C. León Gómez, S. E. de León Aldaco, and J. Aguayo Alquicira,
“A review of hybrid renewable energy systems: Architectures, battery
multiple control tasks while maintaining a simple, expandable systems, and optimization techniques,” Eng, vol. 4, no. 2, pp. 1446–
and human interpretable topology. Further, the EMS operates 1467, 2023.
 [5] F. Valenciaga and P. F. Puleston, “Supervisor control for a stand-alone [28] T. Dotschel, A. Rauh, L. Senkel, and H. Aschemann, “Experimental
hybrid generation system using wind and photovoltaic energy,” IEEE validation of interval-based sliding mode control for solid oxide fuel
Transactions on Energy Conversion, vol. 20, no. 2, pp. 398–405, 2005. cell systems,” in 2013 European Control Conference (ECC). IEEE,
[6] H. P. Berg, A. Himmelberg, M. Lehmann, R. Dückershoff, and M. Neu- 2013, pp. 2489–2494.
mann, “The turbo-fuel-cell 1.0 – family concept,” IOP Conference [29] H. Zhao, P. Zheng, Q. Zhu, X. Zhang, Z. Lan, Z. Fu, T. Luo, R. Shen,
Series: Materials Science and Engineering, vol. 297, p. 012004, 2018. and J. Cheng, “Effect of steam reinjection mass flow rate on the sofc–gt
[7] M. Taufik, T. Taufik, and T. Wong, “Multiple-input single-output con- system with steam reinjection,” Frontiers in Energy Research, vol. 12,
verter for renewable energy sources,” in 2012 IEEE Symposium on 2024.
Industrial Electronics and Applications. IEEE, 2012, pp. 130–135. [30] M. Comanescu, “Influence of the discretization method on the integra-
[8] R. Madhana and G. Mani, “Power enhancement methods of renewable tion accuracy of observers with continuous feedback,” in 2011 IEEE
energy resources using multiport dc-dc converter: A technical review,” International Symposium on Industrial Electronics. IEEE, 2011, pp.
Sustainable Computing: Informatics and Systems, vol. 35, p. 100689, 625–630.
2022. [31] P. C. Shill, Y. Maeda, and K. Murase, “Optimization of fuzzy logic
[9] H. Shadfar, M. Ghorbani Pashakolaei, and A. Akbari Foroud, “Solid– controllers with rule base size reduction using genetic algorithms,” in
state transformers: An overview of the concept, topology, and its 2013 IEEE Symposium on Computational Intelligence in Control and
applications in the smart grid,” International Transactions on Electrical Automation (CICA). IEEE, 2013, pp. 57–64.
Energy Systems, vol. 31, no. 9, 2021. [32] M. S. Wasim, S. Habib, M. Amjad, A. R. Bhatti, E. M. Ahmed, and
[10] L. Gevorkov, J. L. Domı́nguez-Garcı́a, L. T. Romero, and À. F. Martı́nez, M. A. Qureshi, “Battery-ultracapacitor hybrid energy storage system
“Modern multiport converter technologies: A systematic review,” Ap- to increase battery life under pulse loads,” IEEE Access, vol. 10, pp.
plied Sciences, vol. 13, no. 4, p. 2579, 2023. 62 173–62 182, 2022.
[11] S. Ahmad, M. Shafiullah, C. B. Ahmed, and M. Alowaifeer, “A review [33] K. Kwon, D. Park, and M. K. Zadeh, “Load frequency-based power
of microgrid energy management and control strategies,” IEEE Access, management for shipboard dc hybrid power systems,” in 2020 IEEE
vol. 11, pp. 21 729–21 757, 2023. 29th International Symposium on Industrial Electronics (ISIE). IEEE,
[12] T. Alnejaili, A. Alibi, L. Chrifi-Alaoui, D. Mehdi, H. Sahraoui, and 2020, pp. 142–147.
S. Drid, “Control and optimal management of a multi-source energy [34] M. A. Setiawan, A. Abu-Siada, and F. Shahnia, “A new technique
system,” in 2018 7th International Conference on Systems and Control for simultaneous load current sharing and voltage regulation in dc
(ICSC). IEEE, 2018, pp. 164–168. microgrids,” IEEE Transactions on Industrial Informatics, vol. 14, no. 4,
[13] Y. Cao, M. Yao, and X. Sun, “An overview of modelling and energy pp. 1403–1414, 2018.
management strategies for hybrid electric vehicles,” Applied Sciences,
vol. 13, no. 10, p. 5947, 2023.
[14] K. A. Al Sumarmad, N. Sulaiman, N. I. A. Wahab, and H. Hizam,
“Energy management and voltage control in microgrids using artificial
neural networks, pid, and fuzzy logic controllers,” Energies, vol. 15,
no. 1, p. 303, 2022.
[15] A. El Zerk, M. Ouassaid, and Y. Zidani, “Energy management based
fuzzy logic control of hybrid system wind/photovoltaic with batteries,” in
2018 Renewable Energies, Power Systems & Green Inclusive Economy
(REPS-GIE). IEEE, 2018, pp. 1–6.
[16] T. A. Nakabi and P. Toivanen, “Deep reinforcement learning for energy
management in a microgrid with flexible demand,” Sustainable Energy,
Grids and Networks, vol. 25, p. 100413, 2021.
[17] K.-M. Kang, B.-Y. Choi, H. Lee, C.-G. An, T.-G. Kim, Y.-S. Lee,
M. Kim, J. Yi, and C.-Y. Won, “Energy management method of hybrid
ac/dc microgrid using artificial neural network,” Electronics, vol. 10,
no. 16, p. 1939, 2021.
[18] J. Adamy, Fuzzy Logik, neuronale Netze und evolutionäre Algorithmen,
6th ed., ser. Berichte aus der Steuerungs- und Regelungstechnik. Düren:
Shaker Verlag, 2023.
[19] Z. A. Styczynski, K. Rudion, and A. Naumann, Einführung in Experten-
systeme. Berlin, Heidelberg: Springer Berlin Heidelberg, 2017.
[20] K. Michels, F. Klawonn, R. Kruse, and A. Nürnberger, Fuzzy control:
Fundamentals, stability and design of fuzzy controllers, ser. Studies
in fuzziness and soft computing. Berlin and Heidelberg: Springer,
2006, vol. Vol. 200. [Online]. Available: http://www.loc.gov/catdir/
enhancements/fy0901/2006920240-d.html
[21] I. Owusu-Nyarko, M. A. Elgenedy, I. Abdelsalam, and K. H. Ahmed,
“Modified variable step-size incremental conductance mppt technique
for photovoltaic systems,” Electronics, vol. 10, no. 19, p. 2331, 2021.
[22] J. Peng and M. Yao, “Overview of predictive control technology
for permanent magnet synchronous motor systems,” Applied Sciences,
vol. 13, no. 10, p. 6255, 2023.
[23] H. Hinz, “Comparison of lithium-ion battery models for simulating
storage systems in distributed power generation,” Inventions, vol. 4,
no. 3, p. 41, 2019.
[24] J. Xu, S. Xie, and T. Tang, “Active damping-based control for grid-
connected lcl -filtered inverter with injected grid current feedback only,”
IEEE Transactions on Industrial Electronics, vol. 61, no. 9, pp. 4746–
4758, 2014.
[25] C. Ranisch, C. Weiner, and H. P. Berg, “Fuzzy-based energy manage-
ment for a fuel cell micro gas turbine battery power plant,” unpublished.
[26] L.-X. Wang, “Analysis and design of hierarchical fuzzy systems,” IEEE
Transactions on Fuzzy Systems, vol. 7, no. 5, pp. 617–624, 1999.
[27] S. Kamthan and H. Singh, “Hierarchical fuzzy logic for multi-input
multi-output systems,” IEEE Access, vol. 8, pp. 206 966–206 981, 2020.