Helicopter Health and Usage

Monitoring System in China

Shibin Wang, Chaowei Tong, Zhiyu Tao, Ruqiang Yan, and Xuefeng Chen

W
ith the great advantages of hovering and ver- maintenance cost reduction. Through the years, HUMS tech-
tical take-off and landing, helicopters play nology and regulations have developed greatly, and they are
irreplaceable roles in the field of civil aviation, widely used in western countries, especially in the UK, where
including transport, search and rescue, firefighting, medical HUMS is installed mandatorily for registered helicopters [2].
transport, tourism, etc. Collected from public information, by However, the majority of helicopters made in China still lack of
the end of 2019, the total number of the global helicopters had effective HUMS. The onboard maintenance system with func-
exceeded 52 000, in which the number of helicopters in China tions such as condition monitoring, fault detection, and trend
was only about 2000, nearly 4% of the total. With the rapid analysis will be the focus of future development.
development of the national economy and the opening of low-
altitude airspace, there is an increasing demand for helicopters Helicopter Health and Usage
in China. Monitoring System of Hangxin
Compared with fixed-wing aircraft, the power system of There are many organizations that carry out research and de-
the helicopter has the drive train and rotor system in addi- velopment of HUMS technology in China, including Nanjing
tion to aeroengine. The more complex mechanical structure University of Aeronautics and Astronautics, National De-
makes a helicopter’s failure rate higher than fixed-wing air- fense University, Xi’an Jiaotong University, China Helicopter
craft. According to statistics issued by the European Union Research and Development Institute, Shanghai Aero Mea-
Aviation Safety Agency, the numbers of fatal accidents and surement-Control Technology Research Institute, etc. Two
fatalities of western-built turbine helicopters have remained years ago, the first HUMS in China, developed by Guang-
high in the past ten years [1], as illustrated in Fig. 1. Excluding zhou Hangxin Aviation Technology Co., Ltd., was equipped
human factors, the failure of the helicopter’s mechanical struc- on helicopters. The current state of evolution and outlines
ture was the main cause
of accidents. As a result,
helicopter manufactur-
ers and users have always
been focusing on opera-
tional safety improvements
and maintenance cost re-
duction, which paved the
way for the implemen-
tation of the helicopter
Health and Usage Moni-
toring System (HUMS). In
the early 1990s, these sys-
tems were first deployed
on the offshore shuttle he-
licopters in the North Sea
to improve safety and re-
liability, where they also
showed the potential for Fig. 1. Number of fatal accidents and fatalities of western-built turbine helicopters from 2009 to 2019.

April 2022 IEEE Instrumentation & Measurement Magazine 45

1094-6969/22/$25.00©2022IEEE

Authorized licensed use limited to: University of Canberra. Downloaded on November 08,2022 at 12:37:39 UTC from IEEE Xplore. Restrictions apply.
Fig. 2. The schematic diagram of the HX-HUMS.

for Hangxin’s HUMS (HX-HUMS) will be presented in this At least 30 sensors are mounted on the drive train, rotors,
section. engines and airframe to measure the vibration and rotating
The schematic diagram of HX-HUMS is shown in Fig. 2. speed of the key components. In order to obtain more accu-
The system consists of a ground-based system and an onboard rate vibration signals, the accelerometers should be mounted
hardware system, mainly including accelerometers, azimuth as close as possible to the interested components. The function
sensors, a universal tracking device, a data acquisition and of data acquisition and processing board is preprocessing the
processing unit, data storage and a control unit. data measured by different sensors through amplifying, fil-
The HX-HUMS uses accelerometers, azimuth sensors and tering and A/D converting. Additionally, simple condition
the universal tracking device, distributed throughout differ- indicators that reflect the failure conditions of drive train and
ent components and the airframe, to measure the vibration engines are calculated in real-time to give an alarm to pilots
signals, the azimuth signals and the rotor tracking data, which when condition indicators exceed acceptable levels. In order to
are connected to the data acquisition and processing unit for reflect the severity of the failure of different components, am-
amplifying, filtering and A/D conversation. Then, the data ber and red thresholds are set on board for warning or alarm,
storage and control unit receive both raw and processed data respectively, based on levels which are determined by the he-
sent by Ethernet and records it in the cache card through the licopter design via dynamics analysis and experimental test.
SATA bus. Finally, data decoding and management, adjust- During flight, the data acquisition and processing board ac-
ment of the rotor track and balance, fault diagnosis of the quires data at different sampling rates according to the feature
dynamic components and maintenance recommendations can frequencies of different dynamic components, with the highest
be realized in the ground-based system. It should be noted that sampling rate at 102.4 kHz. The vibration signals and azimuth
the processed data could also be transferred to the ground sta- signals are saved based on the regimes and exceedance events.
tion through Ethernet. The onboard hardware system, ground In the case of exceedance, the raw signals for the exceedance
based system and HX-HUMS functions will be described as duration ±5 minutes will be recorded by the data acquisition
follows. and processing board.
The main control board collects the processed signals and
Onboard Hardware System the flight parameters and transfers them to the data storage
The core component of the onboard hardware system is the and control unit through Ethernet for further analysis on the
data acquisition and processing unit, which is composed of ground-based system. All the devices need the power supply
the data acquisition and processing board, main control board, board which converts the +28 V direct battery onboard to the
power supply board and interface board, as shown in Fig. 3. +5 V direct battery and ±24 V direct battery for internal circuit.

46 IEEE Instrumentation & Measurement Magazine April 2022

Authorized licensed use limited to: University of Canberra. Downloaded on November 08,2022 at 12:37:39 UTC from IEEE Xplore. Restrictions apply.
Fig. 3. The schematic diagram of the data acquisition and processing unit.

The interface board acts as a bridge to realize the interconnec-

tion among the data acquisition and processing board, main
control board, power supply board and the external devices.
Fig. 4 presents the architecture of the data acquisition and pro-
cessing unit.

Ground-based System
The ground-based system is the primary platform for users
and maintenance staff to conduct the comprehensive process-
ing and analysis for the data collected and recorded onboard.
Based on the advanced signal processing methods, HX-HUMS
can realize the vibration monitoring, exceedance and alarms,
fault diagnosis, trend analysis and life prediction of the three
major components (rotor system, transmission system and
power system) and generate the final maintenance recommen-
dations. In addition, the ground-based system is responsible
for configurating the aircrafts and parts tracking to support
the analysis of the flight data. The schematic diagram of
HX-HUMS ground based system is shown in Fig. 5.
The HX-HUMS ground station software runs on the win-
dows operating system. The main functional modules include Fig. 4. The architecture of the data acquisition and processing unit.

April 2022 IEEE Instrumentation & Measurement Magazine 47

Authorized licensed use limited to: University of Canberra. Downloaded on November 08,2022 at 12:37:39 UTC from IEEE Xplore. Restrictions apply.
 Fig. 6. The architecture of the HX-HUMS ground station software.

landing) by the flight data, including operating time, flying

time, landing count, engine cycle count, etc. Then, the usage
Fig. 5. The schematic diagram of the HX-HUMS ground-based system.
information of the life parts can be calculated accurately. Fi-
nally, it provides helpful information about flight limitations
condition monitoring and fault diagnosis, usage monitoring to improve greater pilot awareness and computes parts life for
and life analysis, data viewing and management, and mainte- maintenance staff.
nance assistance support. It uses SQL Server database for data In addition, HX-HUMS provides the system function for
storage. The ground-based system is the main part to ensure users and maintainers to configure various parameters and
safety and reduce maintenance costs by providing timely and manage the helicopters or fleets. Users and maintenance staff
accurate diagnostic results to maintenance staff and users. The also have access to the search, replay and analyze vibration sig-
overall architecture interface of the HX-HUMS ground station nals, flight parameters, condition indicators and maintenance
software is shown in Fig. 6. record of all the flights at any instance or period.

HX-HUMS Functions Key Techniques in Dynamic Component

There are three main functions for the HX-HUMS, which are Monitoring and Diagnosis
health monitoring, usage monitoring and system function. The condition monitoring and fault diagnosis of dynamic
Health monitoring for the dynamic components like components, including gears, bearings and shafts, are the
gears, bearings and shafts in the engines and drive train is most important and challenging tasks of the HX-HUMS. There
the most important function of the HX-HUMS. The typical are two main difficulties in dynamic component monitoring
and advanced signal processing methods are integrated in and diagnosis. First, it is difficult to extract the fault feature of
the diagnostic software modularly and upgradeable to sup- each dynamic component. The complex structure and numer-
port the vibration analysis and provide quantitative condition ous components of the drive train cause the vibration signals’
indicators that reflect the health condition of the particular aliasing. Second, setting proper thresholds for exceedance is
components. Finally, maintenance recommendations and re- also difficult.
ports will be generated immediately. Another function of In order to tackle these challenges, a key technique named
health monitoring is rotor track and balance, the process of time synchronous averaging has been used in HX-HUMS
smoothing vibrations in the airframe, which are caused by the [3]. The time synchronous averaging acts as a special case of
main and tail rotors. Rotor vibration is one of the main vibra- comb filters, which can extract the periodic waveforms from
tion sources and affects comfort of flight. When rotor blades a compound signal in time domain. The waveforms of both
have unequal lift and unequal distribution of mass, the vertical non-coherent components and non-synchronous compo-
vibration and lateral vibration increase rapidly. The universal nents will be effectively attenuated, and the wanted signals
tracking device is used to collect track data without attaching will be clearly retained through the “cutting”, “summing” and
tip targets to the blade tips or visually judging the position of “averaging” procedures. “Cutting” means segmenting the vi-
the rotor blades like in traditional approaches. The software bration signals based on the trigger pulse of the interested
will present the adjustment results of each blade quickly. component speed. Due to the fluctuation of the rotational
Different from health monitoring, usage monitoring is im- speed, the samples of segmentations are different. It is neces-
plemented to obtain helicopter condition based on the flight sary to resample the segmentations into the same number of
data computed on board. Manual input or calculating by data points. Then, the interested shaft waveforms and gear
flight hours cannot provide accurate usage information of the waveforms will be extracted by “summing” the resampled
life parts. HX-HUMS contains software to automatically rec- segmentations and “averaging” the results. The residuals ob-
ognize the regime (such as takeoff, hovering, level flight and tained by removing the shaft waveforms and gear waveforms

48 IEEE Instrumentation & Measurement Magazine April 2022

Authorized licensed use limited to: University of Canberra. Downloaded on November 08,2022 at 12:37:39 UTC from IEEE Xplore. Restrictions apply.
 Table 1 – The architecture of the HX-HUMS ground station software

1 N 2
Root Mean Square RMSx   x (n)
N n1
​ ​
N
4
Indicators based on raw vibration signal N   x(n)  x 
Kurtosis K= i 1
2
N 2
  x( n)  x  
 i1  ​
PPx
FM 0  H
FM0
P n
Indicators based on time synchronous averaging ​ n1 ​
esb
R I ,1
( x)  RIesb,1 ( x)
SLF SLF 
​ RMSx ​

 
N 4
N  di  d
FM4 FM 4  i 1
2
N 2
 di  d 
 i1

 ​

​
Indicators based on differential signal
 
N 6
N 2  di  d
M6A M6A  i 1
3
N 2

  di  d 
 i1  ​

​
N
N   riM  rM 
4

NA 4( M )  i 1
Indicators based on residual signal NA4  1 M  N 2 
2

    rij  rj  
M

 
​  j 1 
i 1   ​

OM1 Shaft rotating frequency

OM2 Shaft rotating frequency×2
Frequency indicators
OMx Gear meshing frequency
OM2x Gear meshing frequency×2

from the raw vibration signals can be regarded as the bearing A helicopter equipped with an HX-HUMS triggered a vibra-
waveforms and other components. The time synchronous av- tion exceedance alarm on board during climbing at the airspeed
eraging process is a simple and effective way to separate the of 140 km/h. The raw vibration data and processed data on
signals of different gears, bearings and shafts, and it is imple- board were both recorded for further analysis on the ground-
mented on board. There are two main parameters of the time based system. Using the data replay function, it can be found that
synchronous averaging algorithm, i.e., the number of average the vibration exceedance occurs at 2539~2550 s and 2552~2576 s,
revolutions and the number of the discrete points per revolu- as shown in Fig. 7 with blue circles. The condition indicator OM1
tion. The former is 120 revolutions and the latter is eight points of the main gearbox input shaft exceeds the red threshold and
in the HX-HUMS. lasts 35 s, which activates the alarm condition on board.
Another key technique is choosing proper condition in- In order to verify the fault position, the ground station soft-
dicators to assess the health condition of gears, bearings and ware is used to analyze the data related to main gearbox input
shafts. The primary condition indicators in the HX-HUMS shaft. It can be seen that the condition indicators of vibration
contain five kinds: indicators based on the raw vibration signals collected from the case of right engine and right acces-
signal, indicators based on time synchronous averaging, indi- sory of the main gearbox exceed the amber threshold and red
cators based on differential signal, indicators based on residual threshold, respectively, as shown in Fig. 8. The result on the
signal, and frequency indicators, which are listed in Table 1 [4]. ground is consistent with the result on board, which indicates
In addition, amber thresholds and red thresholds of different that there is an imbalance or misalignment on the right input
condition indicators are set for warning and alarm on board. shaft of the main gearbox. Therefore, the maintenance recom-
In order to illustrate the effectiveness of the HX-HUMS, a mendation is to check the assembly of the right input shaft of
recent case of alarm on board is given as follows. the main gearbox.

April 2022 IEEE Instrumentation & Measurement Magazine 49

Authorized licensed use limited to: University of Canberra. Downloaded on November 08,2022 at 12:37:39 UTC from IEEE Xplore. Restrictions apply.
Fig. 7. Alarm periods on board.

Fig. 8. Trends of condition indicators analyzed by the ground station software.

After the maintenance to adjust the right input shaft phase, shown in Fig. 9. This case verifies the effectiveness of the HX-
there was no vibration exceedance alarm or warning on board. HUMS in health monitoring.
Using the ground station to analyze the data related to the
main gearbox input shaft, it can be seen that the condition in- Summary
dicators of vibration signals collected from the case of right In the past five years, there has been a breakthrough of HUMS
engine and right accessory of the main gearbox are normal, as in China. Not only has HX-HUMS come into use, HUMS

50 IEEE Instrumentation & Measurement Magazine April 2022

Authorized licensed use limited to: University of Canberra. Downloaded on November 08,2022 at 12:37:39 UTC from IEEE Xplore. Restrictions apply.
Fig. 9. Trends of condition indicators analyzed by the ground station software after maintenance.

developed by Shanghai Aero Measurement-Control Technol- [3] E. Bechhoefer and M. Kingsley, “A review of time synchronous
ogy Research Institute have also been equipped on different average algorithms,” in Proc. Ann. Conf. PHM Soc., 2009.
helicopters in China. With growing concern for aviation safety [4] P. D. Samuel and D. J. Pines, “A review of vibration-based
and maintenance costs, many more organizations will get in- techniques for helicopter transmission diagnostics,” J. Sound
volved in HUMS research and bring vigor and vitality into Vibration, vol. 282, pp. 475–508, 2005.
HUMS in China. However, the long-term effectiveness of the
HX-HUMS needs to be confirmed in the future, including the Shibin Wang ([email protected]) is a Professor in the
recently developed sparsity-assisted vibration signal pro- School of Mechanical Engineering, Xi’an Jiaotong University,
cessing algorithm. Seeded fault tests should be carried out to Xi’an, China, where he received the Ph.D. degree in mechanical
support the condition assessment and fault diagnosis of the engineering in 2015. His research interests include time-fre-
major components. More cases on board should be analyzed quency analysis and sparsity-assisted signal processing for
comprehensively to improve the performance of the signal machine condition monitoring and fault diagnosis. In 2017, he
processing algorithm. Moreover, with the development of high- was a Visiting Scholar with the Tandon School of Engineering,
speed microprocessor technology, signal processing technology New York University, NY, USA. Previously, he received the B.S.
and system integration technology, more powerful and efficient and M.S. degrees in electrical engineering from Soochow Uni-
functionality can be integrated in the future HUMS system. versity, Suzhou, China, in 2008 and 2011, respectively.

Acknowledgment Chaowei Tong ([email protected]) received the B.S.

This work was partly supported by the National Natural Sci- and Ph.D. degrees in mechanical engineering from Xi’an Jiao-
ence Foundation of China under Grant Nos. 92060302 and tong University, China in 2014 and 2021. His research interests
52122504, the National Key Research and Development include time-frequency analysis, sparse representation and
Program of China (Grant No. 2018YFB1702400), and the Fun- signal processing in machine condition monitoring and fault
damental Research Funds for the Central Universities. diagnosis.

References Zhiyu Tao ([email protected]) is the Deputy General Man-

[1] “Annual safety review 2020,” European Union Aviation Safety ager of the Research and Development Center of Guangzhou
Agency, 2020. Hangxin Aviation Technology Co., Ltd. He is also the Chief
[2] “Helicopter vibration health monitoring (VHM) guidance Engineer of HUMS in Hangxin and is responsible for the over-
material for operators utilizing VHM in rotor and rotor drive all design, data analysis and technical support of HUMS. He
systems of helicopters,” Civil Aviation Authority, CAP 753, June received the M.B.A. degree from the Northwestern Polytech-
2006. nical University, Xi’an, China in 2002.

April 2022 IEEE Instrumentation & Measurement Magazine 51

Authorized licensed use limited to: University of Canberra. Downloaded on November 08,2022 at 12:37:39 UTC from IEEE Xplore. Restrictions apply.
Ruqiang Yan ([email protected]) joined the School Yan is a fellow of ASME (2019) and IEEE (2021). His hon-
of Mechanical Engineering, Xi’an Jiaotong University, ors and awards include the 2019 IEEE Instrumentation and
Xi’an, China in 2018. His research interests include data Measurement Society Technical Award. He currently serves
analytics, machine learning, and energy-efficient sens- as Editor-in-Chief for IEEE Transactions on Instrumentation
ing and sensor networks for the condition monitoring and and Measurement.
health diagnosis of largescale, complex, dynamical sys-
tems. From 2009 to 2018, he was a Professor of the School Xuefeng Chen ([email protected]) is a Professor of Me-
of Instrument Science and Engineering with the Southeast chanical Engineering with Xi’an Jiaotong University, Xi’an,
University, Nanjing, China. He received the M.S. degree in China, where he received his Ph.D. degree in 2004. His re-
precision instrument and machinery from the University of search interests include finite-element method, mechanical
Science and Technology of China, Hefei, China in 2002 and systems and signal processing, diagnosis and prognosis for
the Ph.D. degree in mechanical engineering from the Uni- complex industrial systems, smart structures, aero-engine
versity of Massachusetts Amherst, MA, USA, in 2007. Dr. fault diagnosis and wind turbine system monitoring.

52 IEEE Instrumentation & Measurement Magazine April 2022

Authorized licensed use limited to: University of Canberra. Downloaded on November 08,2022 at 12:37:39 UTC from IEEE Xplore. Restrictions apply.