Next Article in Journal
CFD Investigation into Flow Characteristics of a Special Splash Lubrication in Light Helicopters
Next Article in Special Issue
Effect of Local Momentum Ratio on Spray Windward Distribution of a Gas–Liquid Pintle Injector Element
Previous Article in Journal
A Pixel-Wise Foreign Object Debris Detection Method Based on Multi-Scale Feature Inpainting
Previous Article in Special Issue
Development of Ultra-Low Specific Speed Centrifugal Pumps Design Method for Small Liquid Rocket Engines
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Aerospace
Volume 9
Issue 9
10.3390/aerospace9090481
aerospace-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Research and Development of Fault Diagnosis Methods for Liquid Rocket Engines

by
Tao Wang
*,
Lin Ding
and
Huahuang Yu
School of Intelligent Systems Engineering, Sun Yat-sen University, Guangzhou 510006, China
*
Author to whom correspondence should be addressed.
Aerospace 2022, 9(9), 481; https://doi.org/10.3390/aerospace9090481
Submission received: 7 July 2022 / Revised: 22 August 2022 / Accepted: 25 August 2022 / Published: 29 August 2022
(This article belongs to the Special Issue Liquid Rocket Engines)
Browse Figure
Versions Notes

Abstract

:
Currently, considerable efforts are being focused on the development of reusable rockets and smart rockets due to the heavy requirements of future next-generation aerospace transportation. Safety, low-launching cost, and repeatability are expected from liquid rocket for fulfilling the big dreams of space transportation, exploration, and travelling. Therefore, research on fault detection of the liquid rocket engines (LRE) is critical for satisfying the above claims. Therefore, a comprehensive survey on the research and development of fault diagnosis systems and methods for the liquid rocket engines is presented. First, development history of liquid rocket engine diagnostic systems is reviewed thoroughly. Then three broad headings of the fault detection approaches of liquid rocket engines are divided through the summary and analysis of the existing methods, including approaches using signal processing, model-driven approach, and approach using artificial intelligence (AI). Then the paper discusses the concrete algorithms according to the classification features of the algorithms. In the end, the future developments of the fault detection approaches are presented, which will mainly pay attention to the reusability and intelligence of the rockets.

1. Introduction

Liquid rocket engine (LRE), as the core power of the flight power of the launch vehicle system, determines the overall performance of the launch vehicle, and is directly related to the success or failure of space missions. LRE is a complex thermo-hydrodynamic system composed of many different independent dynamic links cross-coupling with each other [1,2,3]. It works in extremely harsh environments such as high pressure, high temperature, high-speed airflow ablation, propellant component erosion, and high-frequency oscillation, etc. It works in extremely harsh environments such as high pressure, high temperature, high-speed airflow ablation, propellant component erosion, high-frequency oscillation, etc. In the environment [4], where the working conditions are close to the physical limit of the material, any tiny anomaly may quickly develop into a destructive failure, leading to the failure of the launch mission and huge economic losses.
On 28 June 2015, the Falcon 9-1.1 rocket of the US Space Exploration Technology Corporation (SpaceX) while performing the 7th International Space Station Cargo Supply Mission (SpaceX CRS-7) failed and exploded 139 s after its launch. the Dragon spacecraft and the payload it carried were also damaged. In November of the same year, the SpaceX investigation team located the fault as a problem with the manufacturing materials of the helium cylinder bracket in the second-stage liquid oxygen tank [5]. On 1 December 2016, the Russian space agency used a “Soyuz U” rocket to launch the “Progress MS-04” cargo spacecraft from the Baikonur launch site in Kazakhstan. After flying for 384 s, the oxygen pump caught fire and ruptured, which destroyed the second-stage RD-0110 engine and caused damage to the second-stage oxygen tank. The spacecraft and the rocket separated about 140 s ahead of schedule, and finally crashed. On 2 July 2017, the Long 5 March (CZ-5) Remote 2 rocket was launched at the Wenchang Space Launch Site in Hainan. When the rocket flew to 346 s, an abnormality occurred in the local structure of the turbine exhaust device of the first-level YF-77 engine, the engine thrust dropped sharply in an instant, resulting in the failure of the launch mission [6,7,8].
The failure of related launch missions has had a profound impact on the development of aerospace research worldwide, and with the rapid development of related sensor detection technologies, LRE can be detected in a more precise and timely manner. The complexity also puts forward higher and higher requirements for the reliability and security of LRE, and the research on LRE health monitoring technology has received widespread attention [2]. LRE health monitoring technology mainly covers three aspects of fault detection, fault diagnosis, and fault control [1,9,10]. Fault detection, that is, judging whether the engine has a fault or not according to the current service state of the engine, combined with the structural characteristics, historical operation data of the engine, etc. Fault diagnosis is to judge the type of engine fault, the location of the fault, and the degree of the fault on the basis of detecting the fault. Fault control refers to reducing the losses caused by faults as much as possible through measures such as fault warning, startup safety redundancy, and emergency shutdown. Among them, fault detection is the premise of fault diagnosis and fault control; fault diagnosis is to diagnose faults according to the results of fault detection, and provide directions for the next step of fault control; the purpose of fault control is to reduce the losses caused by faults as much as possible, avoiding mission failure.
Carrying out LRE health monitoring and fault diagnosis can effectively improve the reliability and safety of the launch vehicle power system [11], and related technologies have important engineering application value [12]. Effective anomaly detection of LRE can timely find the defects of rocket engine design or process in the engine development stage, and provide corresponding support for engine model testers and research designers to analyze and locate faults. At the same time, the abnormal detection of the engine can help researchers to diagnose and control the possible problems of the engine on the ground and in flight in real-time, provide direction, further reduce the cost of rocket launch and operation, and improve the safety and reliability of the launch system [13,14].
With the substantial improvement of the performance of hardware equipment, the continuous development of information technology and artificial intelligence algorithms, combined with artificial intelligence technology, the development of smart rockets has become the development trend of a new generation of launch vehicles [15]. In the middle of the 20th century, the aerospace powers represented by the United States have already carried out research in the field of smart rocket-related technologies. On 8 October 2012, during the resupply mission of SpaceX’s Falcon 9 to the International Space Station, one of the engines of the first sub-stage of the rocket failed. Because the fault diagnosis technology is used on the rocket, it did not affect the main task. China’s Long March series of launch vehicles are also speeding up the upgrade to the intelligent direction. The Long March 8 (CZ-8), which made its first flight on 22 December 2020, can identify flight faults online in the taxiing section and autonomously reconfigure attitude control under specific fault conditions [16,17]. The solid–liquid “hybrid version” launch vehicle, the Long March No. 6 modified (CZ-6A) launch vehicle, which made its first flight on 29 March 2022, had a 0.3 s gap between the ignition of the core-stage LRE and the ignition of the solid booster. The engine health diagnosis system was used to complete the fault diagnosis of the core-level power system, and then perform the ignition procedure of the solid booster after ensuring the core-level engine is healthy. This launch mission is the first time that the engine health diagnosis system has been applied to rocket launch in China. Therefore, the online fault detection of the rocket will become an indispensable part of the intelligent control of the launch vehicle, thus improving the reliability of the mission.
In summary, the research on LRE intelligent fault detection technology can effectively improve the reliability and safety of LRE, avoid accidents as much as possible, and also help researchers to improve engine structure and performance. On the one hand, LRE’s intelligent fault detection is an important part of smart rockets and an indispensable part of future “learning” rockets, which has very important engineering application value.

2. Development History of Liquid Rocket Engine Diagnostic System

Foreign countries attach great importance to the failure detection technology of LRE. The history of LRE failure detection can be traced back to the Apollo program in the United States in 1967. In order to reduce the failure rate of rocket engines, the National Aeronautics and Space Administration’s Naval Research Laboratory (sponsored by NASA) started the research on LRE health monitoring technology [1,12]. Since then, with the increasing complexity of space transportation technology, scholars have carried out a lot of research on the LRE fault detection system and detection method to improve the safety and reliability of launch vehicles, which achieved a lot of research results.
Since the 1970s, the United States has developed various LRE fault detection and diagnosis systems, which are widely used in the Space Shuttle Main Engine (SSME) [3,18], mainly including the red line parameter system (Redline), Anomaly and Failure Detection System (System for Anomaly and Failure Detection, SAFD) [19], Rocket Engine Health Management System [20], Health Management System for Rocket Engine (HMSRE), Integrated Vehicle Health Management System (IVHMS), Real-Time Turbine Engine Diagnostic System (RTTEDS) [21], Real-Time Vibration Monitoring System(RTVMS) [22], Post-Test Diagnostic System (PTDS), Health Monitoring System (HMS) [23], Intelligent Integrated Vehicle Management System (IIVMS), etc. [2,24,25]. Entering the 21st century, Stennis Space Center, Ames Research Center and Rocketdyne Company proposed Integrated System Health Management (ISHM) for the A-1 test bed and J-2X engine [26,27]. Space agencies in Russia, Japan, the European Union and other countries or regions have also carried out their own research work on the health management of LREs. Russian company HEO developed a functional diagnosis system that accurately models the engine dynamics called ACФКд [28], automatic adjustment of liquid engine based on accurate mathematical model [29]. The related research works of foreign research institutions provide useful references in the field of LRE health management in China. In the early 1990s, China began research on rocket engine fault diagnosis. The related research units mainly focused on China’s Long March series of launch vehicles, which carried out rocket engine fault pattern recognition, engine fault simulation system design, fault detection and diagnosis related algorithms, etc. Research on ground engine test failure detection system [24,30]. For example, for the YF-75 engine ground test process, Liu et al. [31] of National Defense University of Science and Technology developed a real-time fault detection and alarm system (LRE Real-Time Fault Detection and Alarm System, LRE-RTFDAS), Beijing University of Aeronautics and Astronautics and Beijing Fengyuan Research Institute [32] jointly developed the Condition Monitoring and Fault Diagnosis System (CMFDS), which can monitor and detect the engine in real-time during the actual test run of the engine. Xie et al. [33] developed a turbo pump health monitoring system (Turbo Pump Health Monitoring System, TP-HMS). Bastrygin et al. [34] devised and developed an LRE state diagnosis system. The new system increased the number and types of sensors, reduced the size of sensors and modules, and added data transmission channels. The addition of data transmission channels leads to the increase of monitoring area. A larger number of sensors make the measured data more perfect and ensure all the information needed for diagnosis are selected, as a result of that, it greatly increases the reliability of engine process description. Using the latest small piezoelectric sensor, the stability of the structure is increased, and the data module with wireless channel transmits the sensor information over a short distance to improve the reliability of the data, so as to improve the reliability of the system state diagnosis and reduce the failure rate. In the application on arrows, in addition to the aforementioned CZ-6A carrier rocket engine health diagnosis system, China’s only manned carrier rocket Long March 2F (CZ-2F) is equipped with an automatic fault detection and processing system [35]. The system can detect the failure of the rocket and make autonomous decisions on whether to implement the emergency escape of the astronauts according to the situation.

3. Summary of State of Fault Detection Technology

As the core and foundation of the rocket engine fault detection system [3], the fault detection technology of LRE can be roughly divided into three stages [9]. The second stage is a fault detection technology using signal processing and modeling analysis developed with the development of sensor technology; the third stage uses the core intelligent fault detection technology, such as data mining, deep learning, and information fusion. Scholars have different views on the classification of LRE fault diagnosis techniques [3,9,12,25,36,37]. In this paper, LRE fault detection approaches are divided into three types: signal processing-based approaches, model-driven approaches, and artificial intelligence-based approaches, which are shown in Figure 1.

3.1. Approach Using Signal Processing

The fault detection approach using signal processing refers to determining whether the engine has faults by directly processing the relevant measurable signals, such as red line algorithm, adaptive correlation algorithm, etc.; or through signal models, such as correlation function, autoregressive sliding The average model (Auto-Regressive Moving Average Model, ARMA), etc., extracts the variance, frequency and other characteristics of the signal, and determines whether the detected engine is faulty.

3.1.1. Red Line Algorithm

The red line algorithm [38] can be divided into threshold approach and envelope approach. The threshold approach is a constant space with no time dimension while the envelope law forms upper and lower bounds with a time dimension using historical data. The basic principle is that under the normal working condition of the system, its input and output values fluctuate within the normal range, and if it exceeds the corresponding range, the system is considered to be faulty. The threshold-based discrimination approach is the earliest and most commonly used fault detection approach, which is characterized by simple and clear detection logic. This approach mainly relies on signal processing, and does not need to build a model of the object to be diagnosed, but it only uses the feature information in the detection information and has no deep mining of the structure and internal changes of the system [36]. Zhang et al. [39] used the fixed threshold approach to discriminate the faults when exploring the failure mode and failure control strategy of the high-thrust liquid oxygen kerosene supplementary combustion engine in the starting preparation stage and the starting process. The threshold-based fault detection approach is very simple to implement. The output signal is compared with the threshold set in advance to determine whether there is a fault. However, improper setting of the threshold will lead to false positives and false negatives. Therefore, an adaptive threshold algorithm (Adaptive Threshold Algorithm, ATA) was subsequently developed. The difference is that the set threshold is no longer a fixed value, but is automatically updated as the change of engine operating conditions. In the research on fault detection and isolation of liquid oxygen methane engine, Li [40] explored a fault detection approach of engine pipeline based on ATA. His research showed that this approach has high fault detection accuracy and no leakage or false alarm. Yablochko et al. [41] explored the fault diagnosis method of 11D58M engine and proposed an adaptive algorithm for engine fault diagnosis, which can detect the state of LRE without a large amount of data. The algorithm uses the data of oxidant pressure after pump obtained from 50 successful starts of 11D58M engine in stationary mode and the random normal distribution law to obtain the threshold value of diagnosis state, then uses the threshold value to distinguish whether the LRE has fault or not. The algorithm only limits a single parameter of the engine, which ensures the speed of diagnosis and the reliability of diagnosis results. It has been validated in the stationary mode. However, in the transient mode, the error judgment of the normal transient mode will be made due to the fixed threshold value. Hu et al. [42] proposed an adaptive Gaussian threshold fault detection approach and applied it to the health monitoring of large liquid rocket engine turbopumps. The approach has low computational complexity and is easy to realize in real-time. Although the ATA algorithm can effectively improve the reliability of fault detection, it does not fully consider the interaction between parameters and the impact of abnormal data on the threshold. The fault detection based on the envelope relies on the normal operation of the LRE, the variation rules of relevant sensor parameters are basically the same, and the values are basically concentrated in the envelope determined by a certain threshold.

3.1.2. Autoregressive Moving Average Model

ARMA is an analytical approach, which uses parametric models to process ordered random vibration response data to identify modal parameters. The ARMA model makes a short-term forecast for the system by analyzing the laws of the ordered time series itself. Deng et al. [43,44] devised a real-time fault detection approach in the steady-state stage of LRE based on the ARMA model in the research on the fault diagnosis of the main stage of the high-thrust hydrogen-oxygen supplementary combustion cycle engine, and simulated them through hardware-in-the-loop. The platform verifies the reliability of the approach. Zhao et al. [45] verified the fault detection approach based on the ARMA model in real-time on their hardware-in-the-loop simulation platform for rocket engine fault diagnosis which uses the rapid prototyping approach. Xue et al. [30] applied the ARMA model in the liquid oxygen methane rocket engine fault detection model. The related algorithm successfully diagnosed the common faults of the LRE and met the real-time requirements of engine fault detection. It is noteworthy that the ARMA model has a very important premise when it is used, and the analyzed data must be stable, so this approach is generally used in the fault detection of the engine in the steady state stage.

3.1.3. Other Approaches Using Signal Processing

In addition, researchers also conducted a series of experiments using engine vibration signal [46,47], adaptive correlation algorithm [48], frequency band peak ratio [49,50], characteristic frequency band root mean square value [51], and principal component analysis-based extraction of eigenvectors. Fault detection approaches using signal processing are studied, such as binary time series analysis of phase plane trajectory [52,53], sliding time window principal component analysis [54], fast Fourier transform [55], etc. In order to monitor the state of LRE in transient mode accurately, Kamensky et al. [56] respectively establish different fault detection and evaluation benchmarks in stationary mode and transient mode according to different characteristics of stationary and transient mode. For stationary mode, dramatic parameter changes are the sign that something is wrong. Therefore, evaluating the time series of measured parameters based on the Student’s criterion can detect the occurrence of faults; for transient mode, the measured parameters change stably over time. According to the given cycle curve, the driving angle of the control unit changed by linear law, ensures the transient control of the engine.
Although the fault detection approach using signal processing is simple and intuitive, it also has the disadvantage of low accuracy, and for LRE, due to the complexity of its structure and working process, it is difficult to obtain all fault mode characteristics. The fluctuation is relatively severe, and the approach using signal processing is more focused on the fault detection and analysis of the LRE steady-state working stage.

3.2. Model-Driven Approach

The model-driven approach [57] refers to reconstructing the state of the system that uses an observer or filter to form a residual sequence, and judging whether a fault occurs relies on the statistical analysis of the residual sequence. The model-driven fault detection approach requires that the mathematical model of the detection object is known and can be established. The model-driven fault detection approach can use the deep knowledge inside the system to deeply understand the nature of the detected object system, faults and faults patterns can be traced back to information in a physical sense and have the ability to detect unknown faults [36]. Physical model approach, Kalman/Extended Kalman filter, particle filter and other approaches can be regarded as model-driven fault detection approaches. Model-driven approaches can be divided into analytical model-based approaches and qualitative model-based approaches [1,3].

3.2.1. Fault Detection Approach Using the Analytical Model

Analytical model-based fault detection approaches [1,3,12] first establish a mathematical model representing the working process of the engine according to the working principle of the engine, then compare the output of the model with the measurement information of the actual engine system, and compare the obtained residual error. The value is analyzed and processed to determine whether the system fails. The anomaly detection algorithm (Rocketdyne Safety Algorithm, RSA) was developed by American Rocketdyne Company for SSME to detect engine fault with this thought.
Zhang et al. [58] established a dynamic simulation model of a 50-ton high-thrust liquid-hydrogen-liquid-oxygen rocket engine (referred to as a hydrogen-oxygen rocket engine) based on the AMESim simulation platform, which can realize the fault identification and location of this type of LRE. Zheng et al. [59] used the AMESim simulation software platform to establish a dynamic mathematical model of each component of a high-thrust hydrogen-oxygen rocket engine. By introducing fault factors, the engine faults are analyzed. The model can be used for engine fault analysis and location. Ye et al. [60] detected the thrust drop fault of the launch vehicle power system using the extended Kalman filter. Tsutsumi [61] developed a model-based approach for LRE electromechanical brake failure detection. Kawatsu [62] developed a model-based fault detection and diagnosis program based on Modelica using a dynamic time warping algorithm. Cha et al. [63] studied fault detection in the start-up phase of an open-cycle LRE using a nonlinear Kalman filter method. Lee et al. [64] introduced the application of a fault detection and diagnosis algorithm using the Kalman filter and the fault factor method in the steady-state working phase of an open-cycle LRE. Levochkin et al. [65] proposed a mathematical model, which was based on the steady values of parameters and their derivatives and their relationship between the anomaly modes, forms, and the diagnostic characters, and then shows five fault search algorithms according to the different failure modes. The algorithms can get real-time parameter information, which make it possible to predict whether the state of the engine is healthy or not and avoid failure. Pekarskaya et al. [66] analyzed the components of the engine in non-stationary operation mode based on differential equations and algebraic equations, established the dynamic model of the liquid rocket engine, and obtained the method to ensure the stable operation of LRE. Levochkin et al. [67] use mathematical equations to describe units, calculate the average value of the measured parameters, receive real-time data, then compare real-time data with calculated values, and see whether one deviates from the anther one, deviation indicates a fault.
The accuracy of fault detection results implemented by the analytical model-based fault detection approach directly depends on the degree of agreement between the established model and the actual system. However, for such a complex system as LRE, it is still difficult to establish a system model which can reasonably express its working mechanism. In addition, fault detection approaches using analytical models are usually specific and cannot be used for different models and engine system, these factors limit the use of this method.

3.2.2. Fault Detection Approach Using Qualitative Model

The fault detection approach using the qualitative model [1] needs to establish a qualitative model of the fault detection system according to the dependency of the engine system, then compares the expected behavior of the model with the actual behavior of the system to obtain anomalies, and finally searches and solves the fault source. The detection approach is simple in calculation, but the solution process is prone to false behavior, and the fault detection accuracy of the approach is not high.
In the actual problem of LRE fault detection, it is very difficult to establish an accurate and effective mathematical model. The model has high complexity and poor flexibility, and it is difficult to meet the actual needs. Therefore, the model-driven fault detection approach is used in practice. The application has been greatly restricted.

3.3. Approach Using Artificial Intelligence

Artificial intelligence-based approaches can also be called intelligent detection techniques or data-driven approaches. The approach using artificial intelligence is a fault detection technology developed with the development of artificial intelligence technology and computer technology. The fault detection approach using artificial intelligence does not need to establish an accurate mathematical model, but analyzes the performance of the engine according to the historical data of the sensor, which becomes an important means of effective fault detection. Typical artificial intelligence-based fault detection approaches include expert system-based fault detection approaches, statistical reliability-based fault detection approaches, and neural network-based fault detection approaches.

3.3.1. Fault Detection Approach Using Expert System

The fault detection approach using the expert system needs to establish a fault detection knowledge base that stores the knowledge of fault symptoms, fault modes, etc., and then uses the inference engine to perform fault detection and reasoning. The fault detection approach using the expert system does not need to establish a mathematical model of the engine, but relies on the expert knowledge and engineering experience obtained in practice to detect and diagnose the engine fault by simulating the expert’s decision-making, and has a certain ability to explain the engine fault.
Ali et al. [68] developed an Expert System for Fault Diagnosis for SSME. Gupta et al. [69] designed a real-time fault diagnosis expert system named LEADER for SSME. The fault detection approach using the expert system is subject to expert knowledge, and it is more dependent on the experience obtained in the past, and it is difficult to find new faults. It is generally used for offline analysis of the engine’s working state [1].

3.3.2. Fault Detection Approach Based on Statistical Reliability

The fault detection approach based on statistical reliability relies on statistical reliability or probability, and the fault detection is carried out according to the statistical characteristics of the historical data of the engine in the past. Statistical reliability-based fault detection approaches include Bayesian methods, fuzzy logic, and so on.
Liu et al. [70] constructed a hierarchical polynomial-Dirichlet model and used it for conditional distribution estimation in Bayesian networks. The model has good performance in LRE fault classification with small datasets. Li et al. [71] devised an LRE component-level fault detection and isolation approach using the fuzzy model. The approach first divides the components of the engine, then establishes a fuzzy model of the normal operation of each component, and finally isolates the fault according to the setting of fault detection and isolation strategy. Dong etc. [72] and others combined Fuzzy C-means (FCM) clustering and fuzzy transitive closure method in fault detection of LRE. This method only needs a small number of normal prior samples to be fast and accurate. of detected failures. Li [73] applied the FCM method to the fault detection of the hydrogen-oxygen rocket engine starting process. Compared with the red line system, this method can detect the faults of the engine starting process earlier. Peng et al. [74,75] proposed a cloud classifier-based LRE fault detection approach.

3.3.3. Fault Detection Approach Using Neural Network

A neural network is an abstraction and simulation of several basic properties of the human brain. Neural networks can autonomously learn from samples to find the mapping relationship between sample data, which is currently the most widely used intelligent fault detection approach. Neural networks can overcome the problem that expert systems cannot detect new faults well, and has strong fault-tolerant ability and a good application prospect in the field of fault detection. However, when the neural network is used for fault detection, it relies too much on the empirical knowledge learned from historical data which makes it easy to fall into the problem of local optimum or overfitting. Komlev et al. [76] proposed a fault prediction and diagnosis algorithm for LRE based on artificial neural network(ANN). The algorithm uses the function approximation method and assign the corresponding appropriate proportion to the feature which contains the most useful information to assign the corresponding appropriate proportion, and then compares the calculated results with the reference values of each working condition to detect the state of LRE. However, when the input feature are closely related, the algorithm is difficult to determine their respective weights, the need for a large amount of data support also lowers the speed of detection. Eliseev et al. [77] proposes a LRE state diagnosis model based on recurrent neural network (RNN), which can diagnose and predict the technical state of the system by inputting key control parameters. There are generally three ways to apply neural networks to LRE fault detection and diagnosis [1,2]:
(1)
A cluster-based approach to engine fault detection. For example, Zhang et al. [78] used back propagation (BP) neural network for fault detection of liquid rocket booster delivery system.
(2)
By comparing the measured signal of the engine sensor with the estimated signal of the trained neural network model, the residual values are obtained, and the fault detection is performed according to the residual value. Zhao etc. [79] devised a recursive structure recognition algorithm for fault diagnosis with a feedforward neural network. Flora et al. [80] proposed a neural network-based algorithm for fault detection, isolation and replacement of LRE sensors.
(3)
Feed the engine failure samples into the neural network, learn and establish the correspondence between failure modes and feature parameters, and then apply the similarity measure to the new data samples to obtain the results of engine failure separation.

3.3.4. Fault Detection Approach Using Support Vector Machine

The fault detection approach based on neural network is easy to lead to local optimality. Support vector machine (SVM) can avoid these problems of neural network to a certain extent, so it received certain attention and research.
Yang [81] studied the SVM turbo pump fault detection algorithm using time domain features and frequency domain features. Tao et al. [82] proposed an adaptive fault detection algorithm using wavelet transform and SVM for real-time fault detection of LRE turbo pump. Tao et al. [83] proposed a fault detection algorithm using salient frequency component root mean square and SVM for real-time fault detection of LRE turbo pump. Yuan et al. [84] used boundary sample-based SVM for real-time fault detection of LRE turbo pump. Xiong etc., [85] studied the failure prediction of liquid oxygen methane engine based on a support vector regression machine. Hu et al. [52] used a one-class support vector machine (OCSVM) to analyze the historical vibration signal of the turbo pump, and proposed a method to detect and diagnose faults online, and also to connect the sensor fault with the turbo pump. The pump actually failed to separate.

3.3.5. Fault Detection Approach Using Support Vector Machine

Deep learning mainly involves three types of methods: convolutional neural network (CNN), self-encoding neural network and deep belief network. Soon-Young et al. [86] proposed a deep learning-based fault detection and diagnosis method for the start-up phase of LRE using long-term and short-term memory and CNN. Due to the limitation of lack of fault data in reality, Lv et al. [14] proposed a supervised recognition framework without fault samples. First, a negative sample generation method is proposed, and then a fused recurrent convolutional neural network is constructed for data fusion and feature extraction. Wang et al. [87] proposed an intelligent detection method of LRE health state using Convolutional Auto-Encoder (CAE). Zhu et al. [88] proposed a LRE steady-state stage fault detection approach using CAE and OCSVM.
At present, the application of deep learning in the fault detection of LRE requires a large amount of data support, and the training and operation of such methods put forward high requirements on computing hardware. There are still some difficulties in applying deep learning methods to practical LRE fault detection engineering applications, such as computing resources are limited while real-time computing is required.

3.3.6. Hybrid Fault Detection Approach

As mentioned above, in the study of LRE intelligent fault detection approaches, different types of fault detection approaches have their own advantages and disadvantages. In order to overcome the influence of related shortcomings, researchers combine fault detection approaches with intelligent optimization algorithms and uncertainty theory [89]. The improved fault detection approach can effectively improve the time sensitivity and accuracy of LRE fault detection. Good results have been achieved in the application.
The fault detection approaches combined with intelligent algorithms are mostly combined with swarm intelligence optimization algorithms, and the corresponding swarm intelligence algorithms are used to optimize the parameters of the relevant fault detection approaches, such as the weights and biases of the neural network, so that the corresponding fault detection approaches can be more efficient. Moreover, quick convergence, or better space can be achieved. Particle swarm optimization (PSO) and genetic algorithm are two commonly used swarm intelligence optimization algorithms.
The PSO algorithm is a random search algorithm using group cooperation developed by imitating bird flock foraging. Xu et al. [90] used the improved PSO algorithm to optimize the weights of the wavelet neural network (WNN), and established an LRE fault detection model combining the improved PSO algorithm and WNN, the local convergence ability and more accurate fault prediction ability. Wu et al. [91] established a fault detection approach using PSO optimized least squares support vector machine (LSSVM), and compared the predicted change value of the detected component obtained by the prediction model with the standard threshold to judge the engine and whether a failure occurs. Li et al. [92,93,94] proposed an LRE fault detection approach using the improved PSO optimized BP neural network, and applied the method for the fault detection of the LRE steady-state process. The simulation results show that the convergence speed of the proposed method is obviously high. Because of the BP neural network, and not trapped in local extreme values, the accuracy of fault detection is also significantly improved. In addition, Li et al. [94] combined PSO with LSSVM, used PSO to optimize the kernel function parameters and penalty factor of LSSVM, and established an LRE fault detection model. Su et al. [95] used the improved PSO algorithm to optimize the Bayesian network model.
Genetic algorithm is a search algorithm that draws on biological evolution. Xu et al. [96] designed an LRE fault detection algorithm using quantum genetic algorithm to optimize BP neural network. Zhang et al. [97] combined genetic algorithm with neural network, genetic algorithm, and FCM for fault diagnosis of a certain type of liquid rocket engine.
Combined with uncertainty theories such as dynamic cloud models, the randomness and ambiguity of uncertainty theory can be synthesized. Yao et al. [98] combined cloud model and BP neural network, and proposed a LRE fault diagnosis approach using a dynamic cloud BP network. Huang [99] combined cloud theory and neural network to establish a cloud-Petri network fault diagnosis method for LRE fault diagnosis, and verified the algorithm on the test data of a certain type of liquid oxygen kerosene engine. It can effectively isolate and diagnose the typical faults of the engine oxygen turbo pump. Huang et al. [100] combined fuzzy sets with neural networks and proposed a fuzzy directional neural network for LRE fault detection and isolation.
Although the development of artificial intelligence-based LRE fault detection approaches has achieved a lot of research results, it is still mainly affected by the scale and quality of sensor data. The features are not fully grasped, and the collection, classification, and processing of engine sensor data are not standardized and unified, which limit the training of fault detection models [38].

4. Conclusions

The method of rocket engine fault detection has changed from single algorithm detection and traditional sensor-based diagnosis to multi-algorithm fusion detection and intelligent method-based fault prediction during the development of LRE health monitoring system. Researchers hoped that the working state of the engine can be judged in advance to prepare for the decision of the engine. There is still a certain gap between China’s current launch vehicle engine health management system and related foreign technologies. The fault detection technology in LRE mainly uses the red line algorithm. This method mainly monitors the engine health state according to the preset threshold and the threshold. The settings are mostly based on experience and are prone to false positives and false negatives. The rocket engine failure accidents in recent years show that China urgently needs a reliable engine health management system that can be applied in engineering. Although much research works on the development of the diagnostic system for the LRE have been done by many researchers from major aerospace powers, the methods used for diagnosing the technical conditions of rocket engines are mainly theoretical explorations and offline applications until now. The methods are usually the solutions to the special faults under certain flight circumstances, which cannot deal with complex fault conditions well. With the continuous development and improvement of related fault detection approaches, the related technical achievements that have been obtained are related to practical engineering applications. Combined, it has also become a practical problem that LRE health monitoring technology urgently needs to solve. In the near future, the technical route for solving the problem of real-time onboard predicting rocket engine failures has to draw on the experience of aeromotor health management technology. According to the LRE component modeling and system modeling, a reliable mathematical–physical model of LRE must be established. A model that can be carried onboard and is used for predicting the potential failures using the actual flight data. Moreover, the fault detection and fault degree should give comprehensive considerations, which are directly related with safe flying. Based on the results of exactly and timely diagnosing the technical problems of the LRE, the reconfiguration of flight control system of the rocket can be done very well and it will guarantee the mission capability.

Author Contributions

Conceptualization, T.W. and H.Y.; Methodology, T.W. and L.D.; Formal analysis, T.W., L.D. and H.Y.; Investigation, L.D., H.Y. and T.W.; Resources, T.W.; Data curation, L.D. and H.Y.; Writing—original draft preparation, T.W. and L.D.; Writing—review and editing, L.D. and T.W.; Visualization, L.D. and T.W.; Supervision, T.W.; Project administration, T.W.; Funding acquisition, T.W. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Data is contained within the article.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Ezell, W.; Barkhoudarian, S.; Cross, G. Health Monitoring of Reusable Rocket Engines. SAE Trans. 1992, 101, 175–182. [Google Scholar]
  2. Zhang, Y.; Wu, J.; Huang, M.; Zhu, H.; Chen, Q. Liquid-Propellant Rocket Engine Health-Monitoring Techniques. J. Propuls. Power 1998, 14, 657–663. [Google Scholar] [CrossRef]
  3. Das, D.; Padmanabhan, P.; Kumaresan, V.; Sudhakar, D.P. Case Study on Bearing Fault Diagnosis in Liquid Rocket Engine Using Envelope Detection Technique. In Advances in Mechanical and Materials Technology; Springer: Singapore, 2022; pp. 357–366. [Google Scholar]
  4. Dhital, D.; Lee, J.R.; Farrar, C.; Mascarenas, D. A Review of Flaws and Damage in Space Launch Vehicles: Motors and Engines. J. Intell. Mater. Syst. Struct. 2014, 25, 524–540. [Google Scholar] [CrossRef]
  5. Chen, H.; Gardner, B.M.; Grogan, P.T.; Ho, K. Flexibility Management for Space Logistics via Decision Rules. J. Spacecr. Rocket. 2021, 58, 1314–1324. [Google Scholar] [CrossRef]
  6. Li, C.; Ma, B. Technical Innovation of LH2/LOX Rocket Engines in China. Chin. Ann. Hist. Sci. Technol. 2021, 4, 160–182. [Google Scholar]
  7. Gordon, K.E. Analysis of Chinese Cryogenic Long March Launch Vehicles and YF-100 Liquid Rocket Engine; The Ohio State University: Columbus, OH, USA, 2018. [Google Scholar]
  8. Wang, W.B.; Gong, Y.B. Design and Development of 50-ton LOX/LH2 Rocket Engine. J. Propuls. Technol. 2021, 42, 1458–1465. [Google Scholar]
  9. Wu, J.; Zhu, X.; Cheng, Y.Q. Research Progress of Intelligent Health Monitoring Technology for Liquid-Propellant Rocket Engines. J. Propuls. Technol. 2022, 43, 7–19. [Google Scholar]
  10. Zhao, Y.P.; Wang, J.J.; Li, X.Y.; Peng, G.J.; Yang, Z. Extended least squares support vector machine with applications to fault diagnosis of aircraft engine. ISA Trans. 2020, 97, 189–201. [Google Scholar] [CrossRef]
  11. Yoshida, M.; Kimura, T.; Hashimoto, T.; Moriya, S.; Takada, S. Overview of research and development status of reusable rocket engine (book chapter). Chem. Rocket. Propuls. 2017, 905–931. [Google Scholar]
  12. Lozano-Tovar, P.C. Dynamic Models for Liquid Rocket Engines with Health Monitoring Application; Massachusetts Institute of Technology: Cambridge, MA, USA, 1998. [Google Scholar]
  13. Cha, J.; Ha, C.; Ko, S.; Koo, J. Application of fault factor method to fault detection and diagnosis for space shuttle main engine. Acta Astronaut. 2016, 126, 517–527. [Google Scholar] [CrossRef]
  14. Lv, H.; Chen, J.; Wang, J.; Yuan, J.; Liu, Z. A Supervised Framework for Recognition of Liquid Rocket Engine Health State Under Steady-State Process Without Fault Samples. IEEE Trans. Instrum. Meas. 2021, 70, 3518610. [Google Scholar] [CrossRef]
  15. Lorenzo, C.F.; Musgrave, J.L. Overview of rocket engine control. In AIP Conference Proceedings; American Institute of Physics: College Park, MD, USA, 1992; Volume 246, pp. 446–455. [Google Scholar]
  16. Song, Z.; Xie, Z.; Qiu, L.; Xiang, D.; Li, J. Prospects of sea launches for Chinese cryogenic liquid-fueled medium-lift launch vehicles. Chin. J. Aeronaut. 2021, 34, 424–437. [Google Scholar]
  17. Song, Z.; Wu, Y.; Xu, S.; Chen, X.; Xiao, Y. LM-8: The pioneer of long march rocket series on the innovations of commercialization and intelligence. J. Deep. Space Explor. 2021, 8, 3–16. [Google Scholar]
  18. Duyar, A.; Eldem, V.; Merrill, W.; Guo, T.H. Fault detection and diagnosis in propulsion systems-A fault parameter estimation approach. J. Guid. Control. Dyn. 1994, 17, 104–108. [Google Scholar] [CrossRef]
  19. Panossian, H.V.; Ewing, W.D. Real Time Failure Detection Algorithm for the Space Shuttle Main Engine. IEEE Control. Syst. 1997, 17, 16–23. [Google Scholar]
  20. Nemeth, E.; Division, R.; International, R. Health Management for Rocket Engines System. Rockwell International Corp.: Canoga Park, CA, USA, 1990. [Google Scholar]
  21. Bickford, R.; Malloy, D. Development of a Real-Time Turbine Engine Diagnostic System. In Proceedings of the 38th AIAA/ASME/SAE/ASEE Joint Propulsion Conference, Indianapolis, Indiana, 7–10 July 2002. [Google Scholar]
  22. Fiorucci, T.D.L., II; Reynolds, T. Advanced Engine Health Management Applications of the SSME Real-Time Vibration Monitoring System. In Proceedings of the 36th AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, Las Vegas, NV, USA, 16–19 July 2000. [Google Scholar]
  23. Hawman, M.W. Health Monitoring System for the SSME—Program Overview. In Proceedings of the 26th Joint Propulsion Conference, Orlando, FL, USA, 16–18 July 1990. [Google Scholar]
  24. Paris, D.E.; Trevino, L.C.; Watson, M.D. A framework for integration of IVHM technologies for intelligent integration for vehicle management. In Proceedings of the 2005 IEEE Aerospace Conference, Big Sky, MT, USA, 5–12 March 2005; pp. 3843–3852. [Google Scholar]
  25. Wu, J. Liquid-propellant Rocket Engines Health-monitoring—A Survey. Acta Astronaut. 2005, 56, 347–356. [Google Scholar] [CrossRef]
  26. Figueroa, F.; Schmalzel, J.; Walker, M.; Venkatesh, M.; Kapadia, R.; Morris, J.; Turowski, M.; Smith, H. Integrated system health management: Foundational concepts, approach, and implementation. In Proceedings of the AIAA Infotech: Aerospace Conference and AIAA Unmanned Unlimited Conference, Washington, DC, USA, 6–9 April 2009; p. 1915. [Google Scholar]
  27. Figueroa, F.; Schmalzel, J. Rocket Testing and Integrated System Health Management. In Condition Monitoring and Control for Intelligent Manufacturing; Wang, L., Gao, R.X., Eds.; Springer: London, UK, 2006; pp. 373–391. [Google Scholar]
  28. Dingjun, C. Research on Function Test and Fault Diagnosis Algorithm of Liquid Rocket Engine; Beijing University of Aeronautics and Astronautics: Beijing, China, 1995. [Google Scholar]
  29. Glenn, B.; Mingchu, G.; Minggui, Y.; Mingyu, Q. Automatic Adjustment of Liquid Rocket Motor; Zihang Publishing House: Beijing, China, 1995. [Google Scholar]
  30. Xue, W.; Zhang, Q.; Wu, X.P. Based on the ARMA Model for the Liquid Rocket Propulsion Fault Detection. Comput. Meas. Control. 2019, 27, 4–7. [Google Scholar]
  31. Liu, H.G.; Wei, P.F.; Xie, T.F.; Huang, Q.; Wu, J.J. Research of Real-Time Fault detection approach for Liquid Propellant Rocket Engines in Ground Test. J. Astronaut. 2007, 28, 1660–1663. [Google Scholar]
  32. Yang, E.; Zhang, Z.; Liu, G.Q. Development of an Engineering Application System for Condition Monitoring and Fault Diagnosis of YF-75 Engine. J. Propuls. Technol. 1997, 18, 65–71. [Google Scholar]
  33. Xie, G.J.; Hu, N.Q.; Wen, X.S.; Wu, J.J. Health Monitoring System of Turbopump. In Proceedings of the 41st AIAA/ASME/SAE/ASEE Joint Propulsion Conference & Exhibit, Tucson, Arizona, 10–13 July 2005. [Google Scholar]
  34. Bastrygin, K.I.; trofimov, A.A. The Measurement System, Monitoring, Control, and Diagnostics Parameters of the Rocket Engine. Meas. Moitoring Manag. Control. 2017, 3, 21. (In Russian) [Google Scholar] [CrossRef]
  35. Xu, Y.; Peng, Y.; Wang, H.T.; Wang, Z.Y.; Ren, Y.H. Design and Realization of Highly Reliable Logic Control Software for Pre-Launch Escape Command and Control System. Procedia Comput. Sci. 2020, 166, 269–273. [Google Scholar] [CrossRef]
  36. Song, Z.Y. The survey of launch vehicle long distance fault diagnosis technique. J. Astronaut. 2016, 37, 135–144. [Google Scholar]
  37. Omata, N.; Satoh, D.; Tsutsumi, S.; Kawatsu, K.; Abe, M. Model-based supervised sensor placement optimization to detect propellant leak in a liquid rocket engine. Acta Astronaut. 2022, 195, 234–242. [Google Scholar] [CrossRef]
  38. Zhang, C.; Tang, S.; Tang, K. Strategies of parameter fault detection for rocket engines based on transfer learning. J. Comput. Appl. 2020, 40, 2774–2780. [Google Scholar]
  39. Zhang, X.; Gao, Y.; Ma, D. Staged Startup Technology of High Thrust Staged Combustion LOX/Kerosene Rocket Engine. Missile Space Veh. 2020, 68–72. [Google Scholar]
  40. Li, J. Study on Fault Detection and Isolation Methods for LOX/Methane Rocket Engines; National University of Defense Technology: Changsha, China, 2015. [Google Scholar]
  41. Yablochko, M.A. Adaptive Algorithms for Emergency Protection for Liquid-Fuel Rocket Engines. Mach. Technol. 2014, 4, 42–46. (In Russian) [Google Scholar]
  42. Hu, L.; Hu, N.; Zhang, X.; Gu, F.; Gao, M. Novelty Detection Methods for Online Health Monitoring and Post Data Analysis of Turbopumps. J. Mech. Sci. Technol. 2013, 27, 1933–1942. [Google Scholar] [CrossRef]
  43. Deng, C.; Xue, W.; Zheng, M.; Ma, H. Study on Real-Time Diagnosis Method of the Main Stage Working Condition of Rocket Engine Based on Improved ARMA Model. Comput. Meas. Control. 2020, 28, 33–38. [Google Scholar]
  44. Chen, D. Design and Validation of Fault Diagnosis Algorithm for Hydrogen-Oxygen Combustion Cycle Engine; China Academy of Launch Vehicle Technology: Beijing, China, 2020. [Google Scholar]
  45. Zhao, W.; Guo, Y.; Yang, J.; Sun, H. Hardware-in-the-loop Simulation Platform for Fault Diagnosis of Rocket Engines. In Proceedings of the 2019 IEEE 10th International Conference on Mechanical and Aerospace Engineering (ICMAE), Brussels, Belgium, 22–25 July 2019. [Google Scholar]
  46. Jain, P.C. Vibration Environment and the Rockets. Procedia Eng. 2016, 144, 729–735. [Google Scholar] [CrossRef]
  47. de Sousa, N.; Caporicci, M.; Vandersteen, J.; Rojo-Laguna, J.I.; Saló, E.; Adell, T.; Auletta, G.; van Loon, J.J.W.A. Molecular impact of launch related dynamic vibrations and static hyper-gravity in planarians. Npj Microgravity 2020, 6, 25. [Google Scholar] [CrossRef]
  48. Zhang, L.; LI, C.; Zhang, X.; An, X.Y.; Li, W.; Ren, Z. Improvement of Adaptive Correlation Algorithm for Rocket Engine Fault Detection. Electron. Opt. Control. 2016, 23, 96–99. [Google Scholar]
  49. Tao, H. Study on Real-Time Fault Detection Algorithm for Liquid-Propellant Rocket Engine Turbopump; University of Electronic Science and Technology of China: Chengdu, China, 2012. [Google Scholar]
  50. Hong, T.; Huang, Z.; Zhong, F.L. Turbopump real-time fault detection approach based on maximum amplitude ratio of frequency band. J. Electron. Meas. Instrum. 2012, 26, 24–30. [Google Scholar] [CrossRef]
  51. Zhang, T.; Zhao, X.; Hu, R.; Fu, W. A method for fault diagnosis of power system of carrier rocket engine based on linear-quadratic receding time-domain algorithm. Proc. Inst. Mech. Eng. Part G J. Aerosp. Eng. 2022, 236, 739–752. [Google Scholar] [CrossRef]
  52. Tsutsumi, S.; Hirabayashi, M.; Sato, D.; Kawatsu, K.; Sato, M.; Kimura, T.; Hashimoto, T.; Abe, M. Data-driven Fault Detection in a Reusable Rocket Engine using Bivariate Time-series Analysis. Acta Astronaut. 2021, 179, 685–694. [Google Scholar] [CrossRef]
  53. Tsutsumi, S.; Hirabayashi, M.; Sato, D.; Sato, M.; Kimura, T.; Hashimoto, T. Fault Detection of a Reusable Rocket Engine using Phase Plane Trajectory of Feature Vectors. In Proceedings of the Annual Conference of the PHM Society, Scottsdale, AZ, USA, 23–26 September 2019. [Google Scholar]
  54. Feng, Z. Fault Diagnosis of LPRE Ground-testing Bed Based on PCA-SOM. Int. J. Hybrid Inf. Technol. 2014, 7, 379–396. [Google Scholar] [CrossRef]
  55. Aiswarya, N.; Suja Priyadharsini, S.; Moni, K.S. An Efficient Approach for the Diagnosis of Faults in Turbo Pump of Liquid Rocket Engine by Employing FFT and Time-domain Features. Aust. J. Mech. Eng. 2018, 16, 163–172. [Google Scholar] [CrossRef]
  56. Kamensky, K.V.; Martirosov, D.S. А method for current state monitoring of liquid rocket engine in stationary and transient modes. Aerosp. MAI J. 2021, 28, 46–53. [Google Scholar]
  57. Lang, R.; Pan, L. Fault Diagnosis of Aircraft Based on Support Vector Machine; National Defense Industry Press: Beijing, China, 2016; p. 192. [Google Scholar]
  58. Zhang, J.; Gong, Y.; Liu, Z.; Wang, W. Fault Simulation and Experimental Study on High-Thrust LOX/LH2 Rocket Engine. J. Deep. Space Explor. 2021, 8, 389–398. [Google Scholar]
  59. Zheng, D.Y.; Wang, H.Y.; Hu, J. Transient Characteristics of High-Thrust Oxygen/Hydrogen Rocket Engine. J. Propuls. Technol. 2021, 42, 1761–1769. [Google Scholar]
  60. Ye, S.; Chen, X.; Xiong, C. Thrust Fault Diagnosis of Launch Vehicle Engine Based on Linear-Quadratic Receding Horizon Algorithm. Astronaut. Syst. Eng. Technol. 2020, 4, 29–37. [Google Scholar]
  61. Kawatsu, K.; Tsutsumi, S.; Hirabayashi, M.; Sato, D. Model-based Fault Diagnostics in an Electromechanical Actuator of Reusable Liquid Rocket Engine. In AIAA Scitech 2020 Forum; American Institute of Aeronautics and Astronautics: Orlando, FL, USA, 2020. [Google Scholar]
  62. Kawatsu, K. PHM by Using Multi-Physics System-Level Modeling and Simulation for EMAs of Liquid Rocket Engine. In Proceedings of the 2019 IEEE Aerospace Conference, Big Sky, MT, USA, 2–9 March 2019. [Google Scholar]
  63. Cha, J.; Ko, S.; Park, S.Y.; Jeong, E. Fault Detection and Diagnosis Algorithms for Transient State of an Open-cycle Liquid Rocket Engine using Nonlinear Kalman Filter Methods. Acta Astronaut. 2019, 163, 147–156. [Google Scholar] [CrossRef]
  64. Lee, K.; Cha, J.; Ko, S.; Park, S.Y.; Jung, E. Fault Detection and Diagnosis Algorithms for an Open-cycle Liquid Propellant Rocket Engine using the Kalman Filter and Fault Factor Methods. Acta Astronaut. 2018, 150, 15–27. [Google Scholar] [CrossRef]
  65. Levochkin, P.S.; Martirosov, D.S.; Bukanov, V.T. Problems of Functional Diagnostics of Liquid-Propellant Rocket Engines. Mech. Eng. 2013, 1, 72–88. (In Russian) [Google Scholar]
  66. Pekarskaya, I.V.; Golikovskaya, K.F. Features of Nonstationary Modes of Operation of Rocket Engines. Actual Probl. Aviat. Astronaut. 2017, 1, 172–173. (In Russian) [Google Scholar]
  67. Levochkin, P.S.; Martirosov, D.S.; Kamenskii, S.S.; Kozlov, A.A.; Borovik, I.N.; Belyaeva, N.V.; Rumyantsev, D.S. Liquid rocket engines functional diagnostics system in real-time mode. Aerosp. MAI J. 2019, 26, 147–154. [Google Scholar]
  68. Ali, M.; Gupta, U. An Expert System for Fault Diagnosis in a Space Shuttle Main Engine. In Proceedings of the 26th Joint Propulsion Conference, Orlando, FL, USA, 16–18 July 1990. [Google Scholar]
  69. Gupta, U.K.; Ali, M. Leader—An Integrated Engine Behavior and Design Analyses Based Real-Time Fault Diagnostic Expert System for Space Shuttle Main Engine (SSME); Association for Computing Machinery: New York, NY, USA, 1989. [Google Scholar]
  70. Liu, J.; Ding, X.; Wang, H.; Wang, B.; Liu, H.; Yang, Z.; Wang, Z. Fault Diagnosis of Liquid Rocket Engine Based on Hierarchical Bayesian Network Variational Inference. Trans. Beijing Inst. Technol. 2022, 42, 289–296. [Google Scholar]
  71. Li, Y.J.; Peng, X.H.; Cheng, Y.Q. Research of fault component isolation algorithms for LRE base on fuzzy model. J. Natl. Univ. Def. Technol. 2013, 35, 22–26. [Google Scholar]
  72. Dong, Z.; Guo, Y. Fault diagnosis of liquid rocket engine based on comprehensive fuzzy clustering algorithm. J. Aerosp. Power 2020, 35, 1326–1334. [Google Scholar]
  73. Li, Y. Study on Key Techniques of Fault Detection and Diagnosis for New Generation Large-Scale Liquid-Propellant Rocket Engines; National University of Defense Technology: Changsha, China, 2014. [Google Scholar]
  74. Peng, X.H.; Liu, Y.J.; Peng, Y.Q. A Real-Time Fault detection approach for Liquid-propellant Rocket Engines Based on the Cloud Association Rules. Manned Spacefl. 2013, 19, 84–90. [Google Scholar]
  75. Peng, X.H.; Liu, Y.J.; Cheng, Y.Q. Fault diagnosis method for liquid-propellant rocket engines based on the cloud sorter. J. Natl. Univ. Def. Technol. 2013, 35, 15–19. [Google Scholar]
  76. Komlev, G.V.; Mitrofanova, A.S. To the Question of Forecasting the Technical Condition of Low-Thrust Liquid Rocket Engines. Sib. Aerosp. J. 2020, 21, 70–77. (In Russian) [Google Scholar] [CrossRef]
  77. Eliseev, D.A.; Korolev, P.S. The system for determining the technical state of the liquid rocket engines based on a recurrent neural network. In Proceedings of the Xliii International Youth Scientific Conference “Gagarin Readings 2017”, Moscow, Russia, 5–19 April 2017; pp. 564–565. [Google Scholar]
  78. Zhang, W.; Li, J.; Hou, W.; Wang, G.; Luo, Y.; Xue, Z. BP neural network applied in fault detection of rocket pressurized delivery system. J. Rocket. Propuls. 2021, 47, 85–91. [Google Scholar]
  79. Zhao, W.; Guo, Y.; Yang, J.; Xue, W.; Wu, X. Design of liquid rocket engine fault diagnosis device and its HIL verification. J. Beijing Univ. Aeronaut. Astronaut. 2019, 45, 1995–2002. [Google Scholar]
  80. Flora, J.J.; Auxillia, D.J. Sensor Failure Management in Liquid Rocket Engine using Artificial Neural Network. J. Sci. Ind. Res. 2020, 79, 1024–1027. [Google Scholar]
  81. Hong, T.; Yang, S. Adaptive Algorithm Based on Wavelet and SVM for Turbopump Fault Detection. In Applied Mechanics and Materials; Tech Publications Ltd.: Wallerau, Switzerland; pp. 994–1002.
  82. Hong, T.; Li, H.; Zhong, F. Turbopump Fault Detection Algorithm Based on Wavelet RMS Vectors and SVM. In Proceedings of the 2012 Second International Conference on Electric Information and Control Engineering—(ICEICE ‘12), Washington, DC, USA, 4–6 April 2012; IEEE Computer Society: Washington, DC, USA, 2012; Volume 2, pp. 1950–1954. [Google Scholar]
  83. Hong, T.; Li, H. Turbopump Fault Detection Algorithm Based on Protruding Frequency Components RMS and SVM. In Proceedings of the 2013 IEEE International Conference on Mechatronics and Automation, Takamatsu, Japan, 4–7 August 2013. [Google Scholar]
  84. Yuan, S.F.; Chu, F.L. Support vector machines-based fault diagnosis for turbo-pump rotor. Mech. Syst. Signal Process. 2006, 20, 939–952. [Google Scholar] [CrossRef]
  85. Jingyu, X.; Jianjun, W.; Yuqiang, C. Simulation Based LOX/Methane Rocket Engine Fault Feature Analysis. Int. Congr. Exhib. Conditon Monit. Diagn. Eng. Manag. 2018. [Google Scholar]
  86. Park, S.; Ahn, J. Deep Neural Network Approach for Fault Detection and Diagnosis during Startup Transient of Liquid-propellant Rocket Engine. Acta Astronaut. 2020, 177, 714–730. [Google Scholar] [CrossRef]
  87. Wang, J.; Lv, H.; Chen, J.; Liu, Z.; Yuan, J. Intelligent detection method of liquid rocket engine health state. J. Rocket. Propuls. 2021, 47, 52–58. [Google Scholar]
  88. Zhu, X.; Cheng, Y.; Wu, J.; Hu, R.; Cui, X. Steady-State Process Fault Detection for Liquid Rocket Engines Based on Convolutional Auto-Encoder and One-Class Support Vector Machine. IEEE Access 2020, 8, 3144–3158. [Google Scholar] [CrossRef]
  89. Nie, Y.; Cheng, Y.; Wu, J. Liquid-propellant Rocket Engine Online Health Condition Monitoring Base on Multi-algorithm Parallel Integrated Decision-making. Proceedings of the Institution of Mechanical Engineers. Part G J. Aerosp. Eng. 2017, 231, 1621–1633. [Google Scholar]
  90. Xu, L.; Ma, S.; Xue, W.; Li, N. Improved PSO and Optimized WNN for Liquid Rocket Engine Fault Detection. Aerosp. Control. 2021, 39, 74–80. [Google Scholar]
  91. Mohd Ghazali, M.H.; Rahiman, W. Vibration analysis for machine monitoring and diagnosis: A systematic review. Shock. Vib. 2021, 2021, 9469318. [Google Scholar] [CrossRef]
  92. Li, N.; Xue, W.; Guo, X.; Xu, L.; Wu, Y.; Yao, Y. Fault Detection in Liquid-Propellant Rocket Engines Based on Improved PSO-BP Neural Network. J. Softw. 2019, 14, 380–387. [Google Scholar] [CrossRef]
  93. Li, N.; Xue, W.; Zhao, S.; Gao, Q.; Li, D.; Xu, L. Application of Improved PSO-BP Neural Network in Fault Detection of Liquid-Propellant Rocket Engine. In Proceedings of the 2018 IEEE International Conference on Mechatronics and Automation (ICMA), Changchun, China, 5–8 August 2018. [Google Scholar]
  94. Li, N.; Xue, W.; Guo, X.; Zhao, S.B.; Xu, L. Fault detection of liquid-propellant rocket engines based on LSSVM. In Proceedings of the 2018 International Conference on Information, Electronic and Communication Engineering (IECE 2018), Beijing, China, 28–29 October 2018; pp. 104–110. [Google Scholar] [CrossRef]
  95. Su, J.; Zhang, H.; Ling, C.X.; Matwin, S. Discriminative parameter learning for bayesian networks. In Proceedings of the 25th International Conference on Machine Learning, Helsinki, Finland, 5–9 July 2008; pp. 1016–1023. [Google Scholar]
  96. Xu, L.; Zhao, S.; Li, N.; Gao, Q.; Wang, T.; Xue, W. Application of QGA-BP for fault detection of liquid rocket engines. IEEE Trans. Aerosp. Electron. Syst. 2019, 55, 2464–2472. [Google Scholar] [CrossRef]
  97. Zhang, W.; Tian, G.; Xu, Z.; Yang, Z. Failure Characteristics Analysis and Fault Diagnosis for Liquid Rocket Engines; Springer: Cham, Switzerland, 2016. [Google Scholar]
  98. Nie, Y.; Cheng, Y.; Wu, J. Dynamic Cloud Back-propagation Networks and its Application in Fault Diagnostic for Liquid-propellant Rocket Engines. Proc. Inst. Mech. Eng. Part G J. Aerosp. Eng. 2018, 232, 583–594. [Google Scholar] [CrossRef]
  99. Huang, Q. Study on the Techniques of Fault Detection and Diagnosis for High Pressure Staged Combustion LOX/Kerosene Rocket Engine; National University of Defense Technology: Changsha, China, 2012. [Google Scholar]
  100. Huang, M.; Xing, B. LRE Fault Detection and Isolation Based on Fuzzy Direction Neural Network. Appl. Mech. Mater. 2015, 727–728, 880–883. [Google Scholar] [CrossRef]
Aerospace 09 00481 g001 550
Figure 1. Fault detection approaches for LRE.
Figure 1. Fault detection approaches for LRE.
Aerospace 09 00481 g001
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Wang, T.; Ding, L.; Yu, H. Research and Development of Fault Diagnosis Methods for Liquid Rocket Engines. Aerospace 2022, 9, 481. https://doi.org/10.3390/aerospace9090481

AMA Style

Wang T, Ding L, Yu H. Research and Development of Fault Diagnosis Methods for Liquid Rocket Engines. Aerospace. 2022; 9(9):481. https://doi.org/10.3390/aerospace9090481

Chicago/Turabian Style

Wang, Tao, Lin Ding, and Huahuang Yu. 2022. "Research and Development of Fault Diagnosis Methods for Liquid Rocket Engines" Aerospace 9, no. 9: 481. https://doi.org/10.3390/aerospace9090481

APA Style

Wang, T., Ding, L., & Yu, H. (2022). Research and Development of Fault Diagnosis Methods for Liquid Rocket Engines. Aerospace, 9(9), 481. https://doi.org/10.3390/aerospace9090481

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop