A self-diagnosis method and system for avionics system fault

Abstract

The present invention relates to the field of fault self-diagnosis technology, and in particular to a method and system for self-diagnosis of faults in an avionics system. The method comprises the following steps: dynamically sampling fault modules of an avionics system, and performing temporal data association between monitoring modules to generate temporal association data between modules; modeling module time-varying causal relationships based on temporal association data between modules to generate causal network data between modules; performing fault confidence calculation using the causal network data between modules to obtain fault location confidence data; performing intelligent fault diagnosis processing on the avionics system using the fault location confidence data to generate system fault self-diagnosis report data. The present invention realizes rapid positioning and intelligent diagnosis of fault modules by tracing the fault propagation path, and significantly improves the diagnostic accuracy and fault handling efficiency for the intelligent control system integration of the avionics system, thereby ensuring the efficient and safe operation of the avionics system.

Description

Technical Field

[0001] The present invention relates to the technical field of fault self-diagnosis, and in particular to a method and system for self-diagnosis of faults of an avionics system. Background Art

[0002] A large number of electronic devices are integrated on modern aircraft. These devices are interconnected and influence each other, forming a complex electronic network system. Once a module fails, its impact may spread to other modules along the signal transmission path, resulting in a series of chain reactions, and finally showing complex fault phenomena. For example, when the output signal of the upstream module is abnormal, the downstream module will show abnormal behavior due to the error of the received input signal. In this case, if the diagnosis is only based on the abnormal performance of the downstream module, it is easy to misdiagnose the downstream module as the source of the fault, while ignoring that the real root cause of the fault lies in the upstream module. This misdiagnosis will lead to unnecessary maintenance and replacement operations, which not only increases maintenance costs, but may also delay troubleshooting and even cause more serious system failures. However, most of the existing avionics system fault diagnosis methods rely on the internal self-diagnosis mechanism of a single module. It is difficult to accurately locate the faults caused by the interaction between modules, especially when there is complex data interaction between multiple modules. There is a lack of effective cross-verification capability for faults between multiple modules, resulting in incorrect maintenance instructions and even more serious system safety issues. Summary of the invention

[0003] Based on this, the present invention provides an avionics system fault self-diagnosis method and system to solve at least one of the above technical problems.

[0004] To achieve the above object, a method for self-diagnosis of avionics system faults includes the following steps:

[0005] Step S1: Analyze the characteristics of the electronic monitoring module of the avionics system to generate characteristic parameters of the monitoring module; dynamically configure the sampling rate according to the characteristic parameters of the monitoring module to obtain a sampling strategy for the monitoring module; dynamically sample the fault module using the sampling strategy for the monitoring module to generate original fault module monitoring data;

[0006] Step S2: Slice the original fault module monitoring data in time series to generate time series fault module monitoring slice data; associate the time series data between monitoring modules according to the time series fault module monitoring slice data to generate time series association data between modules;

[0007] Step S3: Modeling the module time-varying causal relationship according to the inter-module timing correlation data to generate inter-module causal relationship network data; using the inter-module causal relationship network data to perform fault response verification processing to generate fault response feature data; performing fault confidence calculation on the fault response feature data to obtain fault location confidence data;

[0008] Step S4: locate the fault module of the avionics system through the fault location confidence data to obtain the target fault module; perform fault module fuzzification processing on the target fault module to generate a fault self-diagnosis fuzzy decision model; use the fault self-diagnosis fuzzy decision model to perform intelligent fault diagnosis processing to generate system fault self-diagnosis report data.

[0009] The present invention analyzes the characteristics of the electronic monitoring module of the avionics system, obtains the operating characteristic parameters of each module, and dynamically configures the sampling rate according to these parameters, thereby realizing accurate dynamic sampling of the faulty module. This dynamic sampling strategy avoids unnecessary data collection of the entire system, greatly reduces the burden of data processing and improves the diagnostic efficiency while ensuring the validity of the data. Compared with the traditional fixed sampling rate method, dynamic sampling can adaptively adjust the sampling frequency according to the changes in the system operating state, so as to more effectively capture the key information related to the fault. After obtaining the original data, the continuous monitoring data is divided into discrete time segments by using time series slicing. By analyzing the monitoring slice data of the timing fault module, the state changes of each module in different time periods can be identified, the timing correlation data between modules is constructed, and the timing relationship of signal transmission between modules is clearly displayed. Based on the timing correlation data between modules, a causal relationship network between modules is constructed, which can clearly display the signal transmission path and mutual influence relationship between modules. By analyzing the causal relationship network, the propagation path of the fault can be accurately tracked and the real root cause of the fault can be identified. Even if the fault of the upstream module causes the abnormality of the downstream module, the upstream module can be accurately identified as the fault source, avoiding misdiagnosis of the downstream module. This is particularly important for complex multi-module interactive systems, and effectively solves the problem that traditional methods are difficult to handle fault propagation between modules. By simulating the propagation process of the fault in the system, the accuracy of the causal relationship network is verified, and fault response feature data is generated. The fault confidence is calculated for the fault response feature data to obtain the fault location confidence data, which further improves the reliability of fault diagnosis. The fault module of the avionics system is located by the fault location confidence data, and the target fault module is fuzzified to generate a fault self-diagnosis fuzzy decision model. The model is used for intelligent fault diagnosis and processing, and system fault self-diagnosis report data is generated, providing maintenance personnel with more comprehensive and accurate fault information, assisting them in more effective troubleshooting and maintenance work. The introduction of the fuzzy decision model effectively handles various uncertainties and fuzzy factors in the avionics system, such as sensor data noise, module performance degradation, etc., thereby improving the robustness and reliability of fault diagnosis. By integrating the fault self-diagnosis intelligent control system into the avionics system in civil aircraft, the diagnostic accuracy, cost control and fault handling efficiency are significantly improved, and the efficient and safe operation of the intelligent control of the avionics system in civil aircraft is realized. Therefore, the avionics system fault self-diagnosis method of the present invention realizes accurate fault location and intelligent diagnosis by real-time monitoring of module characteristic parameters, dynamically configuring the sampling rate, and using the time series correlation data between fault modules to build a causal relationship network, thereby significantly improving the fault response speed and processing efficiency, and realizing effective self-diagnosis of cross-module faults of the avionics system.

[0010] Preferably, step S1 comprises the following steps:

[0011] Step S11: performing aviation operation status identification on the avionics system to generate aviation operation status data; performing electronic monitoring module characteristic analysis on the avionics system to generate monitoring module characteristic parameters;

[0012] Step S12: dynamically configure the sampling rate according to the aviation operation status data and the characteristic parameters of the monitoring module to obtain the monitoring module sampling strategy;

[0013] Step S13: performing initial sampling of the monitoring module status according to the monitoring module sampling strategy to generate initial aviation monitoring module status data;

[0014] Step S14: performing abnormal state detection on the initial aviation monitoring module state data through a preset module monitoring state threshold to obtain abnormal module detection mark data;

[0015] Step S15: Mark the suspicious monitoring module according to the abnormal module detection mark data to obtain the electronic suspected fault module; perform data interaction module discrimination on the avionics system according to the electronic suspected fault module to obtain the electronic adjacent monitoring module and other electronic monitoring modules respectively;

[0016] Step S16: Based on the preset monitoring frequency adjustment data, the monitoring module sampling strategy is used to perform high-frequency sampling adjustment on the electronic suspected fault module to generate key module monitoring data; the electronic adjacent monitoring module is performed medium-frequency sampling adjustment to generate associated module monitoring data; the other electronic monitoring modules are performed low-frequency sampling adjustment to generate basic module monitoring data;

[0017] Step S17: integrating the key module monitoring data, the associated module monitoring data and the basic module monitoring data, and performing timestamp processing to generate original fault module monitoring data.

[0018] The present invention can identify the operating status of the aircraft, and dynamically configure the sampling rate in combination with the characteristic parameters of the electronic monitoring module. This means that in different flight phases (such as take-off, cruising, landing), and in the face of different module importance, the system will automatically adjust the frequency of data collection. This dynamic adjustment strategy avoids unnecessary high-frequency sampling of the entire system, thereby reducing the pressure of data storage and processing, and reducing the system load. Through the preset module monitoring state threshold, the system can quickly identify potential faulty modules and mark them as suspected faulty modules. For these suspicious modules, the system will adopt a high-frequency sampling strategy to capture their state changes more finely, thereby improving the accuracy of fault diagnosis. At the same time, in order to analyze the impact range of the faulty module, the system will also perform medium-frequency sampling on adjacent modules that interact with the suspicious module data, and low-frequency sampling on other modules. This hierarchical monitoring strategy effectively reduces the overall data collection volume while ensuring the integrity of key data collection.

[0019] Preferably, step S12 comprises the following steps:

[0020] Step S121: dividing the aviation operation status data into flight mission stages to obtain mission stage characteristic data;

[0021] Step S122: dynamically evaluate the system criticality according to the monitoring module characteristic parameters and the task phase characteristic data to obtain module priority data;

[0022] Step S123: Allocate monitoring sampling resources according to the module priority data to generate monitoring resource configuration data;

[0023] Step S124: Optimize the module sampling rate of the monitoring resource configuration data in real time, and dynamically generate the sampling strategy to obtain the monitoring module sampling strategy.

[0024] The present invention divides the aviation operation status data into flight mission phases and obtains characteristic data of different flight phases. This means that the system can adjust the monitoring strategy according to the characteristics of different flight phases (such as take-off, climb, cruise, descent, landing, etc.). The importance of each module of the avionics system is different in different flight phases, so it is necessary to allocate monitoring resources in a targeted manner. Combined with the characteristic parameters of the monitoring module and the characteristic data of the mission phase, the system criticality is dynamically evaluated to obtain the module priority data. This means that the system can identify which modules are critical to flight safety in a specific flight phase, so as to allocate more monitoring resources to these key modules. This priority-based resource allocation strategy ensures the monitoring accuracy of key modules and improves the detection sensitivity of key faults. The system can dynamically adjust the sampling rate, data storage space and other resources of each module according to the importance of the module. Important modules will obtain higher sampling rates and more storage space to ensure the integrity and accuracy of the data, while secondary modules will allocate fewer resources to avoid resource waste. According to the real-time situation, the sampling strategy is adjusted to adapt to the changes in flight status and fault conditions. For example, when a module is abnormal, the system can automatically increase the sampling rate of the module to monitor its state changes in more detail.

[0025] Preferably, step S2 comprises the following steps:

[0026] Step S21: preprocessing the original fault module monitoring data to obtain preprocessed fault module monitoring data;

[0027] Step S22: performing time series slicing according to the pre-processed fault module monitoring data to generate time series fault module monitoring slice data;

[0028] Step S23: Calculate the monitoring parameter change rate of the sequential fault module monitoring slice data, determine the adaptive time window, and generate adaptive time window data;

[0029] Step S24: performing inter-module data association screening on the pre-processed fault module monitoring data to generate associated module data pairs;

[0030] Step S25: performing inter-module signal delay calculation on the associated module data pair through the adaptive time window data to generate inter-module delay characteristic data;

[0031] Step S26: Perform signal propagation analysis based on the inter-module delay characteristic data, and monitor the correlation of timing data between modules to generate inter-module timing correlation data.

[0032] The present invention preprocesses the original fault module monitoring data, such as removing noise, filling missing values, etc., to improve the quality and reliability of the data. Data preprocessing is a prerequisite for any data analysis. It can eliminate outliers and noise in the data and ensure the accuracy of subsequent analysis results. The preprocessed data is sliced ​​into time series to divide the continuous monitoring data into discrete time segments. Time series slicing can better show the changing trend of module status over time and help capture transient fault information. The concept of adaptive time window is introduced. By calculating the monitoring parameter change rate of the time series fault module monitoring slice data, the size of the time window can be adaptively determined. This means that the system can automatically adjust the width of the time window according to the change of the data, so as to more accurately capture the time period when the fault occurs. In the stage where the parameters change drastically, a smaller time window can be used to observe the changes in the data more finely; while in the stage where the parameters change slowly, a larger time window can be used to reduce the amount of calculation and improve efficiency. The preprocessed fault module monitoring data is screened for inter-module data association, and the associated module data pairs with data interaction relationships are extracted. By analyzing these associated module data pairs, we can better understand the mutual influence between modules and help identify the propagation path of the fault. Adaptive time window data is used to calculate the signal delay between modules for the associated module data pairs, and the inter-module delay feature data is generated. These delay feature data reflect the time difference of signal transmission between modules, which is the key information for analyzing the fault propagation path and building the causal relationship network.

[0033] Preferably, step S3 comprises the following steps:

[0034] Step S31: segmenting the inter-module timing correlation data to obtain normal module status data and fault module status data respectively;

[0035] Step S32: sorting the electronic suspected faulty modules according to the faulty module status data to obtain faulty module sorting data;

[0036] Step S33: Modeling the module time-varying causal relationship of the faulty module sorting data using the original faulty module monitoring data and the normal module status data to generate inter-module causal relationship network data;

[0037] Step S34: using the inter-module causal relationship network data to perform fault response verification processing on the fault module sorting data to generate fault response feature data;

[0038] Step S35: Calculate the fault confidence of the fault response characteristic data according to the fault module status data and the causal relationship network data between modules to obtain the fault location confidence data.

[0039] The present invention divides the inter-module time-series correlation data into normal module state data and fault module state data, which respectively represent the inter-module interaction information in the normal operation state and fault state of the system. This data separation strategy helps to more clearly identify the impact of the fault on the system and build a more accurate causal relationship model. The electronic suspected fault modules are sorted using the fault module state data to obtain the fault module sorting data. This means that the system can evaluate the possibility of the fault according to the performance of the fault module in the fault state, so as to prioritize the fault modules with greater possibility, thereby improving the efficiency of fault diagnosis. The fault module sorting data is modeled by the module time-varying causal relationship using the original fault module monitoring data and the normal module state data, and the inter-module causal relationship network data is generated. The network data describes the causal relationship between modules at different time points, can more accurately reflect the propagation path and impact range of the fault, and provide a more reliable basis for fault location. Compared with the traditional static causal relationship model, the time-varying causal relationship model can better capture the dynamic changes of the system, thereby more accurately identifying the root cause of the fault. By simulating the propagation process of the fault in the system, the accuracy of the causal relationship model can be verified, and interference factors can be further eliminated to improve the reliability of fault diagnosis. Fault confidence reflects the reliability of fault location results, providing maintenance personnel with more comprehensive reference information and helping them make more accurate maintenance decisions.

[0040] Preferably, step S33 includes the following steps:

[0041] Step S331: extracting module benchmark response characteristics from the original fault module monitoring data to generate module benchmark response characteristic data;

[0042] Step S332: using the fault module sorting data to perform Granger causality test on the temporal correlation data between modules to generate fault module causal relationship data;

[0043] Step S333: constructing a causal network model according to the fault module causal relationship data to generate fault module network structure data;

[0044] Step S334: performing time-varying causal strength calculation on the faulty module network structure data based on the normal module state data, the faulty module state data and the module reference response characteristic data to generate time-varying causal strength data;

[0045] Step S335: performing fault propagation path analysis according to the time-varying causal strength data, and performing causal relationship direction processing to generate fault propagation path data;

[0046] Step S336: quantify the fault propagation path abnormality weight of the fault propagation path data, and reconstruct the causal link using the fault module network structure data to generate inter-module causal relationship network data.

[0047] The present invention extracts the module benchmark response characteristics from the original fault module monitoring data, which provides a benchmark reference for the subsequent causal strength calculation, making the quantification of causal relationship more accurate. The Granger causality test is performed on the inter-module time series correlation data using the fault module sorting data. The Granger causality test can determine whether the change of one time series has a predictive effect on the change of another time series, thereby determining the causal relationship between the modules. A causal network model is constructed based on the fault module causal relationship data, and the time-varying causal strength is calculated by combining the normal module state data, the fault module state data and the module benchmark response feature data. The time-varying causal strength not only reflects the strength of the causal relationship between the modules, but also considers the change of the causal relationship over time, thereby more accurately describing the dynamic behavior of the system. The fault propagation path analysis is performed based on the time-varying causal strength data, and the causal relationship direction is processed, which enables the system to clearly show how the fault propagates in the system, thereby locating the fault source more quickly. The abnormal degree of the propagation path is weighted quantified, which can highlight the abnormal propagation path, thereby more effectively identifying the root cause of the fault. The causal link reconstruction can further optimize the causal relationship network so that it can more accurately reflect the true relationship between the modules.

[0048] Preferably, step S34 includes the following steps:

[0049] Step S341: performing module safety threshold evaluation on the fault module sorting data, and setting control constraint conditions to generate safety control constraint condition data;

[0050] Step S342: setting a reverse excitation signal according to the safety control constraint condition data to obtain reverse excitation strategy data;

[0051] Step S343: using the inter-module causal relationship network data to identify the faulty downstream modules of the faulty module sorting data, and obtaining the faulty downstream module identification data;

[0052] Step S344: applying a control signal to the fault downstream module identification data using the reverse excitation strategy data, and monitoring the module response data to generate downstream module response data;

[0053] Step S345: Analyze the upstream module status of the electronic suspected fault module by using the causal relationship network data between modules to obtain the upstream fault module traceability data;

[0054] Step S346: Perform fault response verification processing on the upstream fault module traceability data and the downstream module response data to generate fault response feature data.

[0055] The present invention performs safety threshold assessment and control constraint setting on the fault module to ensure that no secondary damage is caused to the system during the verification process. This reflects the safety and reliability of the solution and avoids new problems caused by diagnostic operations. According to the set safety control constraint, the reverse excitation signal is set. By applying a control signal to the faulty downstream module and monitoring its response data, the influence of the faulty module on its downstream module can be verified, thereby determining the causal relationship between them. This reverse verification method can effectively avoid misdiagnosing the fault of the downstream module as a root fault. The upstream module state analysis of the electronic suspected fault module is performed using the causal relationship network data between modules to trace the source of the fault. By analyzing the state of the upstream module, it can be determined whether the upstream module is the real root cause of the fault, and further verify the conclusion of the reverse excitation. Combined with the upstream fault module traceability data and the downstream module response data, the fault response verification process is performed to generate fault response feature data. This two-way verification strategy can more comprehensively analyze the causal relationship of the fault and improve the accuracy and reliability of fault diagnosis.

[0056] Preferably, step S35 includes the following steps:

[0057] Step S351: performing time domain decomposition on the fault module state data to obtain time domain feature data of the fault module;

[0058] Step S352: performing frequency domain feature analysis on the fault module state data based on the fault module time domain feature data to generate the fault module frequency domain feature data;

[0059] Step S353: extracting causal domain features of the fault module according to the causal relationship network data between modules, and generating causal domain feature data of the fault module;

[0060] Step S354: performing fault feature consistency assessment based on the fault module time domain feature data, the fault module frequency domain feature data, and the fault module causal domain feature data, and performing aviation state weight optimization to generate fault feature weight data;

[0061] Step S355: Calculate the fault location reliability of the fault response feature data using the fault feature weight data to obtain the fault location reliability data.

[0062] The present invention performs time domain decomposition on the fault module state data, and extracts the time domain feature data of the fault module, such as the time, duration, and change trend of the fault. Based on the time domain feature data, the fault module state data is subjected to frequency domain feature analysis, and the frequency domain feature data of the fault module is generated, such as the frequency, amplitude, and phase of the fault signal. The frequency domain feature analysis can reveal the characteristics of the fault signal in the frequency domain, which is helpful to identify different types of faults. The causal domain features of the fault module are extracted according to the causal relationship network data between modules, such as the degree of influence of the fault module on other modules, the fault propagation path, etc. The causal domain features can reflect the propagation relationship of the fault in the system, which is helpful to locate the fault source more accurately. The extracted multi-domain features are evaluated for consistency, and the weights are optimized in combination with the aviation status to generate the fault feature weight data. This means that the system will comprehensively consider the time domain, frequency domain, and causal domain features, and assign different weights to different features according to the current aviation operation status. For example, in the critical flight phase, the weight of the causal domain features will be higher, because at this time, more attention needs to be paid to the impact of the fault on the entire system. The fault location confidence calculation is performed on the fault response feature data using the fault feature weight data, and more accurate fault location confidence data is obtained. This confidence calculation method based on multi-domain features and aviation status can more comprehensively evaluate the reliability of the fault location results, thereby improving the accuracy of fault diagnosis.

[0063] Preferably, step S4 comprises the following steps:

[0064] Step S41: locating the fault module of the avionics system by using the fault location reliability data to obtain the target fault module;

[0065] Step S42: searching the fault type of the target fault module through a preset fault type library to generate a module fault type data set;

[0066] Step S43: performing fault module fuzzification processing on the module fault type data set to generate fault module fuzzy feature data;

[0067] Step S44: constructing a fuzzy decision tree according to the fuzzy feature data of the fault module to generate a fault self-diagnosis fuzzy decision model;

[0068] Step S45: Obtain historical fault module diagnosis data; perform historical diagnosis accuracy statistics based on the historical fault module diagnosis data to obtain historical diagnosis accuracy data;

[0069] Step S46: performing accuracy performance trend analysis on historical diagnostic accuracy data, and performing adaptive diagnostic confidence threshold adjustment to obtain adaptive diagnostic threshold data;

[0070] Step S47: Optimize the decision threshold of the fault self-diagnosis fuzzy decision model through adaptive diagnosis threshold data, perform intelligent fault diagnosis processing on the target fault module, and generate system fault self-diagnosis report data.

[0071] The present invention uses fault location reliability data to accurately locate the target fault module, and retrieves the fault type of the module through a preset fault type library, thereby narrowing the scope of fault troubleshooting. The module fault type data set is fuzzified to generate fault module fuzzy feature data. Fuzzification can better handle the uncertainty and fuzziness of fault information and improve the robustness of the diagnosis model. A fuzzy decision tree is constructed based on the fuzzy feature data of the fault module, and a fault self-diagnosis fuzzy decision model is generated. The fuzzy decision tree can effectively handle complex fault conditions and perform reasoning and judgment based on the fuzzy feature data, thereby improving the accuracy of fault diagnosis. Historical fault module diagnosis data is introduced, and historical diagnosis accuracy is statistically analyzed and trended. By analyzing the accuracy of historical diagnosis data, the deficiencies of the diagnosis model can be identified, and a reference basis can be provided for the optimization of the model. According to the trend analysis of historical diagnosis accuracy, an adaptive diagnosis confidence threshold is adjusted to obtain adaptive diagnosis threshold data. This means that the system can automatically adjust the diagnosis confidence threshold according to the performance of historical diagnosis data, thereby improving the reliability of the diagnosis result.

[0072] Preferably, the present invention further provides an avionics system fault self-diagnosis system, which executes the avionics system fault self-diagnosis method as described above, and the avionics system fault self-diagnosis system comprises:

[0073] The system monitoring and acquisition module is used to analyze the characteristics of the electronic monitoring module of the avionics system and generate the characteristic parameters of the monitoring module; dynamically configure the sampling rate according to the characteristic parameters of the monitoring module to obtain the sampling strategy of the monitoring module; dynamically sample the fault module using the sampling strategy of the monitoring module to generate the original fault module monitoring data;

[0074] The time series data association module is used to perform time series slicing on the original fault module monitoring data to generate time series fault module monitoring slice data; perform time series data association between monitoring modules based on the time series fault module monitoring slice data to generate time series association data between modules;

[0075] The fault causal reasoning module is used to model the module time-varying causal relationship based on the inter-module temporal correlation data and generate the inter-module causal relationship network data; use the inter-module causal relationship network data to perform fault response verification processing and generate fault response feature data; calculate the fault confidence of the fault response feature data and obtain the fault location confidence data;

[0076] The fault self-diagnosis decision module is used to locate the fault module of the avionics system through the fault location confidence data to obtain the target fault module; perform fault module fuzzification processing on the target fault module to generate a fault self-diagnosis fuzzy decision model; use the fault self-diagnosis fuzzy decision model to perform intelligent fault diagnosis processing to generate system fault self-diagnosis report data. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 A schematic diagram of the steps of a self-diagnosis method for an avionics system fault according to the present invention;

[0078] Figure 2 for Figure 1 Detailed implementation steps of step S2 in the flowchart;

[0079] Figure 3 for Figure 1 Detailed implementation steps of step S4 in FIG.

[0080] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0081] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0082] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0083] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0084] To achieve this, please refer to Figures 1 to 3 The present invention provides an avionics system fault self-diagnosis method, comprising the following steps:

[0085] Step S1: Analyze the characteristics of the electronic monitoring module of the avionics system to generate characteristic parameters of the monitoring module; dynamically configure the sampling rate according to the characteristic parameters of the monitoring module to obtain a sampling strategy for the monitoring module; dynamically sample the fault module using the sampling strategy for the monitoring module to generate original fault module monitoring data;

[0086] Step S2: Slice the original fault module monitoring data in time series to generate time series fault module monitoring slice data; associate the time series data between monitoring modules according to the time series fault module monitoring slice data to generate time series association data between modules;

[0087] Step S3: Modeling the module time-varying causal relationship according to the inter-module timing correlation data to generate inter-module causal relationship network data; using the inter-module causal relationship network data to perform fault response verification processing to generate fault response feature data; performing fault confidence calculation on the fault response feature data to obtain fault location confidence data;

[0088] Step S4: locate the fault module of the avionics system through the fault location confidence data to obtain the target fault module; perform fault module fuzzification processing on the target fault module to generate a fault self-diagnosis fuzzy decision model; use the fault self-diagnosis fuzzy decision model to perform intelligent fault diagnosis processing to generate system fault self-diagnosis report data.

[0089] In the embodiment of the present invention, reference Figure 1 The above is a schematic flow chart of the steps of a method for self-diagnosis of avionics system faults of the present invention. In this embodiment, the method for self-diagnosis of avionics system faults includes the following steps:

[0090] Step S1: Analyze the characteristics of the electronic monitoring module of the avionics system to generate characteristic parameters of the monitoring module; dynamically configure the sampling rate according to the characteristic parameters of the monitoring module to obtain a sampling strategy for the monitoring module; dynamically sample the fault module using the sampling strategy for the monitoring module to generate original fault module monitoring data;

[0091] In the embodiment of the present invention, the avionics system of a certain type of aircraft is taken as an example. The system includes key subsystems such as a flight control system, a navigation system, and a communication system. Each subsystem is composed of multiple electronic monitoring modules, such as a flight control computer, an attitude sensor, an inertial navigation unit, etc. First, a characteristic analysis is performed on each electronic monitoring module, such as analyzing its function, importance, data update frequency, fault sensitivity, etc., to generate characteristic parameters of the monitoring module. For example, the flight control computer is given a high priority due to its importance, and its data update frequency is 100 times per second. Then, the sampling rate is dynamically configured according to these characteristic parameters. In different flight phases (such as takeoff, cruising, and landing), the importance of different modules is different, so the sampling rate needs to be dynamically adjusted. For example, in the takeoff phase, the sampling rate of the flight control system will be higher than that in the cruising phase. The generated monitoring module sampling strategy contains the sampling rate of each module in different flight phases. Finally, the system is dynamically sampled according to the monitoring module sampling strategy to collect data from each module. When the system detects a potential fault, it will dynamically adjust the sampling rate of related modules according to preset rules or algorithms. For example, it will increase the sampling rate of the suspected fault module and its related modules to monitor their state changes more finely and generate original fault module monitoring data, which includes the sampling data of each module and the corresponding timestamp.

[0092] Step S2: Slice the original fault module monitoring data in time series to generate time series fault module monitoring slice data; associate the time series data between monitoring modules according to the time series fault module monitoring slice data to generate time series association data between modules;

[0093] In an embodiment of the present invention, after the original fault module monitoring data is obtained, it is first necessary to slice the time series. The size of the slice window can be adjusted according to the specific situation, for example, it can be set to 1 second, 5 seconds or 10 seconds, so as to better analyze the module state changes in different time periods. For example, the original fault module monitoring data of the flight control computer is sliced ​​according to a 1-second window to generate a series of time series fault module monitoring slice data. Each slice contains various monitoring data of the flight control computer within 1 second. Then, according to the time series fault module monitoring slice data, the time series data between the monitoring modules are associated. For example, the output control signal of the flight control computer will affect the measurement value of the attitude sensor, so it is necessary to associate the data of the two modules. The association method can be determined according to the physical connection relationship or data transmission relationship between the modules. For example, according to the system architecture diagram of the aircraft, the connection relationship between the flight control computer and the attitude sensor can be determined, and according to the delay time of the signal transmission, the output control signal of the flight control computer at time t is associated with the measurement value of the attitude sensor at time t+Δt, where Δt represents the signal transmission delay, and the time series association data between the modules is generated, which contains the data association relationship and time sequence between different modules.

[0094] Step S3: Modeling the module time-varying causal relationship according to the inter-module timing correlation data to generate inter-module causal relationship network data; using the inter-module causal relationship network data to perform fault response verification processing to generate fault response feature data; performing fault confidence calculation on the fault response feature data to obtain fault location confidence data;

[0095] In an embodiment of the present invention, based on the inter-module temporal association data generated in step S2, module time-varying causal relationship modeling is performed. Methods such as Granger causality test and transfer entropy can be used to analyze whether there is a causal relationship between modules and the strength of the causal relationship. For example, by analyzing the temporal relationship between the output control signal of the flight control computer and the measured value of the attitude sensor, it can be determined whether the output of the flight control computer has a causal effect on the measured value of the attitude sensor. The modeling result is represented in the form of inter-module causal relationship network data, for example, a directed graph is used to represent the causal relationship between modules, where nodes represent modules, edges represent causal relationships, and the direction of the edges represents the direction of causal influence. Then, the inter-module causal relationship network data is used to perform fault response verification processing. For example, if the flight control computer is suspected to be a faulty module, the impact of its fault on other modules can be predicted based on the causal relationship network data, and the predicted results can be compared with the actual monitoring data. The comparison results can be used to verify the fault hypothesis and generate fault response feature data, for example, to record the consistency between the predicted results and the actual monitoring data. Finally, the fault confidence is calculated based on the fault response feature data. For example, if the predicted results are highly consistent with the actual monitoring data, the confidence of the faulty module can be improved. The calculation results are expressed in the form of fault location confidence data, for example, the probability of each module being a faulty module is expressed in numerical values.

[0096] Step S4: locate the fault module of the avionics system through the fault location confidence data to obtain the target fault module; perform fault module fuzzification processing on the target fault module to generate a fault self-diagnosis fuzzy decision model; use the fault self-diagnosis fuzzy decision model to perform intelligent fault diagnosis processing to generate system fault self-diagnosis report data.

[0097] In an embodiment of the present invention, the fault module is located according to the fault location confidence data. A confidence threshold is set, such as 80%. If the fault confidence of a module is higher than the threshold, the module is located as the target fault module. In this example, the flight control computer has the highest confidence and is higher than the threshold, so it is located as the target fault module. Then, the target fault module is fuzzified. Since the characteristics of the fault are often imprecise, the use of fuzzy logic can better describe the fault. For example, the fault type of the flight control computer is fuzzified into fuzzy sets such as "minor sensor fault", "medium sensor fault", and "serious sensor fault", and the membership function is used to describe the degree to which the fault belongs to each fuzzy set. These fuzzified fault information will be used to construct a fault self-diagnosis fuzzy decision model, for example, to construct a fuzzy decision tree for fault diagnosis based on the fuzzified fault characteristics. Finally, the constructed fault self-diagnosis fuzzy decision model is used for intelligent fault diagnosis processing. The characteristic data of the target fault module is input into the fuzzy decision model, and the model will output the fault diagnosis result according to the fuzzy reasoning rule, for example, the diagnosis result is "medium sensor fault" with a confidence of 85%. Integrate the diagnostic results, confidence levels, and other relevant information (such as fault occurrence time, fault characteristic data, etc.) into the system fault self-diagnosis report data to facilitate maintenance personnel to conduct troubleshooting and repairs.

[0098] Preferably, step S1 comprises the following steps:

[0099] Step S11: performing aviation operation status identification on the avionics system to generate aviation operation status data; performing electronic monitoring module characteristic analysis on the avionics system to generate monitoring module characteristic parameters;

[0100] Step S12: dynamically configure the sampling rate according to the aviation operation status data and the characteristic parameters of the monitoring module to obtain the monitoring module sampling strategy;

[0101] Step S13: performing initial sampling of the monitoring module status according to the monitoring module sampling strategy to generate initial aviation monitoring module status data;

[0102] Step S14: performing abnormal state detection on the initial aviation monitoring module state data through a preset module monitoring state threshold to obtain abnormal module detection mark data;

[0103] Step S15: Mark the suspicious monitoring module according to the abnormal module detection mark data to obtain the electronic suspected fault module; perform data interaction module discrimination on the avionics system according to the electronic suspected fault module to obtain the electronic adjacent monitoring module and other electronic monitoring modules respectively;

[0104] Step S16: Based on the preset monitoring frequency adjustment data, the monitoring module sampling strategy is used to perform high-frequency sampling adjustment on the electronic suspected fault module to generate key module monitoring data; the electronic adjacent monitoring module is performed medium-frequency sampling adjustment to generate associated module monitoring data; the other electronic monitoring modules are performed low-frequency sampling adjustment to generate basic module monitoring data;

[0105] Step S17: integrating the key module monitoring data, the associated module monitoring data and the basic module monitoring data, and performing timestamp processing to generate original fault module monitoring data.

[0106] In the embodiment of the present invention, the avionics system of a certain type of aircraft is taken as an example, and the system includes a flight control system, a navigation system, a communication system, etc. First, by accessing the aircraft data bus, the real-time flight data of the aircraft, such as flight altitude, speed, attitude, engine status, etc., is obtained, and the current aviation operation status, such as take-off, cruising, landing, etc., is judged in combination with the preset flight status logic, and the aviation operation status data is generated. Then, for the electronic monitoring module of the aircraft, such as the flight control computer, the inertial navigation unit, the radio altimeter, etc., its working principle, performance parameters, data transmission rate and other characteristics are analyzed to generate monitoring module characteristic parameters, such as module importance, data update frequency, fault sensitivity, etc. For example, the module importance of the flight control computer is defined as the highest level, the data update frequency is 100 times per second, and the fault sensitivity is high. For example, in the cruising state of the aircraft, the flight control computer is set to high-frequency sampling due to its importance and high data update frequency, and the sampling rate is 50 times per second; the inertial navigation unit is set to medium-frequency sampling, and the sampling rate is 10 times per second; the radio altimeter is set to low-frequency sampling due to its low frequency of use in the cruising state, and the sampling rate is 1 time per second. During the take-off and landing of the aircraft, due to the higher requirements for flight control and attitude stability, the sampling rates of all relevant modules are increased, such as the sampling rate of the flight control computer is increased to 100 times per second, and the sampling rate of the inertial navigation unit is increased to 20 times per second. For example, after the aircraft enters the cruising state, according to the preset sampling strategy, the flight control computer is sampled 50 times per second to obtain the control signal, state parameters and other data output by it; the inertial navigation unit is sampled 10 times per second to obtain the attitude, position, speed and other data output by it; the radio altimeter is sampled once per second to obtain the aircraft height data measured by it. The collected data is preliminarily processed and formatted to generate the initial aviation monitoring module status data. The normal working range threshold of each monitoring module is pre-set. For example, the output control signal amplitude range of the flight control computer is -5V to 5V, and the attitude angular rate range is -10° / s to 10° / s. The initial aviation monitoring module status data obtained in step S13 is compared with the preset threshold. For example, if the amplitude of the output control signal of the flight control computer exceeds 5V or is less than -5V, the module is determined to be abnormal, and the abnormal module detection flag data is set to "1" to indicate that an abnormality exists; if the data is within the normal range, the flag data is set to "0" to indicate that it is normal. If the abnormal module detection flag data of the flight control computer is detected as "1" in step S14, the flight control computer is marked as an electronic suspected fault module.According to the topological structure and data interaction relationship of the aircraft's avionics system, the modules that have direct data interaction with the flight control computer, such as attitude sensors and control units, are identified and marked as electronic adjacent monitoring modules; other modules that do not have direct data interaction with the flight control computer, such as navigation systems and communication systems, are marked as other electronic monitoring modules. The sampling frequency of the flight control computer marked as a suspected electronic fault module is adjusted to high frequency, for example, from 50 times per second to 100 times per second, to generate key module monitoring data. The sampling frequency of the attitude sensor and control unit adjacent to it is adjusted to medium frequency, for example, from 10 times per second to 20 times per second, to generate associated module monitoring data. For other electronic monitoring modules, such as navigation systems and communication systems, the original low-frequency sampling strategy is maintained, for example, once per second, to generate basic module monitoring data. The integrated data is sorted by timestamp to generate the original fault module monitoring data.

[0107] Preferably, step S12 comprises the following steps:

[0108] Step S121: dividing the aviation operation status data into flight mission stages to obtain mission stage characteristic data;

[0109] Step S122: dynamically evaluate the system criticality according to the monitoring module characteristic parameters and the task phase characteristic data to obtain module priority data;

[0110] Step S123: Allocate monitoring sampling resources according to the module priority data to generate monitoring resource configuration data;

[0111] Step S124: Optimize the module sampling rate of the monitoring resource configuration data in real time, and dynamically generate the sampling strategy to obtain the monitoring module sampling strategy.

[0112] In the embodiment of the present invention, a certain type of aircraft performs a flight mission as an example, and the mission includes take-off, climb, cruise, descent, landing and other stages. By analyzing the real-time flight data of the aircraft, such as altitude, speed, attitude, engine status, etc., the aviation operation status data is divided into different flight mission stages. For example, when the aircraft altitude is lower than 1500 meters, the speed is lower than 250 knots, the landing gear is lowered, and the flaps are deployed, it is determined to be the landing stage; when the aircraft altitude is higher than 10000 meters, the speed is higher than 400 knots, and the engine is in the cruise thrust state, it is determined to be the cruise stage. Each stage corresponds to different mission stage characteristic data, for example, the characteristic data of the landing stage includes: low altitude, low speed, high flap deflection, high landing gear state, etc.; the characteristic data of the cruise stage includes: high altitude, high speed, low engine thrust change rate, etc. Combined with the mission stage characteristic data obtained in step S121 and the pre-set monitoring module characteristic parameters, the criticality dynamic evaluation of the avionics system is performed. For example, in the landing phase, since the landing gear and flap systems are crucial to flight safety, the criticality of the landing gear control module and the flap control module is evaluated as the highest level; at the same time, since low-altitude flight requires higher stability of the flight control system, the criticality of the flight control computer is also higher. In the cruising phase, the stability of the engine and navigation system is more important, so the criticality of the engine control module and the inertial navigation unit is evaluated as the highest level. According to the criticality evaluation results, module priority data is generated, for example, the module with the highest criticality is given the highest priority, and the module with the lowest criticality is given the lowest priority. The system has limited monitoring sampling resources, such as data bus bandwidth, processor computing power, etc. According to the module priority data obtained in step S122, these limited monitoring sampling resources are allocated. Modules with high priority are allocated more sampling resources, such as higher sampling frequency and larger data transmission bandwidth; modules with low priority are allocated fewer sampling resources. For example, in the landing phase, the landing gear control module and the flap control module will be allocated higher sampling frequency and larger data transmission bandwidth due to their highest priority to ensure that their status can be monitored in real time; while some secondary systems, such as the cabin entertainment system, will be allocated fewer resources. Finally, monitoring resource configuration data is generated, which contains the specific sampling resources allocated to each module, such as sampling frequency, data transmission bandwidth, etc. For example, even in the same flight phase, there are different working conditions, and the sampling rate needs to be dynamically adjusted. During the final approach of the landing phase, the sampling rate of the flight control computer can be further increased due to the need for more precise control of the aircraft attitude. At the same time, in order to avoid resource conflicts, the sampling rates of other modules can be adjusted according to the real-time situation, such as reducing the sampling rates of some non-critical modules to ensure that the sampling requirements of critical modules are met.Finally, a monitoring module sampling strategy is dynamically generated, which includes the specific sampling rate of each module under different working conditions and can be dynamically adjusted according to the real-time situation.

[0113] As an example of the present invention, refer to Figure 2 As shown, Figure 1 Detailed implementation steps of step S2 in the embodiment are shown in the flowchart. In this embodiment, step S2 includes:

[0114] Step S21: preprocessing the original fault module monitoring data to obtain preprocessed fault module monitoring data;

[0115] In an embodiment of the present invention, for example, the original fault module monitoring data contains noise, missing values, etc., and needs to be preprocessed. For example, Kalman filtering can be used to remove noise, and linear interpolation or spline interpolation can be used to fill missing values. In addition, data normalization processing can also be performed to convert data of different dimensions into the same range for subsequent analysis. For example, the output control signal of the flight control computer and the pitch angle data of the attitude sensor are normalized to between 0 and 1.

[0116] Step S22: performing time series slicing according to the pre-processed fault module monitoring data to generate time series fault module monitoring slice data;

[0117] In an embodiment of the present invention, the size of the slicing window needs to be selected according to the specific situation, for example, it can be determined according to the sampling frequency of the module and the expected fault duration. Assuming that the sampling frequency of the flight control computer is 100 Hz and the expected fault duration is about 1 second, 1 second can be selected as the size of the slicing window. The pre-processed flight control computer output control signal data is sliced ​​according to a 1-second time window, and a series of data segments with a length of 100 can be obtained, each segment representing the change of the output control signal of the flight control computer within 1 second. Similarly, the pre-processed monitoring data of other modules can be sliced ​​in time series to generate corresponding timing fault module monitoring slice data. Time series slicing can divide long-term monitoring data into smaller segments, which is convenient for subsequent separate analysis of each segment, such as calculating parameter change rate, performing spectrum analysis, etc.

[0118] Step S23: Calculate the monitoring parameter change rate of the sequential fault module monitoring slice data, determine the adaptive time window, and generate adaptive time window data;

[0119] In an embodiment of the present invention, slice data is monitored for each timing fault module, and the change rate of each monitoring parameter is calculated. For example, for the output control signal of the flight control computer, the change rate of the control signal in each slice can be calculated, that is, the difference between the control signal of two adjacent sampling points. The change rate can reflect the speed of the module state change. If the parameter change rate in a slice is large, it means that the module state changes quickly and there is an abnormal situation; conversely, if the parameter change rate is small, it means that the module state is relatively stable. According to the size of the parameter change rate, the size of the time window can be adaptively adjusted. For example, if the change rate of the output control signal of the flight control computer in a slice exceeds a preset threshold, the time window can be reduced, for example, from 1 second to 0.5 seconds, so as to more finely observe the module state changes in this time period. Conversely, if the parameter change rate is small, the size of the time window can be maintained or increased.

[0120] Step S24: performing inter-module data association screening on the pre-processed fault module monitoring data to generate associated module data pairs;

[0121] In an embodiment of the present invention, in an avionics system, there is usually data interaction and functional association between different modules. For example, the output control signal of the flight control computer will affect the deflection angle of the rudder control unit, and the pitch angle of the attitude sensor will serve as the input signal of the flight control computer. In order to analyze the mutual influence between modules, it is necessary to perform inter-module data association screening on the pre-processing fault module monitoring data to generate associated module data pairs. The method for screening associated module data pairs can determine the connection relationship and data flow between different modules based on the system's design documents, circuit diagrams and other materials. For example, based on the flight control system design document of the aircraft, it can be determined that there is a data association between the flight control computer and the rudder control unit, and the output control signal data of the flight control computer and the rudder deflection angle data of the rudder control unit are paired to form an associated module data pair.

[0122] Step S25: performing inter-module signal delay calculation on the associated module data pair through the adaptive time window data to generate inter-module delay characteristic data;

[0123] In an embodiment of the present invention, in an associated module data pair, a signal change of one module causes a signal change of another module, but there is usually a certain time delay between the two. For example, after the flight control computer sends a control signal, the rudder control unit needs a certain time to respond and adjust the rudder deflection angle. This time delay is called the inter-module signal delay, which can reflect the response speed and signal transmission efficiency between modules. In order to calculate the inter-module signal delay, the associated module data pair can be analyzed using adaptive time window data. For example, for the associated module data pair of the flight control computer and the rudder control unit, the time difference between the two can be calculated based on the change time of the control signal output by the flight control computer and the change time of the rudder deflection angle of the rudder control unit, that is, the inter-module signal delay. Signal processing methods such as cross-correlation algorithm and wavelet transform can be used to calculate the time delay. The calculated inter-module signal delay can be used as inter-module delay feature data.

[0124] Step S26: Perform signal propagation analysis based on the inter-module delay characteristic data, and monitor the correlation of timing data between modules to generate inter-module timing correlation data.

[0125] In the embodiment of the present invention, for example, it is assumed that the flight control computer sends a control signal, and after a certain delay, the control unit receives the signal and adjusts the deflection angle of the control unit, and after a certain delay, the attitude sensor detects the change in the attitude of the aircraft. By analyzing the delay characteristic data between these modules, the propagation link of the signal between different modules can be constructed, and the role of each module in the signal propagation process can be determined. For example, the propagation path of the signal from the flight control computer to the control unit and then to the attitude sensor can be determined based on the delay data between the flight control computer, the control unit and the attitude sensor. In addition, the timing data between the monitoring modules can also be associated based on the delay characteristic data between the modules. For example, the control signal sent by the flight control computer at time t is associated with the deflection angle data of the control unit at time t+Δt, where Δt is the signal delay between the flight control computer and the control unit. Similarly, the deflection angle data of the control unit can be associated with the attitude angle data of the attitude sensor at time t+Δt+Δt', where Δt' is the signal delay between the control unit and the attitude sensor.

[0126] Preferably, step S3 comprises the following steps:

[0127] Step S31: segmenting the inter-module timing correlation data to obtain normal module status data and fault module status data respectively;

[0128] Step S32: sorting the electronic suspected faulty modules according to the faulty module status data to obtain faulty module sorting data;

[0129] Step S33: Modeling the module time-varying causal relationship of the faulty module sorting data using the original faulty module monitoring data and the normal module status data to generate inter-module causal relationship network data;

[0130] Step S34: using the inter-module causal relationship network data to perform fault response verification processing on the fault module sorting data to generate fault response feature data;

[0131] Step S35: Calculate the fault confidence of the fault response characteristic data according to the fault module status data and the causal relationship network data between modules to obtain the fault location confidence data.

[0132] In the embodiment of the present invention, for example, according to a preset threshold or rule, it is determined whether each module is in a normal state at each time point. For example, if the output control signal of the flight control computer exceeds the preset safety range, it is considered that the flight control computer is in a faulty state at this time point. By summarizing the state judgment results of all modules at all time points, the inter-module temporal correlation data can be divided into two parts: one part is the data that all modules are in a normal state, which is called normal module state data; the other part is the data that at least one module is in a faulty state, which is called faulty module state data. For example, if the output control signal of the flight control computer is abnormal in a certain time period, while the data of the attitude sensor and the control unit of the control surface are normal, the data of the time period is divided into faulty module state data. The faulty module state data obtained in step S31 is used to sort these suspected faulty modules. The sorting can be based on the frequency of module failure, fault duration, fault severity, etc. For example, if the flight control computer appears most frequently in the faulty module state data and has the longest fault duration, the flight control computer is considered to be the faulty module and is ranked first. Other suspected faulty modules are sorted according to similar rules, and finally the faulty module sorting data is obtained. For example, the sorting result is: flight control computer, attitude sensor, and control unit. The faulty module sorting data obtained in step S32 is modeled for module time-varying causal relationships. Methods such as Granger causality test and transfer entropy can be used to analyze the data correlation and causal relationship between modules. For example, the relationship between the output control signal of the flight control computer and the pitch angle data of the attitude sensor is analyzed. If it is found that the change of the output control signal of the flight control computer leads the change of the pitch angle of the attitude sensor, it can be inferred that the flight control computer has a causal effect on the attitude sensor. The causal relationship between all modules is represented by a directed graph, in which the node represents the module, the edge represents the causal relationship, and the direction of the edge represents the direction of the causal influence, and finally the inter-module causal relationship network data is generated. For example, if the flight control computer ranks first in the faulty module sorting data, then according to the inter-module causal relationship network data, it is checked whether the failure of the flight control computer will cause abnormal responses in other modules. For example, if the flight control computer fails, according to the causal relationship network, the data of the attitude sensor and the control unit should also change accordingly. By analyzing the original faulty module monitoring data, it is verified whether these expected responses occur. If the actual observed data is consistent with the expected response, the confidence in the flight control computer failure can be enhanced and fault response characteristic data can be generated. For example, if the flight control computer appears frequently in the fault module status data, and during the fault response verification process, the responses of other modules are consistent with expectations, then the flight control computer failure confidence can be considered high.Bayesian networks, fuzzy logic and other methods can be used to calculate fault confidence. Finally, the fault location confidence data is obtained. For example, the fault confidence of the flight control computer is 90%, the fault confidence of the attitude sensor is 50%, and the fault confidence of the control unit is 10%.

[0133] Preferably, step S33 includes the following steps:

[0134] Step S331: extracting module benchmark response characteristics from the original fault module monitoring data to generate module benchmark response characteristic data;

[0135] Step S332: using the fault module sorting data to perform Granger causality test on the temporal correlation data between modules to generate fault module causal relationship data;

[0136] Step S333: constructing a causal network model according to the fault module causal relationship data to generate fault module network structure data;

[0137] Step S334: performing time-varying causal strength calculation on the faulty module network structure data based on the normal module state data, the faulty module state data and the module reference response characteristic data to generate time-varying causal strength data;

[0138] Step S335: performing fault propagation path analysis according to the time-varying causal strength data, and performing causal relationship direction processing to generate fault propagation path data;

[0139] Step S336: quantify the fault propagation path abnormality weight of the fault propagation path data, and reconstruct the causal link using the fault module network structure data to generate inter-module causal relationship network data.

[0140] In the embodiment of the present invention, for example, the original fault module monitoring data includes data such as the output control signal of the flight control computer, the pitch angle of the attitude sensor, and the deflection angle of the rudder control unit. Under normal working conditions, there is a specific response relationship between different modules. For example, after the flight control computer sends a control signal, the deflection angle of the rudder control unit will change accordingly. In order to extract the module benchmark response characteristics, the original fault module monitoring data under normal flight conditions can be analyzed. For example, a system identification method can be used to establish a transfer function model between the output control signal of the flight control computer and the deflection angle of the rudder control unit. The model can describe the dynamic response relationship between the two and serve as the module benchmark response characteristic data. The Granger causality test can determine whether the change of a time series can be used to predict the change of another time series. For example, it is tested whether the output control signal of the flight control computer has Granger causality to the pitch angle of the attitude sensor. If the test result shows that there is a causal relationship, the causal relationship is added to the fault module causal relationship data, for example, it is recorded that the output control signal of the flight control computer has a causal effect on the pitch angle of the attitude sensor. The causal network model can be represented by a directed graph, in which nodes represent modules and directed edges represent causal relationships between modules. For example, if the output control signal of the flight control computer has a causal effect on the pitch angle of the attitude sensor, a directed edge is drawn from the flight control computer node to the attitude sensor node in the causal network model, and the constructed causal network model is the fault module network structure data. By comparing the difference between the effect of the output control signal of the flight control computer on the pitch angle of the attitude sensor in the fault state and the benchmark response characteristics in the normal state, the change in the strength of the causal relationship in the fault state can be quantified. The time-varying causal strength can be calculated using methods such as transfer entropy and conditional mutual information. The calculation results are stored as time-varying causal strength data, for example, it is recorded that the causal strength of the flight control computer to the attitude sensor increased by 50% in a certain time period. For example, if it is found that the causal strength of the flight control computer to the attitude sensor increases significantly, and the causal strength of the attitude sensor to the control unit also increases, it can be inferred that the fault propagates from the flight control computer to the attitude sensor and then to the control unit. By analyzing the change of time-varying causal strength, the direction and path of fault propagation can be determined, and the result can be stored as fault propagation path data. For example, the fault propagation path is recorded as: flight control computer->attitude sensor->rudder control unit. For example, according to the change amplitude of time-varying causal strength, the abnormality degree of each propagation path is quantified. For example, the abnormality degree of the propagation path from the flight control computer to the attitude sensor is quantified as 0.8, and the abnormality degree of the propagation path from the attitude sensor to the rudder control unit is quantified as 0.5. Then, combined with the fault module network structure data constructed in step S333, the causal link is reconstructed.For example, if a propagation path is found to be highly abnormal, the weight of the path in the causal network model can be increased, or a new causal connection can be added. The reconstructed causal network model contains more accurate causal relationships and fault propagation path information, and ultimately generates inter-module causal network data.

[0141] Preferably, step S34 includes the following steps:

[0142] Step S341: performing module safety threshold evaluation on the fault module sorting data, and setting control constraint conditions to generate safety control constraint condition data;

[0143] Step S342: setting a reverse excitation signal according to the safety control constraint condition data to obtain reverse excitation strategy data;

[0144] Step S343: using the inter-module causal relationship network data to identify the faulty downstream modules of the faulty module sorting data, and obtaining the faulty downstream module identification data;

[0145] Step S344: applying a control signal to the fault downstream module identification data using the reverse excitation strategy data, and monitoring the module response data to generate downstream module response data;

[0146] Step S345: Analyze the upstream module status of the electronic suspected fault module by using the causal relationship network data between modules to obtain the upstream fault module traceability data;

[0147] Step S346: Perform fault response verification processing on the upstream fault module traceability data and the downstream module response data to generate fault response feature data.

[0148] In an embodiment of the present invention, the safety thresholds of the flight control computer related parameters are evaluated, for example, the maximum safe amplitude of the output control signal, the maximum safe rate of change of the attitude angle, etc. Then, control constraints are set according to these safety thresholds, for example, the amplitude and frequency of the reverse excitation signal are limited to ensure that the flight safety will not be endangered when the fault response verification is performed. These control constraints can be adjusted according to the specific flight state and mission requirements. The control constraint data finally generated includes the safety threshold and the corresponding control constraint of each fault module, for example, the amplitude of the control signal output by the flight control computer is limited to ±5V, and the frequency is limited to 10Hz. The type, amplitude, frequency and other parameters of the reverse excitation signal need to be selected according to the specific fault module and control constraint. For example, for the flight control computer, a sine wave signal with a small amplitude and a low frequency can be injected as a reverse excitation signal to test its response to the excitation signal. The reverse excitation strategy data includes the excitation signal parameters of each fault module, for example, a sine wave signal with an amplitude of 0.5V and a frequency of 1Hz is injected into the flight control computer. For example, if the inter-module causal relationship network data shows that the flight control computer has a causal effect on the attitude sensor and the control unit, the attitude sensor and the control unit are identified as the faulty downstream modules of the flight control computer. The faulty downstream module identification data contains a list of downstream modules corresponding to each faulty module. For example, the faulty downstream modules of the flight control computer include the attitude sensor and the control unit. For example, after injecting a reverse excitation signal into the flight control computer, the response data of the attitude sensor and the control unit are monitored, such as the change in attitude angle, the change in the deflection angle of the control unit, etc. The monitored data is recorded to generate downstream module response data. For example, if the flight control computer is marked as a suspected faulty module, and the inter-module causal relationship network data shows that the attitude sensor has a causal effect on the flight control computer, it is necessary to analyze the state data of the attitude sensor to determine whether the attitude sensor is the root cause of the flight control computer failure. The upstream fault module traceability data contains the upstream module state analysis results for each suspected faulty module. For example, the state data of the attitude sensor is recorded to be normal, which excludes the possibility that it is the root cause of the flight control computer failure. After injecting a reverse excitation signal into the flight control computer, the response data of the attitude sensor and the control surface control unit are consistent with expectations, and the upstream module status analysis results show that no other modules cause the flight control computer failure. It can be considered that there is a high possibility of a failure in the flight control computer, and fault response characteristic data is generated.

[0149] Preferably, step S35 includes the following steps:

[0150] Step S351: performing time domain decomposition on the fault module state data to obtain time domain feature data of the fault module;

[0151] Step S352: performing frequency domain feature analysis on the fault module state data based on the fault module time domain feature data to generate the fault module frequency domain feature data;

[0152] Step S353: extracting causal domain features of the fault module according to the causal relationship network data between modules, and generating causal domain feature data of the fault module;

[0153] Step S354: performing fault feature consistency assessment based on the fault module time domain feature data, the fault module frequency domain feature data, and the fault module causal domain feature data, and performing aviation state weight optimization to generate fault feature weight data;

[0154] Step S355: Calculate the fault location reliability of the fault response feature data using the fault feature weight data to obtain the fault location reliability data.

[0155] In the embodiment of the present invention, the fault module state data includes the output control signal data collected during the fault period. The data is decomposed in the time domain to extract a variety of time domain features, such as: 1. Fault occurrence time: record the time point when the fault first occurs. 2. Fault duration: calculate the total duration of the fault. 3. Signal mean: calculate the average value of the output control signal during the fault period. 4. Signal variance: calculate the variance of the output control signal during the fault period, reflecting the degree of signal fluctuation. 5. Signal peak: extract the maximum and minimum values ​​of the output control signal during the fault period. 6. Signal change rate: calculate the rate of change of the signal between adjacent time points, reflecting the speed of signal change. Extracting these time domain features can help better understand the dynamic characteristics of the fault, for example, whether the fault occurs suddenly or develops gradually, how serious the fault is, etc. These feature data will be used as the time domain feature data of the fault module. Use Fast Fourier Transform (FFT) to convert the time domain signal to the frequency domain, and analyze the spectrum characteristics of the output control signal during the fault period. The following frequency domain features can be extracted: 1. Spectral peak: identify the main frequency components in the spectrum and record their corresponding amplitudes. 2. Spectral energy distribution: Analyze the energy distribution in different frequency bands, for example, whether the fault causes the energy in a specific frequency band to increase or decrease. 3. Harmonic components: Analyze whether there are harmonic components, as well as the amplitude and frequency of the harmonic components. These frequency domain features can help identify signal frequency changes caused by the fault, for example, whether the fault causes new frequency components to appear, or the amplitude of existing frequency components to change. For example, if the flight control computer is the faulty module, and the inter-module causal network data shows that the flight control computer has a causal impact on the attitude sensor and the control unit, the following causal domain features can be extracted: 1. Number of affected modules: Count the number of downstream modules directly affected by the faulty module. 2. Causal strength: Quantify the strength of the causal impact of the faulty module on the downstream modules, for example, use methods such as transfer entropy to calculate the causal strength. 3. Causal path length: Calculate the causal path length from the faulty module to the affected module, reflecting the range of fault propagation. These features are evaluated for consistency and weighted according to the current state of the aircraft. For example, if the time domain characteristics show that the output control signal of the flight control computer has a large amplitude fluctuation, the frequency domain characteristics show that the energy in a specific frequency band has increased significantly, and the causal domain characteristics show that the causal influence of the flight control computer on the downstream modules has increased, then these characteristics are consistent, which can enhance the confidence in the flight control computer failure. The consistency assessment can be based on expert knowledge or machine learning algorithms. In addition, the weights of different features need to be optimized according to the current state of the aircraft (e.g., takeoff, cruise, landing, etc.). For example, during the takeoff and landing phases, higher weights are given to features that are closely related to flight safety. The fault location confidence calculation is performed on the fault response feature data generated in step S346.For example, the time domain features, frequency domain features, and causal domain features of the flight control computer are multiplied by the corresponding weights, and then the weighted feature values ​​are integrated to calculate the fault location confidence of the flight control computer. Bayesian networks, fuzzy logic, and other methods can be used for confidence calculation. The final fault location confidence data contains the confidence value of each fault module. For example, the fault confidence of the flight control computer is 95%, and the fault confidence of the attitude sensor is 60%. The higher the confidence, the greater the possibility of failure of the module.

[0156] As an example of the present invention, refer to Figure 3 As shown, Figure 1 Detailed implementation steps of step S4 in the flowchart, in this example, step S4 includes:

[0157] Step S41: locating the fault module of the avionics system by using the fault location reliability data to obtain the target fault module;

[0158] In an embodiment of the present invention, for example, in the fault location confidence data generated in step S355, the fault confidence of the flight control computer is 95%, the fault confidence of the attitude sensor is 60%, and the fault confidence of the control surface control unit is 10%. Set a confidence threshold, such as 80%. The module with a confidence higher than the threshold is located as the target fault module. In this example, the confidence of the flight control computer is higher than the threshold, so it is located as the target fault module. If the confidence of multiple modules is higher than the threshold, they can be sorted according to the confidence level, and the module with the highest confidence is processed first. The result of fault module location will be used for subsequent fault type retrieval and diagnosis. If the confidence of all modules is lower than the threshold, it can be considered that there is no obvious fault module at present, or the confidence threshold needs to be adjusted for re-evaluation.

[0159] Step S42: searching the fault type of the target fault module through a preset fault type library to generate a module fault type data set;

[0160] In an embodiment of the present invention, after determining that the flight control computer is the target fault module, it is necessary to retrieve the fault type according to the preset fault type library. The fault type library stores various fault types and their corresponding feature descriptions. For example, for the flight control computer, the fault types include: sensor fault, actuator fault, software fault, hardware fault, etc. Each fault type has its specific characteristic manifestations. For example, sensor faults are manifested as data anomalies, increased noise, etc.; actuator faults are manifested as output signal anomalies, response delays, etc.; software faults are manifested as program crashes, logical errors, etc.; hardware faults are manifested as circuit short circuits, component damage, etc. By matching the features of the target fault module with the features in the fault type library, the fault type of the target fault module can be preliminarily determined. For example, if the output control signal of the flight control computer fluctuates abnormally, it can be preliminarily determined that there is a sensor fault or an actuator fault. The retrieval results will be stored in the module fault type data set, including all fault types and their corresponding feature descriptions.

[0161] Step S43: performing fault module fuzzification processing on the module fault type data set to generate fault module fuzzy feature data;

[0162] In an embodiment of the present invention, after obtaining the module fault type data set of the flight control computer, it is necessary to perform fuzzy processing on the fault types therein. Since the characteristics of actual faults are often fuzzy and uncertain, the use of fuzzy logic can better describe the fault characteristics. For example, for the fault type of "sensor fault", it can be fuzzified into fuzzy sets such as "minor sensor fault", "medium sensor fault", and "serious sensor fault". Each fuzzy set corresponds to a membership function, which is used to describe the degree to which the fault feature belongs to the set. For example, a triangular membership function or a Gaussian membership function can be used to define a fuzzy set. Through fuzzification processing, the original precise fault type description can be converted into a fuzzy feature description to better reflect the complexity and uncertainty of the actual fault. The generated fault module fuzzy feature data contains all fuzzified fault types and their corresponding membership functions.

[0163] Step S44: constructing a fuzzy decision tree according to the fuzzy feature data of the fault module to generate a fault self-diagnosis fuzzy decision model;

[0164] In an embodiment of the present invention, a fuzzy decision tree is constructed based on the fuzzy feature data of the fault module of the flight control computer. The construction process of the fuzzy decision tree is similar to that of the traditional decision tree, but the division of nodes and the judgment of branches are based on fuzzy logic. For example, the data can be divided into three branches of "minor fault", "moderate fault" and "serious fault" according to the membership function of "sensor fault". Each branch corresponds to a subtree, which is used to further judge the specific type of fault. In the process of constructing the fuzzy decision tree, it is necessary to select a suitable fuzzy reasoning method, such as Mamdani reasoning or TS fuzzy reasoning. The constructed fuzzy decision tree is a fault self-diagnosis fuzzy decision model, which is used to perform intelligent fault diagnosis on the target fault module.

[0165] Step S45: Obtain historical fault module diagnosis data; perform historical diagnosis accuracy statistics based on the historical fault module diagnosis data to obtain historical diagnosis accuracy data;

[0166] In the embodiment of the present invention, historical fault module diagnostic data of the flight control computer is obtained, and these data contain the results of the previous fault diagnosis of the flight control computer, including information such as the diagnosed fault type, the actual fault type, and whether the diagnosis is accurate. Based on these historical data, the historical diagnosis accuracy rate is calculated. For example, in the past year, the accuracy rate of the fault diagnosis of the flight control computer is 90%.

[0167] Step S46: performing accuracy performance trend analysis on historical diagnostic accuracy data, and performing adaptive diagnostic confidence threshold adjustment to obtain adaptive diagnostic threshold data;

[0168] In an embodiment of the present invention, the historical diagnostic accuracy data of the flight control computer is analyzed, for example, the diagnostic accuracy of each month in the past year is analyzed, and its changing trend is observed. If it is found that the diagnostic accuracy shows a downward trend, it indicates that the performance of the fault diagnosis model has declined and needs to be adjusted. According to the accuracy performance trend analysis results, the diagnostic confidence threshold is adaptively adjusted. For example, if the historical diagnostic accuracy is high, the diagnostic confidence threshold can be appropriately increased to reduce false alarms; conversely, if the historical diagnostic accuracy is low, the diagnostic confidence threshold can be appropriately lowered to increase the recall rate. The adjusted diagnostic confidence threshold will be used as the adaptive diagnostic threshold data.

[0169] Step S47: Optimize the decision threshold of the fault self-diagnosis fuzzy decision model through adaptive diagnosis threshold data, perform intelligent fault diagnosis processing on the target fault module, and generate system fault self-diagnosis report data.

[0170] In an embodiment of the present invention, the decision threshold of the fault self-diagnosis fuzzy decision model constructed in step S44 is optimized. For example, the decision threshold of each node in the fuzzy decision tree is adjusted according to the adaptive diagnostic threshold data. The optimized fuzzy decision model can perform intelligent fault diagnosis processing according to the fuzzy feature data of the target fault module. For example, the fuzzy feature data of the flight control computer is input into the fuzzy decision model, and the model will output the final fault diagnosis result according to the fuzzy reasoning rules. For example, the diagnosis result is "moderate sensor fault" with a confidence level of 85%. The diagnostic results, confidence levels and other relevant information (for example, fault occurrence time, fault feature data, etc.) are integrated into the system fault self-diagnosis report data to facilitate maintenance personnel to troubleshoot and repair faults.

[0171] Preferably, the present invention further provides an avionics system fault self-diagnosis system, which executes the avionics system fault self-diagnosis method as described above, and the avionics system fault self-diagnosis system comprises:

[0172] The system monitoring and acquisition module is used to analyze the characteristics of the electronic monitoring module of the avionics system and generate the characteristic parameters of the monitoring module; dynamically configure the sampling rate according to the characteristic parameters of the monitoring module to obtain the sampling strategy of the monitoring module; dynamically sample the fault module using the sampling strategy of the monitoring module to generate the original fault module monitoring data;

[0173] The time series data association module is used to perform time series slicing on the original fault module monitoring data to generate time series fault module monitoring slice data; perform time series data association between monitoring modules based on the time series fault module monitoring slice data to generate time series association data between modules;

[0174] The fault causal reasoning module is used to model the module time-varying causal relationship based on the inter-module temporal correlation data and generate the inter-module causal relationship network data; use the inter-module causal relationship network data to perform fault response verification processing and generate fault response feature data; calculate the fault confidence of the fault response feature data and obtain the fault location confidence data;

[0175] The fault self-diagnosis decision module is used to locate the fault module of the avionics system through the fault location confidence data to obtain the target fault module; perform fault module fuzzification processing on the target fault module to generate a fault self-diagnosis fuzzy decision model; use the fault self-diagnosis fuzzy decision model to perform intelligent fault diagnosis processing to generate system fault self-diagnosis report data.

[0176] The present application is to accurately capture the fault module in the avionics system through dynamic sampling strategy and time series data association analysis, and accurately track the fault propagation path through time-varying causal relationship modeling, so as to realize rapid fault location. This method not only reduces the possibility of misdiagnosis and missed diagnosis, but also significantly improves the accuracy and efficiency of fault diagnosis. By introducing fuzzy processing of fault modules and fuzzy decision models, the system's ability to handle uncertainty and fuzziness problems is enhanced, making the fault diagnosis results more robust. In addition, through statistical analysis of historical diagnostic data and adaptive diagnostic confidence threshold adjustment, the system can continuously learn and optimize, improve the intelligence and adaptability of the diagnostic model, and thus achieve more accurate fault diagnosis. Through real-time monitoring and intelligent diagnosis, the dependence on manual intervention is reduced, maintenance costs are reduced, and troubleshooting time is shortened. This is of great significance for improving the operating efficiency of aircraft and reducing flight delays caused by faults. At the same time, through the precise positioning and type identification of fault modules, the system can provide maintenance personnel with more specific and detailed fault information, guide them to carry out targeted maintenance operations, and reduce the cost waste caused by blindly replacing parts.

[0177] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0178] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for self-diagnosis of avionics system faults, characterized in that: The following steps are involved: Step S1: Analyze the characteristics of the electronic monitoring module of the avionics system to generate characteristic parameters of the monitoring module; dynamically configure the sampling rate according to the characteristic parameters of the monitoring module to obtain a sampling strategy for the monitoring module; dynamically sample the fault module using the sampling strategy for the monitoring module to generate original fault module monitoring data; wherein, step S1 is specifically as follows: Step S11: performing aviation operation status identification on the avionics system to generate aviation operation status data; performing electronic monitoring module characteristic analysis on the avionics system to generate monitoring module characteristic parameters; Step S12: dynamically configure the sampling rate according to the aviation operation status data and the characteristic parameters of the monitoring module to obtain the monitoring module sampling strategy; Step S13: performing initial sampling of the monitoring module status according to the monitoring module sampling strategy to generate initial aviation monitoring module status data; Step S14: performing abnormal state detection on the initial aviation monitoring module state data through a preset module monitoring state threshold to obtain abnormal module detection mark data; Step S15: Mark the suspicious monitoring module according to the abnormal module detection mark data to obtain the electronic suspected fault module; perform data interaction module discrimination on the avionics system according to the electronic suspected fault module to obtain the electronic adjacent monitoring module and other electronic monitoring modules respectively; Step S16: Based on the preset monitoring frequency adjustment data, the monitoring module sampling strategy is used to perform high-frequency sampling adjustment on the electronic suspected fault module to generate key module monitoring data; the electronic adjacent monitoring module is performed medium-frequency sampling adjustment to generate associated module monitoring data; the other electronic monitoring modules are performed low-frequency sampling adjustment to generate basic module monitoring data; Step S17: integrating the key module monitoring data, the associated module monitoring data and the basic module monitoring data, and performing timestamp processing to generate original fault module monitoring data; Step S2: Slice the original fault module monitoring data in time series to generate time series fault module monitoring slice data; associate the time series data between monitoring modules according to the time series fault module monitoring slice data to generate time series association data between modules; Step S3: Modeling the module time-varying causal relationship according to the inter-module timing correlation data to generate inter-module causal relationship network data; using the inter-module causal relationship network data to perform fault response verification processing to generate fault response feature data; performing fault confidence calculation on the fault response feature data to obtain fault location confidence data; Step S4: locate the fault module of the avionics system through the fault location confidence data to obtain the target fault module; perform fault module fuzzification processing on the target fault module to generate a fault self-diagnosis fuzzy decision model; use the fault self-diagnosis fuzzy decision model to perform intelligent fault diagnosis processing to generate system fault self-diagnosis report data.

2. The avionics system fault self-diagnosis method according to claim 1, characterized in that: Step S12 includes the following steps: Step S121: dividing the aviation operation status data into flight mission stages to obtain mission stage characteristic data; Step S122: dynamically evaluate the system criticality according to the monitoring module characteristic parameters and the task phase characteristic data to obtain module priority data; Step S123: Allocate monitoring sampling resources according to the module priority data to generate monitoring resource configuration data; Step S124: Optimize the module sampling rate of the monitoring resource configuration data in real time, and dynamically generate the sampling strategy to obtain the monitoring module sampling strategy.

3. The avionics system fault self-diagnosis method according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: preprocessing the original fault module monitoring data to obtain preprocessed fault module monitoring data; Step S22: performing time series slicing according to the pre-processed fault module monitoring data to generate time series fault module monitoring slice data; Step S23: Calculate the monitoring parameter change rate of the sequential fault module monitoring slice data, determine the adaptive time window, and generate adaptive time window data; Step S24: performing inter-module data association screening on the pre-processed fault module monitoring data to generate associated module data pairs; Step S25: performing inter-module signal delay calculation on the associated module data pair through the adaptive time window data to generate inter-module delay characteristic data; Step S26: Perform signal propagation analysis based on the inter-module delay characteristic data, and monitor the correlation of timing data between modules to generate inter-module timing correlation data.

4. The avionics system fault self-diagnosis method according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: segmenting the inter-module timing correlation data to obtain normal module status data and fault module status data respectively; Step S32: sorting the electronic suspected faulty modules according to the faulty module status data to obtain faulty module sorting data; Step S33: Modeling the module time-varying causal relationship of the faulty module sorting data using the original faulty module monitoring data and the normal module status data to generate inter-module causal relationship network data; Step S34: using the inter-module causal relationship network data to perform fault response verification processing on the fault module sorting data to generate fault response feature data; Step S35: Calculate the fault confidence of the fault response characteristic data according to the fault module status data and the causal relationship network data between modules to obtain the fault location confidence data.

5. The avionics system fault self-diagnosis method according to claim 4, characterized in that: Step S33 includes the following steps: Step S331: extracting module benchmark response characteristics from the original fault module monitoring data to generate module benchmark response characteristic data; Step S332: using the fault module sorting data to perform Granger causality test on the temporal correlation data between modules to generate fault module causal relationship data; Step S333: constructing a causal network model according to the fault module causal relationship data to generate fault module network structure data; Step S334: performing time-varying causal strength calculation on the faulty module network structure data based on the normal module state data, the faulty module state data and the module reference response characteristic data to generate time-varying causal strength data; Step S335: performing fault propagation path analysis according to the time-varying causal strength data, and performing causal relationship direction processing to generate fault propagation path data; Step S336: quantify the fault propagation path abnormality weight of the fault propagation path data, and reconstruct the causal link using the fault module network structure data to generate inter-module causal relationship network data.

6. The avionics system fault self-diagnosis method according to claim 4, characterized in that: Step S34 includes the following steps: Step S341: performing module safety threshold evaluation on the fault module sorting data, and setting control constraint conditions to generate safety control constraint condition data; Step S342: setting a reverse excitation signal according to the safety control constraint condition data to obtain reverse excitation strategy data; Step S343: using the inter-module causal relationship network data to identify the faulty downstream modules of the faulty module sorting data, and obtaining the faulty downstream module identification data; Step S344: applying a control signal to the fault downstream module identification data using the reverse excitation strategy data, and monitoring the module response data to generate downstream module response data; Step S345: Analyze the upstream module status of the electronic suspected fault module by using the causal relationship network data between modules to obtain the upstream fault module traceability data; Step S346: Perform fault response verification processing on the upstream fault module traceability data and the downstream module response data to generate fault response feature data.

7. The avionics system fault self-diagnosis method according to claim 4, characterized in that: Step S35 includes the following steps: Step S351: performing time domain decomposition on the fault module state data to obtain time domain feature data of the fault module; Step S352: performing frequency domain feature analysis on the fault module state data based on the fault module time domain feature data to generate the fault module frequency domain feature data; Step S353: extracting causal domain features of the fault module according to the causal relationship network data between modules, and generating causal domain feature data of the fault module; Step S354: performing fault feature consistency assessment based on the fault module time domain feature data, the fault module frequency domain feature data, and the fault module causal domain feature data, and performing aviation state weight optimization to generate fault feature weight data; Step S355: Calculate the fault location reliability of the fault response feature data using the fault feature weight data to obtain the fault location reliability data.

8. The avionics system fault self-diagnosis method according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: locating the fault module of the avionics system by using the fault location reliability data to obtain the target fault module; Step S42: searching the fault type of the target fault module through a preset fault type library to generate a module fault type data set; Step S43: performing fault module fuzzification processing on the module fault type data set to generate fault module fuzzy feature data; Step S44: constructing a fuzzy decision tree according to the fuzzy feature data of the fault module to generate a fault self-diagnosis fuzzy decision model; Step S45: Obtain historical fault module diagnosis data; perform historical diagnosis accuracy statistics based on the historical fault module diagnosis data to obtain historical diagnosis accuracy data; Step S46: performing accuracy performance trend analysis on historical diagnostic accuracy data, and performing adaptive diagnostic confidence threshold adjustment to obtain adaptive diagnostic threshold data; Step S47: Optimize the decision threshold of the fault self-diagnosis fuzzy decision model through adaptive diagnosis threshold data, perform intelligent fault diagnosis processing on the target fault module, and generate system fault self-diagnosis report data.

9. An avionics system fault self-diagnosis system, characterized in that: Used to execute the avionics system fault self-diagnosis method as claimed in claim 1, the avionics system fault self-diagnosis system comprises: The system monitoring and acquisition module is used to analyze the characteristics of the electronic monitoring module of the avionics system and generate the characteristic parameters of the monitoring module; dynamically configure the sampling rate according to the characteristic parameters of the monitoring module to obtain the sampling strategy of the monitoring module; dynamically sample the fault module using the sampling strategy of the monitoring module to generate the original fault module monitoring data; The time series data association module is used to perform time series slicing on the original fault module monitoring data to generate time series fault module monitoring slice data; perform time series data association between monitoring modules based on the time series fault module monitoring slice data to generate time series association data between modules; The fault causal reasoning module is used to model the module time-varying causal relationship based on the inter-module temporal correlation data and generate the inter-module causal relationship network data; use the inter-module causal relationship network data to perform fault response verification processing and generate fault response feature data; calculate the fault confidence of the fault response feature data and obtain the fault location confidence data; The fault self-diagnosis decision module is used to locate the fault module of the avionics system through the fault location confidence data to obtain the target fault module; perform fault module fuzzification processing on the target fault module to generate a fault self-diagnosis fuzzy decision model; use the fault self-diagnosis fuzzy decision model to perform intelligent fault diagnosis processing to generate system fault self-diagnosis report data.

Images