Satellite fault diagnosis method and system based on multi-source information fusion

Abstract

The application provides a satellite fault diagnosis method and system based on multi-source information fusion. First, a multi-source monitoring data set of platform system state data, load system working data and test control link interaction data generated in the process of testing the satellite based on an emergency quick response satellite test system is obtained. Then, multi-source information correlation extraction processing is performed on the multi-source monitoring data set to generate a correlation feature set reflecting the coupling relationship between systems. Then, the correlation feature set is input into a pre-trained satellite fault diagnosis model for fault mode analysis processing to obtain fault feature descriptions containing fault trigger nodes and influence links. Then, based on the fault feature descriptions, a pre-set fault knowledge base is matched to generate a fault diagnosis result containing fault type identification and propagation path. Finally, a satellite fault handling strategy is generated according to the fault diagnosis result. The method can comprehensively and accurately diagnose satellite faults and improve satellite reliability.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of aerospace technology, in particular, to a satellite fault diagnosis method and system based on multi-source information fusion. BACKGROUND

[0002] The structure of modern satellite systems is increasingly complex and highly integrated, and they are operated in harsh space environments (such as extreme temperatures, high-energy particle radiation, and vacuum microgravity) for a long time in orbit. The reliability and safety of the satellite systems are directly related to the safety of valuable assets and the success of critical space missions. Through ground environment simulation tests during the development phase and comprehensive tests before launch, the core carriers of the working status and health of each subsystem, single machine, and even components of the satellite can be truly reflected. These data usually have characteristics of multi-dimensionality, high frequency, strong time sequence, and large data volume, and contain rich state information and potential fault signs.

[0003] Currently, satellite fault diagnosis mainly relies on a single type of data monitoring and analysis method. For example, only the state data of the platform system is monitored, or only the working data of the payload system is focused on. However, the satellite system is a highly complex coupled system, and there is a close interaction and dependency relationship between each subsystem. The fault diagnosis method of a single data source often cannot comprehensively and accurately reflect the actual running state of the satellite system, and it is difficult to capture potential faults caused by the coupling relationship between systems. In addition, the existing fault diagnosis methods also have deficiencies in fault mode analysis and fault knowledge base matching, which cannot quickly and accurately determine the fault type and propagation path, resulting in a lack of pertinence and effectiveness of the fault handling scheme, which seriously affects the reliability and service life of the satellite. SUMMARY

[0004] Therefore, the purpose of the present application is to provide a satellite fault diagnosis method and system based on multi-source information fusion.

[0005] According to a first aspect of the present application, a satellite fault diagnosis method based on multi-source information fusion is provided, which comprises:

[0006] a multi-source monitoring data set measured during the testing of a satellite by an emergency quick-response satellite test system, the multi-source monitoring data set containing platform system state data, payload system working data, and measurement and control link interaction data;

[0007] performing multi-source information correlation extraction processing on the multi-source monitoring data set to generate a correlation feature set reflecting the coupling relationship between systems;

[0008] inputting the correlation feature set into a pre-trained satellite fault diagnosis model to perform fault mode analysis processing, and obtaining fault feature descriptions containing fault trigger nodes and influence links;

[0009] Based on the fault feature description, a fault diagnosis result containing fault type identification and propagation path is generated by matching a preset fault knowledge base;

[0010] A satellite fault handling strategy is generated according to the fault diagnosis result.

[0011] According to a second aspect of the present application, a satellite fault diagnosis system based on multi-source information fusion is provided, which comprises a machine readable storage medium and a processor, the machine readable storage medium stores machine executable instructions, and the processor, when executing the machine executable instructions, implements the aforementioned satellite fault diagnosis method based on multi-source information fusion.

[0012] According to a third aspect of the present application, a computer readable storage medium is provided, which stores computer executable instructions, and when the computer executable instructions are executed, the aforementioned satellite fault diagnosis method based on multi-source information fusion is implemented.

[0013] According to any one of the above aspects, the technical effect of the present application is that:

[0014] By acquiring a multi-source monitoring data set including platform system state data, load system working data and TT&C link interaction data during satellite operation, and performing multi-source information correlation extraction processing, the coupling relationship between each subsystem of the satellite system can be fully and deeply mined, and a correlation feature set accurately reflecting the correlation characteristics between systems is generated. By inputting the correlation feature set into a pre-trained satellite fault diagnosis model for fault mode analysis processing, the fault trigger node and influence link can be accurately identified, and a detailed fault feature description is obtained. Based on the fault feature description, a fault diagnosis result containing fault type identification and propagation path can be quickly generated by matching a preset fault knowledge base, which provides an accurate basis for fault handling. Finally, a satellite fault handling scheme containing positioning coordinates and handling strategies is generated according to the fault diagnosis result, realizing an integrated process from fault diagnosis to handling, greatly improving the accuracy and timeliness of satellite fault diagnosis, effectively improving the reliability and operation efficiency of the satellite, and reducing the loss caused by satellite faults. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope, and for those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0016] Figure 1 A flowchart of a satellite fault diagnosis method based on multi-source information fusion provided by an embodiment of the present application is shown.

[0017] Figure 2 A component structure diagram of a satellite fault diagnosis system based on multi-source information fusion provided by an embodiment of the present application is shown. DETAILED DESCRIPTION

[0018] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described below in detail with reference to the drawings in the embodiments of the present application. It should be understood that the drawings in the present application only serve the purpose of illustration and description, and are not used to limit the scope of protection of the present application. In addition, it should be understood that the schematic drawings are not drawn according to the actual proportions. The flowchart in the present application shows the operations implemented according to some embodiments of the present application. It should be understood that the operations of the flowchart can not be implemented in sequence, and the steps without logical context relationship can be reversed in sequence or implemented simultaneously. In addition, under the guidance of the content of the present application, a person skilled in the art can add multiple other operations to the flowchart, or delete multiple operations from the flowchart.

[0019] Figure 1 A flowchart of a satellite fault diagnosis method based on multi-source information fusion provided by an embodiment of the present application is shown. It should be understood that in other embodiments, the order of some steps of the satellite fault diagnosis method based on multi-source information fusion of the present embodiment can be shared according to actual needs, or some steps can be omitted or maintained. The detailed steps of the satellite fault diagnosis method based on multi-source information fusion include:

[0020] Step S110: Based on the multi-source monitoring data set measured in the process of testing the satellite by the emergency quick-response satellite test system, the multi-source monitoring data set contains platform system state data, load system working data and measurement and control link interaction data.

[0021] In this embodiment, in order to effectively diagnose the satellite, it is necessary to comprehensively collect various types of data generated in the satellite test. When the satellite performs various tasks, the platform system, the load system and the TT&C link will generate a large amount of data related to the running state. The platform system state data mainly reflects the running state of the satellite platform itself. Among them, the attitude control parameters are the key data for describing the attitude of the satellite in space, for example, the angular offset of the satellite in three-dimensional space, which can be obtained in real time through the attitude sensor on the satellite. The power output parameters reflect the working state of the satellite power system, including the output voltage, current and other conditions of the power supply, which can reflect whether the power supply normally provides energy for each system of the satellite. The thermal control adjustment parameters record the working conditions of the satellite thermal control system, such as the opening state of the thermal control equipment, the temperature adjustment amplitude, etc., to ensure that the internal equipment of the satellite runs in a suitable temperature environment.

[0022] The load system working data mainly involves the working state of the satellite load equipment. The load running mode reflects the current working way of the load equipment, and different tasks may require the load to adopt different running modes. The data acquisition rate represents the amount of data collected by the load equipment in unit time, which directly affects the efficiency of the satellite in obtaining information. The processing delay parameter refers to the time spent by the load equipment from data acquisition to processing completion, which is very important for tasks that require timely data processing.

[0023] The TT&C link interaction data describes the communication between the satellite and the ground control center. The uplink command response time refers to the time interval from when the satellite receives the command sent by the ground to when it starts to execute the corresponding operation, which reflects the response speed of the satellite to the ground command. The downlink data packet loss rate refers to the proportion of data lost by the satellite during data transmission to the ground, which directly affects the reliability of data transmission. The signal strength fluctuation parameter records the change of the communication signal strength between the satellite and the ground, and the stability of the signal strength has an important influence on the communication quality. Through the data acquisition module on the satellite, these multi-source monitoring data can be collected in real time, and integrated into a multi-source monitoring data set.

[0024] Step S120: performing multi-source information correlation extraction processing on the multi-source monitoring data set to generate a correlation feature set reflecting the coupling relationship between systems.

[0025] In this embodiment, after obtaining the multi-source monitoring data set, further processing of these data is needed to mine the coupling relationship between different systems. The purpose of multi-source information correlation extraction processing is to extract features from the multi-source monitoring data that can reflect the mutual relationship between systems.

[0026] Step S121: Extract attitude control parameters, power output parameters, and thermal control adjustment parameters from the platform system state data as a platform feature subset.

[0027] In the platform system state data, attitude control parameters, power output parameters, and thermal control adjustment parameters are key parameters that can reflect the running state of the platform system. Attitude control parameters can be obtained through attitude sensors such as gyroscopes and accelerometers on the satellite. These sensors monitor the attitude changes of the satellite in space in real time and transmit data to the data processing module. Power output parameters are collected by power monitoring circuits in real time, including output voltage, current, and other data of the power supply. Thermal control adjustment parameters are obtained through thermal control sensors distributed in key parts of the satellite, which monitor temperature changes in real time and feedback to the thermal control system. Extracting these parameters forms a platform feature subset for subsequent analysis of the correlation between the platform system and other systems.

[0028] Step S122: Extract load running mode, data acquisition rate, and processing delay parameters from the load system working data as a load feature subset.

[0029] For load system working data, the load running mode can be determined according to the control instructions of the load device, and different tasks will set different running modes. The data acquisition rate can be calculated by recording the amount of data collected by the load device per unit time, such as by counting the number of data points collected in a period of time to determine the data acquisition rate. The processing delay parameter can be obtained by recording the difference between the data acquisition time and the processing completion time. Extracting these parameters forms a load feature subset for analyzing the mutual influence between the load system and other systems.

[0030] Step S123: Extract uplink instruction response time, downlink data packet loss rate, and signal strength fluctuation parameters from the TT&C link interaction data as a link feature subset.

[0031] In the TT&C link interaction data, the uplink instruction response time can be calculated by recording the time when the satellite receives the ground instruction and the time when the instruction starts to be executed. The downlink data packet loss rate can be determined by comparing the amount of data sent by the satellite and the amount of data received on the ground, such as by calculating the proportion of the difference to the amount of data sent to obtain the packet loss rate. The signal strength fluctuation parameter can be monitored by a signal strength sensor in real time to monitor the change of signal strength, and the fluctuation amplitude and frequency of signal strength are recorded. Extracting these parameters forms a link feature subset for analyzing the coupling relationship between the TT&C link and other systems.

[0032] Step S124: Calculate the dynamic correlation degree between the platform feature subset and the load feature subset, and generate the first correlation feature reflecting the support capability of the platform and the working state of the load.

[0033] In this embodiment, in order to deeply understand the mutual relationship between the platform system and the load system, it is necessary to calculate the dynamic correlation degree between the platform feature subset and the load feature subset.

[0034] Step S1241: Extract the time series data of the power output parameter in the platform feature subset and the load running mode in the load feature subset.

[0035] The time series data of the power output parameter is extracted from the platform feature subset, that is, the change of the power output parameter with time in a period of time is recorded. Similarly, the time series data of the load running mode is extracted from the load feature subset, and the change of the load running mode in the same period of time is recorded. Through analysis of these time series data, the dynamic relationship between the power output and the load running mode can be understood.

[0036] Step S1242: Calculate the matching degree of the power output parameter and the load running mode at the same time point, and the higher the matching degree, the more coordinated the platform power supply capability and the load working demand.

[0037] At the same time point, the matching between the power output parameter and the load running mode is analyzed. For example, when the load is in a high-power consumption running mode, a larger power output is required; when the load is in a low-power consumption mode, the demand for power output is relatively small. By comparing the corresponding relationship between the power output parameter and the load running mode at different time points, the matching degree can be calculated. A matching degree evaluation standard can be set, and the score is given according to the matching degree of the power output parameter and the load running mode, and the higher the score, the higher the matching degree.

[0038] Step S1243: Extract the change trend data of the thermal control adjustment parameter in the platform feature subset and the processing delay parameter in the load feature subset.

[0039] The change trend data of the thermal control adjustment parameter is extracted from the platform feature subset, and the change direction and amplitude of the thermal control adjustment parameter in a period of time are observed. At the same time, the change trend data of the processing delay parameter is extracted from the load feature subset, and the change of the processing delay parameter with time is understood. Through analysis of these two groups of change trend data, the potential relationship between the thermal control adjustment and the load processing delay can be found.

[0040] Step S1244: Calculate the synchronization of the thermal control adjustment parameter change rate and the processing delay parameter change rate, and the higher the synchronization, the more coordinated the thermal environment control and the load processing efficiency.

[0041] The change rate of the thermal control adjustment parameter and the change rate of the processing delay parameter are calculated, and the synchronization of the two change rates is evaluated by comparing the change of the two change rates. For example, a correlation coefficient between the two change rates can be calculated, and the closer the correlation coefficient is to 1, the higher the synchronization is. High synchronization means that the control of the thermal environment can better match the processing efficiency of the load, reducing the impact of temperature changes on load processing.

[0042] Step S1245: Combining the matching degree and the synchronization to form a first association feature reflecting the support capability of the platform and the working state of the load.

[0043] The calculated matching degree of the power output parameter and the load operation mode and the synchronization of the change rate of the thermal control adjustment parameter and the change rate of the processing delay parameter are combined to form a first association feature. The first association feature can comprehensively reflect the coupling relationship between the support capability of the platform system to the load system and the working state of the load system.

[0044] Step S125: Calculate the time sequence matching degree of the load feature subset and the link feature subset to generate a second association feature reflecting the load data output and the link transmission capability.

[0045] In this embodiment, in order to analyze the relationship between the load system and the TT&C link, the time sequence matching degree of the load feature subset and the link feature subset needs to be calculated.

[0046] Step S1251: Align the time stamps of the data acquisition rate in the load feature subset and the downlink data packet loss rate in the link feature subset.

[0047] The time sequence data of the data acquisition rate is extracted from the load feature subset, and the time stamp corresponding to each data acquisition rate is recorded. Similarly, the time sequence data of the downlink data packet loss rate is extracted from the link feature subset, and the corresponding time stamp is recorded. The time stamps of the two groups of data are aligned to ensure that the data comparison is performed at the same time point.

[0048] Step S1252: Calculate the time offset of the peak value of the data acquisition rate and the peak value of the downlink data packet loss rate, and the smaller the time offset, the better the matching between the load data output and the link transmission capability.

[0049] In the timestamp-aligned data, find the peak value of the data acquisition rate and the peak value of the downlink data packet loss rate. Calculate the time offset of the two peak values, that is, the difference between the two peak time points. The smaller the time offset, the better the matching between the peak of the load data output and the change of the link transmission capability, and the link can timely process the data output by the load, reducing the possibility of data loss.

[0050] Step S1253: Extracting the continuous sampling data of the processing delay parameter in the load feature subset and the uplink instruction response time in the link feature subset.

[0051] The continuous sampling data of the processing delay parameter is extracted from the load feature subset, and a plurality of sampling values of the processing delay parameter in a period of time are recorded. Meanwhile, the continuous sampling data of the uplink instruction response time is extracted from the link feature subset. Through analysis of the two groups of continuous sampling data, the relationship between the load processing time consumption and the link instruction response time consumption can be understood.

[0052] Step S1254: Calculating the difference between the mean value of the processing delay parameter and the mean value of the uplink instruction response time, and the smaller the difference is, the more coordinated the load processing time consumption and the link instruction response time consumption are.

[0053] The mean value of the processing delay parameter and the mean value of the uplink instruction response time are calculated, and then the difference between the two mean values is calculated. The smaller the difference is, the closer the time of the load processing data and the time of the link response ground instruction are, and the more coordinated the operation of the whole system is, which can improve the working efficiency of the satellite.

[0054] Step S1255: Combining the time offset and the difference to form a second correlation feature reflecting the load data output and the link transmission capability.

[0055] The calculated time offset and difference are combined to form a second correlation feature. The second correlation feature can comprehensively reflect the coupling relationship between the data output of the load system and the transmission capability of the TT&C link.

[0056] Step S126: Calculating the environmental adaptability correlation degree of the platform feature subset and the link feature subset to generate a third correlation feature reflecting the platform running state and the link communication stability.

[0057] In this embodiment, in order to analyze the environmental adaptability relationship between the platform system and the TT&C link, the environmental adaptability correlation degree of the platform feature subset and the link feature subset needs to be calculated.

[0058] Step S1261: Extracting the spatial coverage data of the attitude control parameter in the platform feature subset and the signal strength fluctuation parameter in the link feature subset.

[0059] The spatial coverage data of the attitude control parameter is extracted from the platform feature subset, that is, the distribution of the attitude control parameter at different spatial positions. Meanwhile, the spatial coverage data of the signal strength fluctuation parameter is extracted from the link feature subset, and the variation of the signal strength fluctuation at different spatial positions is recorded. Through analysis of the two groups of spatial coverage data, the relationship between the platform attitude and the link signal strength can be understood.

[0060] Step S1262: Calculate the deviation value between the satellite pointing angle corresponding to the attitude control parameter and the best communication angle corresponding to the signal strength fluctuation parameter. The smaller the deviation value, the more consistent the platform pointing is with the link communication direction.

[0061] The satellite pointing angle corresponding to the attitude control parameter and the best communication angle corresponding to the signal strength fluctuation parameter are determined. The deviation value between the two angles is calculated. The smaller the deviation value, the closer the satellite pointing is to the best communication direction, which can improve the quality and stability of the link communication.

[0062] Step S1263: Extract the power output parameter in the platform feature subset and the energy consumption data of the uplink instruction response time in the link feature subset.

[0063] The energy consumption data of the power output parameter is extracted from the platform feature subset, i.e. the energy output of the power supply in different time periods. At the same time, the energy consumption data of the uplink instruction response time is extracted from the link feature subset, recording the energy consumption during the execution of the uplink instruction. Through the analysis of these two sets of energy consumption data, the relationship between platform power supply and link communication energy consumption can be understood.

[0064] Step S1264: Calculate the ratio of the redundant power in the power output parameter to the energy consumption corresponding to the uplink instruction response time. The larger the ratio, the more suitable the platform power supply redundancy is to the link communication energy consumption demand.

[0065] The redundant power in the power output parameter is calculated, i.e. the remaining power after subtracting the power required for the normal operation of each system of the satellite from the power output. The energy consumption corresponding to the uplink instruction response time is calculated, i.e. the power consumed during the execution of the uplink instruction. Then the ratio of the redundant power to the energy consumption is calculated. The larger the ratio, the better the platform power supply redundancy can meet the energy consumption demand of the link communication, ensuring the stability of the link communication.

[0066] Step S1265: Combine the deviation value and the ratio to form a third associated feature reflecting the platform running state and the link communication stability.

[0067] The calculated deviation value and ratio are combined to form a third associated feature. This third associated feature can comprehensively reflect the coupling relationship between the platform system running state and the TT&C link communication stability.

[0068] Step S127: Combine the first associated feature, the second associated feature and the third associated feature to form an associated feature set reflecting the coupling relationship between the systems.

[0069] In this embodiment, the first associated feature, the second associated feature and the third associated feature are combined to form an associated feature set. The associated feature set can comprehensively reflect the coupling relationship between the platform system, the payload system and the measurement and control link.

[0070] Step S130: inputting the associated feature set into a pre-trained satellite fault diagnosis model for fault mode analysis processing to obtain a fault feature description containing a fault triggering node and an influence link.

[0071] In this embodiment, after generating the associated feature set, it is input into a pre-trained satellite fault diagnosis model for fault mode analysis processing. The satellite fault diagnosis model is trained by a large amount of historical data and can analyze and process the associated feature set to find hidden fault information.

[0072] Step S131: receiving the associated feature set through the feature encoding layer, and performing time sequence feature extraction processing on the first associated feature, the second associated feature and the third associated feature to generate an encoded feature vector with time dimension representation.

[0073] The satellite fault diagnosis model includes a feature encoding layer, a coupling analysis layer and a fault analysis layer. The feature encoding layer first receives the associated feature set and performs time sequence feature extraction processing on the first associated feature, the second associated feature and the third associated feature. Time sequence feature extraction refers to extracting features that can reflect the change rule from the time sequence data of these associated features. For example, by analyzing the numerical change and trend change of the associated features at different time points, an encoded feature vector with time dimension representation is generated. The encoded feature vector can more effectively represent the change of the associated feature over time.

[0074] Step S132: performing system interdependence relationship modeling processing on the encoded feature vector through the coupling analysis layer to identify key coupling nodes between the platform system, the payload system and the measurement and control link.

[0075] In this embodiment, after the coupling analysis layer receives the encoded feature vector, it performs system interdependence relationship modeling processing on it.

[0076] Step S1321: performing interactive influence analysis processing on the platform feature encoding, the payload feature encoding and the link feature encoding in the encoded feature vector through the coupling analysis layer to calculate the influence weight of platform feature change on payload feature, the influence weight of payload feature change on link feature and the influence weight of link feature change on platform feature.

[0077] In the encoding feature vector, platform feature encoding, load feature encoding and link feature encoding are respectively included. The coupling analysis layer performs interaction analysis and processing on these encodings, calculates the influence weight of platform feature change on load feature, the influence weight of load feature change on link feature and the influence weight of link feature change on platform feature by analyzing the correlation and change relationship between different feature encodings. For example, the correlation coefficient between different feature encodings can be calculated by statistical analysis method, and the influence weight is determined according to the size of the correlation coefficient.

[0078] Step S1322: constructing an inter-system dependency graph according to the influence weight, wherein the nodes in the dependency graph represent system modules and the edges represent the influence relationship between modules.

[0079] According to the calculated influence weight, an inter-system dependency graph is constructed. In the inter-system dependency graph, the nodes represent the modules in the platform system, the load system and the measurement and control link, and the edges represent the influence relationship between the modules, and the weight of the edge is determined by the influence weight. Through the inter-system dependency graph, the mutual dependency relationship between different system modules can be intuitively displayed.

[0080] Step S1323: calculating the centrality index of each node in the dependency graph, wherein the higher the centrality index, the more critical the node is in system interaction.

[0081] In the constructed inter-system dependency graph, the centrality index of each node is calculated. The centrality index can measure the importance of the node in the whole system interaction, and common centrality indexes include degree centrality, betweenness centrality, etc. The degree centrality refers to the number of connected edges of the node, and the betweenness centrality refers to the frequency of the node appearing in the shortest path between other nodes. The higher the centrality index, the more critical the node plays in the system interaction.

[0082] Step S1324: screening out the nodes with a centrality index exceeding a preset threshold as key coupling nodes between the platform system, the load system and the measurement and control link.

[0083] A preset threshold is set, and the nodes with a centrality index exceeding the threshold are screened out as key coupling nodes. The state change of the above key coupling nodes may have a greater impact on other system modules.

[0084] Step S133: performing abnormal feature detection processing based on the key coupling nodes by the fault analysis layer, and positioning the original monitoring data source node corresponding to the abnormal feature as a fault trigger node.

[0085] In this embodiment, the fault analysis layer receives the key coupling node information and performs abnormal feature detection processing.

[0086] Step S1331: Extract the feature values corresponding to each key coupling node in the encoding feature vector through the fault analysis layer, and calculate the deviation degree of the feature values from the feature mean value in the historical normal operation state.

[0087] The feature values corresponding to each key coupling node are extracted from the encoding feature vector, and the feature mean values of these key coupling nodes in the historical normal operation state are obtained. The deviation degree between the current feature values and the historical feature mean values can be measured by calculating the difference or ratio between them. The greater the deviation degree, the more abnormal the state of the key coupling node.

[0088] Step S1332: Screen out the key coupling nodes with a deviation degree exceeding a preset threshold as abnormal coupling nodes.

[0089] A preset threshold is set to screen out the key coupling nodes with a deviation degree exceeding the threshold as abnormal coupling nodes. The above abnormal coupling nodes may be potential positions of the fault occurrence.

[0090] Step S1333: Trace the original data mapping relationship of the abnormal coupling node in the feature encoding process to determine the original monitoring data type and specific parameters corresponding to the abnormal coupling node.

[0091] By tracing the original data mapping relationship of the abnormal coupling node in the feature encoding process, the original monitoring data type and specific parameters corresponding to the abnormal coupling node are found. For example, a certain abnormal coupling node may correspond to the attitude control parameter in the platform system state data. Through further analysis, it can be determined that which attitude control parameter is abnormal.

[0092] Step S1334: According to the original monitoring data type and specific parameters, locate the satellite hardware module or software function module generating the data as the fault trigger node.

[0093] In this embodiment, according to the determined original monitoring data type and specific parameters, the satellite hardware module or software function module generating the data can be located. For example, if the abnormal original monitoring data is a parameter related to satellite attitude control, then through the hardware architecture and software logic of the satellite, the hardware modules responsible for attitude control such as gyroscopes and accelerometers, or the software function modules corresponding to the attitude control algorithm can be traced to determine the fault trigger node. In this way, through in-depth analysis of the abnormal coupling node, the source of the fault occurrence can be accurately found.

[0094] Step S134: According to the connection path of the fault trigger node in the coupling relationship, trace the system module sequence of abnormal feature propagation as the influence link.

[0095] In this embodiment, after the fault trigger node is determined, the propagation path of the fault needs to be further analyzed. Due to the coupling relationship between systems, the abnormal characteristics of the fault trigger node will propagate to other system modules along a certain path. By viewing the inter-system dependency graph, the connection edges between the fault trigger node and other modules are found, and according to the influence relationship represented by these connection edges, the sequence of system modules through which the abnormal characteristics propagate is tracked. For example, if the fault trigger node is a hardware module in the platform system, it is connected to a module of the payload system through an edge in the dependency graph, and the edge indicates that the abnormality of the platform module will affect the payload module, so it can be determined that the payload module is the next node through which the abnormal characteristics propagate. In this way, the sequence of system modules through which the abnormal characteristics propagate can be obtained, which is the influence link. The influence link can clearly show how the fault propagates between different system modules starting from the trigger node.

[0096] Step S135: combining the fault trigger node and the influence link to form a fault feature description containing the fault trigger node and the influence link.

[0097] In this embodiment, the determined fault trigger node and the influence link are combined to form a fault feature description. The fault feature description is a comprehensive description of the satellite fault situation, which not only identifies the source of the fault (the fault trigger node), but also shows the propagation path of the fault in the system (the influence link).

[0098] Step S140: matching the fault feature description with a preset fault knowledge base to generate a fault diagnosis result containing a fault type identifier and a propagation path.

[0099] In this embodiment, after the fault feature description is obtained, it needs to be matched with a preset fault knowledge base to generate a fault diagnosis result. The fault knowledge base is a database that stores a large number of historical fault cases, each case containing information such as fault type, fault trigger node, influence link, etc.

[0100] Step S141: extracting the module identifier of the fault trigger node and the module sequence of the influence link from the fault feature description.

[0101] The module identifier of the fault trigger node is extracted from the fault feature description, which can uniquely identify the specific module where the fault occurs. At the same time, the module sequence of the influence link is extracted, which records the sequence of modules through which the fault propagates in the system. The above information is the key basis for fault matching.

[0102] Step S142: querying the historical fault cases in the fault knowledge base that match the module identifier and the module sequence.

[0103] The module identification of the extracted fault trigger node and the module sequence of the affected link are used to query the fault knowledge base. By comparing these information with the corresponding information in the historical fault cases, the historical fault case that matches is found. The matching process can use the method of similarity calculation, for example, calculating the consistency of the module identification and the similarity of the module sequence, the higher the similarity, the more likely the case is the matching case.

[0104] Step S143: Extract the fault type label in the historical fault case as the fault type identification of the current fault.

[0105] After finding the matching historical fault case, the fault type label is extracted from the historical fault case as the fault type identification of the current fault. The fault type identification can clearly indicate the type of the current fault, such as power failure, communication failure, attitude control failure, etc.

[0106] Step S144: Extract the abnormal propagation path in the historical fault case as the propagation path of the current fault.

[0107] Similarly, the abnormal propagation path is extracted from the matching historical fault case as the propagation path of the current fault. The abnormal propagation path can further verify the accuracy of the affected link.

[0108] Step S145: Combine the fault type identification and the propagation path to form a fault diagnosis result containing the fault type identification and the propagation path.

[0109] The extracted fault type identification and the propagation path are combined to form a fault diagnosis result. The fault diagnosis result can comprehensively describe the type and propagation of the current fault.

[0110] Step S150: Generate a satellite fault handling scheme containing positioning coordinates and disposal strategies according to the fault diagnosis result.

[0111] In this embodiment, after obtaining the fault diagnosis result, a satellite fault handling scheme needs to be generated according to the result to effectively handle the fault.

[0112] Step S151: Extract the module identification of the fault trigger node from the fault diagnosis result, query the satellite cabin layout corresponding to the module identification, and obtain the three-dimensional coordinates of the module in the satellite cabin as the positioning coordinates.

[0113] A module identifier of the fault triggering node is extracted from the fault diagnosis result, and then a satellite cabin layout map is queried. The satellite cabin layout map records the position information of each module of the satellite in the cabin in detail. Through the module identifier, the corresponding module can be found in the layout map, and its three-dimensional coordinates in the satellite cabin can be obtained. The three-dimensional coordinates are the positioning coordinates of the fault triggering node, which can help the ground control personnel accurately understand the location of the fault.

[0114] Step S152: According to the fault type identifier, query the preset disposal strategy library to extract the emergency operation instruction and repair priority matching the fault type identifier.

[0115] According to the fault type identifier in the fault diagnosis result, the preset disposal strategy library is queried. The disposal strategy library stores emergency operation instructions and repair priorities for different fault types. By matching the fault type identifier in the disposal strategy library, the emergency operation instruction and repair priority corresponding to the current fault type are extracted. The emergency operation instruction can guide the ground control personnel to take appropriate measures to alleviate the impact of the fault, such as adjusting the working mode of the satellite, restarting the related module, etc. The repair priority can help determine the sequence of repairing the fault, ensuring that limited resources can be used to handle critical faults first.

[0116] Step S153: According to the propagation path, predict the extension direction of the fault within a preset time window, and generate operation constraint conditions containing extension direction restrictions.

[0117] According to the propagation path in the fault diagnosis result, the extension direction of the fault within a preset time window is predicted. Since the propagation of the fault has certain regularity, the next system module that may be affected by the fault can be roughly predicted by analyzing the propagation path. According to the predicted extension direction, operation constraint conditions containing extension direction restrictions are generated. For example, if it is predicted that the fault may extend to a certain specific system module, then when performing fault handling operations, further impact on the module needs to be avoided, and corresponding operation constraint conditions are developed to ensure the safety and effectiveness of the fault handling process.

[0118] Step S154: Combine the positioning coordinates, emergency operation instruction, repair priority and operation constraint conditions to form a satellite fault handling scheme containing positioning coordinates and disposal strategies.

[0119] The acquired positioning coordinates, the extracted emergency operation instructions, the repair priority, and the generated operation constraint conditions are combined to form a satellite fault processing scheme. The satellite fault processing scheme comprehensively considers the position of the fault, the processing measures, the repair priority, and the operation restrictions, and can provide ground control personnel with a comprehensive and effective fault processing guide, helping them to quickly and accurately process satellite faults and ensure the normal operation of the satellite.

[0120] In this embodiment, the training process of the satellite fault diagnosis model enables the model to learn the fault patterns and rules between systems, thereby improving the accuracy of fault diagnosis. The satellite fault diagnosis model includes a feature encoding layer, a coupling analysis layer, and a fault analysis layer.

[0121] First, a large amount of satellite historical operation data is collected, including a multi-source monitoring data set and corresponding fault labels. The multi-source monitoring data set includes platform system state data, payload system working data, and measurement and control link interaction data. The fault labels clearly indicate the fault type and fault trigger node corresponding to each data sample. The data is preprocessed, including data cleaning, normalization, and other operations, to ensure data quality and consistency. Data cleaning can remove noise and outliers in the data, and normalization can convert data of different dimensions to the same scale, facilitating model processing.

[0122] Then, the parameters of the feature encoding layer, the coupling analysis layer, and the fault analysis layer of the satellite fault diagnosis model are initialized. Random initialization can be used to set the initial weights and biases of the neurons in each layer. The above initial parameters will be continuously adjusted during the training process to enable the model to better fit the training data.

[0123] Next, the preprocessed training data is input into the satellite fault diagnosis model for forward propagation. In the feature encoding layer, the input multi-source monitoring data is processed for time series feature extraction to generate an encoded feature vector. In the coupling analysis layer, the encoded feature vector is processed for system interdependence modeling to identify key coupling nodes. In the fault analysis layer, abnormal feature detection is performed based on the key coupling nodes to output the fault prediction result.

[0124] The fault prediction result of the model is compared with the real fault label to calculate the loss function. The loss function can measure the difference between the model's prediction result and the true value. Common loss functions include cross-entropy loss function, which can make the model's prediction result closer to the true value.

[0125] According to the calculated loss function, backpropagation is performed. The process of backpropagation is to calculate the gradient of the loss function with respect to each parameter of the model through chain rule. The gradient represents the influence of each parameter on the loss function. According to the size and direction of the gradient, the parameters of the model can be adjusted to make the loss function gradually decrease.

[0126] According to the gradient calculated by backpropagation, the parameters of the model are updated using optimization algorithms such as stochastic gradient descent algorithm. The optimization algorithm adjusts the weights and biases of the model according to the gradient information with a certain learning rate, so that the performance of the model on the training data becomes better and better.

[0127] Repeat the above steps until the training stopping condition is met. The training stopping condition can be that the pre-set training round is reached, or the loss function converges to a smaller value. Through continuous training, the model can learn the failure patterns and rules between systems, and improve the accuracy of fault diagnosis.

[0128] Finally, the trained satellite fault diagnosis model is evaluated using test data. Test data is a separate data set from training data. The performance of the model is evaluated by calculating the accuracy, recall rate and other indicators on the test data. If the performance of the model meets the requirements, it can be used for actual satellite fault diagnosis; if the performance does not meet the requirements, the training parameters need to be adjusted or re-training is needed.

[0129] During data collection, data desensitization techniques are used for data that may involve privacy-sensitive data. For example, sensitive data such as the identity of satellite operators is replaced or masked to ensure that the data does not contain identifiable personal information without affecting fault diagnosis analysis. During data transmission, encryption algorithms are used to encrypt data, such as symmetric encryption algorithms to encrypt data. Only authorized parties with decryption keys can decrypt the data. In terms of data storage, data is stored in secure databases with strict access control, and only authorized personnel can access the data. At the same time, regular backups are made to prevent data loss or damage. Through these technical means, the security of privacy-sensitive data can be effectively protected to prevent data leakage.

[0130] Figure 2The illustrated satellite fault diagnosis system 100 based on multi-source information fusion comprises a processor 1001 and a memory 1003. The processor 1001 and the memory 1003 are connected, for example, through a bus 1002. Optionally, the satellite fault diagnosis system 100 based on multi-source information fusion can further comprise a transceiver 1004, which can be used for data interaction, such as data sending and / or data receiving, between the server and other servers. It should be noted that the transceiver 1004 is not limited to one in actual scheduling, and the structure of the satellite fault diagnosis system 100 based on multi-source information fusion does not constitute a limitation to the embodiments of the present application.

[0131] The processor 1001 can be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It can implement or execute various exemplary logical blocks, modules and circuits described in combination with the disclosure of the present application. The processor 1001 can also be a combination of computing functions, such as one or more microprocessor combinations, combinations of DSP and microprocessor, etc.

[0132] The bus 1002 can comprise a channel for transmitting information between the above-mentioned components. The bus 1002 can be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The bus 1002 can be divided into an address bus, a data bus, a control bus, etc. For the sake of convenience, Figure 2 In the figure, only one thick line is used to represent the bus, but it does not mean that there is only one bus or only one type of bus.

[0133] The memory 1003 can be a ROM (Read Only Memory), or other type of static storage device that can store static information and instructions, a RAM (Random Access Memory), or other type of dynamic storage device that can store information and instructions, an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, a magnetic disk storage or other magnetic storage devices, or any other medium capable of storing and carrying program codes that can be read by a computer, without limitation.

[0134] The memory 1003 is configured to store program codes for implementing the embodiments of the present application, and the processor 1001 is configured to control the execution of the program codes stored in the memory 1003. The processor 1001 is configured to execute the program codes stored in the memory 1003 to implement the steps shown in the foregoing method embodiments.

[0135] The embodiments of the present application provide a computer readable storage medium, and the computer readable storage medium stores program codes. The program codes are executed by a processor to implement the steps and corresponding contents of the foregoing method embodiments.

[0136] It should be understood that, although each operation step in the flowcharts of the embodiments of the present application is represented by a directional connection, the implementation order of the steps is not limited to the order covered by the directional connection. Unless otherwise specified herein, in some implementation scenarios of the embodiments of the present application, the implementation steps in each flowchart can be executed in other orders based on requirements. In addition, part or all of the steps in each flowchart can include a plurality of sub-steps or a plurality of stages, and part or all of the sub-steps or stages can be executed in the same stage, or each of the sub-steps or stages can be executed in different stages. In the scenario where the execution stages are different, the execution order of the sub-steps or stages can be flexibly configured based on requirements, and the embodiments of the present application do not limit this.

[0137] The above description is only optional implementation manners of some implementation scenarios of the present application, and it should be pointed out that, for those skilled in the art, other similar implementation manners according to the technical concept of the present application can be adopted without departing from the technical concept of the present application, and these also belong to the protection scope of the embodiments of the present application.

Claims

1. A satellite fault diagnosis method based on multi-source information fusion, characterized in that, The method comprises: Based on the multi-source monitoring data set generated by the emergency quick-response satellite test system in the process of testing the satellite, the multi-source monitoring data set contains platform system state data, load system operation data and measurement and control link interaction data; The multi-source information correlation extraction processing is performed on the multi-source monitoring data set to generate a correlation feature set reflecting the coupling relationship between systems; The correlation feature set is input into a pre-trained satellite fault diagnosis model for fault mode analysis processing to obtain a fault feature description containing a fault trigger node and an influence link; Based on the fault feature description, a fault knowledge base is matched to generate a fault diagnosis result containing a fault type identifier and a propagation path; According to the fault diagnosis result, a satellite fault handling strategy is generated. 2.The satellite fault diagnosis method based on multi-source information fusion according to claim 1, characterized in that, The multi-source information correlation extraction processing is performed on the multi-source monitoring data set to generate a correlation feature set reflecting the coupling relationship between systems, comprising: Extracting attitude control parameters, power output parameters and thermal control adjustment parameters from the platform system state data as a platform feature subset; Extracting load operation mode, data acquisition rate and processing delay parameters from the load system operation data as a load feature subset; Extracting uplink instruction response time, downlink data packet loss rate and signal strength fluctuation parameters from the measurement and control link interaction data as a link feature subset; Calculate the dynamic correlation degree of the platform feature subset and the load feature subset to generate a first correlation feature reflecting the platform support capability and the load working state; Calculate the time sequence matching degree of the load feature subset and the link feature subset to generate a second correlation feature reflecting the load data output and the link transmission capability; Calculate the environmental adaptability correlation degree of the platform feature subset and the link feature subset to generate a third correlation feature reflecting the platform running state and the link communication stability; The first correlation feature, the second correlation feature and the third correlation feature are combined to form a correlation feature set reflecting the coupling relationship between systems. 3.The satellite fault diagnosis method based on multi-source information fusion according to claim 2, characterized in that, The dynamic correlation degree of the platform feature subset and the load feature subset is calculated to generate a first correlation feature reflecting the platform support capability and the load working state, comprising: Extracting time series data of power output parameters in the platform feature subset and load operation mode in the load feature subset; Calculate the matching degree of power output parameters and load operation mode at the same time point. The higher the matching degree, the more coordinated the platform power supply capability and the load working demand are; Extracting change trend data of thermal control adjustment parameters in the platform feature subset and processing delay parameters in the load feature subset; Calculate the synchronization of thermal control adjustment parameter change rate and processing delay parameter change rate. The higher the synchronization, the more coordinated the thermal environment control and load processing efficiency are; The matching degree and the synchronization are combined to form the first correlation feature reflecting the platform support capability and the load working state. 4.The satellite fault diagnosis method based on multi-source information fusion according to claim 2, characterized in that, The time sequence matching degree of the load feature subset and the link feature subset is calculated to generate a second correlation feature reflecting the load data output and the link transmission capability, comprising: extracting timestamp alignment data of the data acquisition rate in the load feature subset and the downlink data packet loss rate in the link feature subset; calculating a time offset of the data acquisition rate peak value and the downlink data packet loss rate peak value, and the smaller the time offset, the more matched the load data output and the link transmission capability are; extracting continuous sampling data of the processing delay parameter in the load feature subset and the uplink instruction response time in the link feature subset; calculating a difference value of the processing delay parameter mean value and the uplink instruction response time mean value, and the smaller the difference value, the more coordinated the load processing time consumption and the link instruction response time consumption are; combining the time offset and the difference value to form a second correlation feature reflecting the load data output and the link transmission capability. 5.The satellite fault diagnosis method based on multi-source information fusion according to claim 2, characterized in that, The calculation of the environmental adaptability correlation degree of the platform feature subset and the link feature subset generates a third correlation feature reflecting the platform running state and the link communication stability, including: extracting spatial coverage data of the attitude control parameter in the platform feature subset and the signal strength fluctuation parameter in the link feature subset; calculating a deviation value of the satellite pointing angle corresponding to the attitude control parameter and the best communication angle corresponding to the signal strength fluctuation parameter, and the smaller the deviation value, the more matched the platform pointing and the link communication direction are; extracting energy consumption data of the power output parameter in the platform feature subset and the uplink instruction response time in the link feature subset; calculating a ratio of the redundant power in the power output parameter and the energy consumption corresponding to the uplink instruction response time, and the larger the ratio, the more adaptive the platform power supply redundancy and the link communication energy consumption demand are; combining the deviation value and the ratio to form a third correlation feature reflecting the platform running state and the link communication stability. 6.The satellite fault diagnosis method based on multi-source information fusion according to claim 2, characterized in that, The satellite fault diagnosis model includes a feature encoding layer, a coupling analysis layer and a fault analysis layer; The input of the correlation feature set into the pre-trained satellite fault diagnosis model for fault mode analysis processing generates a fault feature description containing a fault trigger node and an affected link, including: The feature encoding layer receives the correlation feature set and performs time sequence feature extraction processing on the first correlation feature, the second correlation feature and the third correlation feature, respectively, to generate an encoded feature vector with time dimension representation; The coupling analysis layer performs system interdependence relationship modeling processing on the encoded feature vector to identify key coupling nodes between the platform system, the load system and the TT&C link; The fault analysis layer performs abnormal feature detection processing based on the key coupling nodes to locate the original monitoring data source node corresponding to the abnormal feature as the fault trigger node; According to the connection path of the fault trigger node in the coupling relationship, the system module sequence of abnormal feature propagation is tracked as the affected link; The fault trigger node and the affected link are combined to form a fault feature description containing the fault trigger node and the affected link.

7. The satellite fault diagnosis method based on multi-source information fusion according to claim 6, characterized in that, The coupling analysis layer performs system interdependence relationship modeling processing on the encoded feature vector to identify key coupling nodes between the platform system, the load system and the TT&C link, including: The platform feature encoding, the load feature encoding and the link feature encoding in the encoded feature vector are interactively analyzed and processed by a coupling analysis layer to calculate the influence weight of platform feature change on load feature, the influence weight of load feature change on link feature and the influence weight of link feature change on platform feature; An inter-system dependency graph is constructed according to the influence weights, wherein the nodes represent system modules and the edges represent the influence relationship between modules; The centrality index of each node in the dependency graph is calculated, and the higher the centrality index, the more critical the node is in system interaction; The nodes with a centrality index exceeding a preset threshold are screened out as key coupling nodes between platform system, load system and measurement and control link. 8.The satellite fault diagnosis method based on multi-source information fusion according to claim 6, characterized in that, The abnormal feature detection processing based on the key coupling nodes by the fault analysis layer locates the original monitoring data source node corresponding to the abnormal feature as a fault trigger node, including: The feature values corresponding to each key coupling node in the encoded feature vector are extracted by the fault analysis layer, and the deviation degree of the feature values from the feature mean value in the historical normal operation state is calculated; The key coupling nodes with a deviation degree exceeding a preset threshold are screened out as abnormal coupling nodes; The original data mapping relationship of the abnormal coupling nodes in the feature encoding process is tracked to determine the original monitoring data type and specific parameters corresponding to the abnormal coupling nodes; According to the original monitoring data type and specific parameters, the satellite hardware module or software functional module generating the data is located as the fault trigger node. 9.The satellite fault diagnosis method based on multi-source information fusion according to claim 1, characterized in that, The fault feature description is matched with a preset fault knowledge base to generate a fault diagnosis result containing a fault type identifier and a propagation path, including: The module identifier of the fault trigger node and the module sequence of the influence link are extracted from the fault feature description; The historical fault cases matching the module identifier and the module sequence in the fault knowledge base are queried; The fault type label in the historical fault cases is extracted as the fault type identifier of the current fault; The abnormal propagation path in the historical fault cases is extracted as the propagation path of the current fault; The fault type identifier and the propagation path are combined to form a fault diagnosis result containing a fault type identifier and a propagation path; And a satellite fault handling scheme containing a positioning coordinate and a disposal strategy is generated according to the fault diagnosis result, including: The module identifier of the fault trigger node is extracted from the fault diagnosis result, the satellite cabin layout map corresponding to the module identifier is queried, and the three-dimensional coordinate of the module identifier in the satellite cabin is obtained as the positioning coordinate; The emergency operation instruction and the repair priority matching the fault type identifier are extracted from the preset disposal strategy library according to the fault type identifier; The expansion direction of the fault in a preset time window is predicted according to the propagation path to generate an operation constraint condition containing an expansion direction restriction; The positioning coordinate, the emergency operation instruction, the repair priority and the operation constraint condition are combined to form a satellite fault handling scheme containing a positioning coordinate and a disposal strategy.

10. A satellite fault diagnosis system based on multi-source information fusion, characterized in that, The application relates to a satellite fault diagnosis method based on multi-source information fusion, comprising a processor and a computer readable storage medium, wherein the computer readable storage medium stores machine executable instructions, and the machine executable instructions are executed by the processor to realize the satellite fault diagnosis method based on multi-source information fusion in any one of claims 1-9.

Images