Satellite autonomous health management system and method implemented in stages

Abstract

The invention discloses a satellite autonomous health management method implemented by stages, which is implemented according to a satellite autonomous health management system implemented by stages, and comprises the following steps: S1, training an intelligent algorithm by a first-stage ground algorithm training module by using health data and a rule-based result, the trained intelligent algorithm is uploaded to an on-satellite intelligent algorithm management module; s2, in the second stage, an on-satellite ground algorithm training module trains the intelligent algorithm trained in the step S1 by using health data and a final result, and uploads the trained intelligent algorithm to an on-satellite intelligent algorithm management module; and S3, in the third stage, the ground algorithm training module trains the intelligent algorithm trained in the step S2 by using the health data and the final result, and uploads the trained intelligent algorithm to an on-satellite intelligent algorithm management module. According to the invention, the management strategy can be automatically adjusted according to different stages of satellite operation so as to realize the best health management effect.

Description

Technical Field

[0001] The present invention belongs to the technical field of spacecraft health management, and in particular relates to a satellite autonomous health management system and method implemented in stages. Background Art

[0002] With the rapid development of space technology, satellites play an increasingly important role in communications, navigation, remote sensing, etc. However, satellites face challenges in extreme environments and complex working conditions in space, and effective health management strategies are needed to ensure their normal operation.

[0003] Traditional satellite health management mainly relies on monitoring and analysis by ground personnel. This approach has the following limitations:

[0004] Slow response time: Ground control personnel need time to analyze data and make decisions, which may lead to delayed response in emergency situations.

[0005] Human Error: Reliance on manual analysis increases the likelihood of error, especially in complex and high-pressure work environments.

[0006] High resource consumption: A large number of ground personnel and equipment are required to support satellite monitoring and maintenance, which increases costs.

[0007] Insufficient data utilization: A large amount of satellite operation data is not fully utilized, and there is a lack of effective data analysis and processing methods.

[0008] The development of artificial intelligence technology, especially machine learning and deep learning technology, provides new solutions for satellite health management. Artificial intelligence algorithms are able to process and analyze large amounts of data, identify patterns and trends, predict failures, and provide decision support. This enables satellite health management to achieve the following advantages:

[0009] Automation: Reduce reliance on manual intervention and improve response speed and processing efficiency.

[0010] Accuracy: Improve the accuracy of fault diagnosis and prediction through algorithm optimization.

[0011] Predictive: Leverages historical data and pattern recognition to predict potential failures and problems.

[0012] Cost-effectiveness: Reduced reliance on ground personnel, lower operating costs.

[0013] Traditional satellite health management is mainly based on rules, and satellite fault diagnosis is performed through expert systems. This method has a high diagnostic accuracy for known problems, but it is difficult to detect gradual and trending problems, and it can only be diagnosed based on existing rules, and new rules cannot be automatically generated. Health management based on intelligent algorithms is trained through a large amount of data, and the accuracy of intelligent algorithms is greatly dependent on training data. In order to achieve a relatively high accuracy, it is necessary to train with a large amount of data for a period of time. Therefore, traditional health management has problems such as rule limitations, insufficient data utilization, and limited processing capabilities, while health management based on intelligent algorithms requires long-term data training to achieve a certain accuracy. Both have their advantages and disadvantages. Summary of the invention

[0014] The purpose of the present invention is to provide a satellite autonomous health management system and method that is implemented in stages. The present invention can automatically adjust the management strategy according to the different stages of satellite operation to achieve the best health management effect.

[0015] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0016] The satellite autonomous health management system implemented in stages includes on-board data acquisition module, on-board knowledge management module, on-board intelligent algorithm management module, on-board business module, ground data management module, ground algorithm training module, and ground comprehensive management module, among which:

[0017] The onboard data collection module is deployed onboard the satellite to directly collect the satellite's health data;

[0018] The on-board knowledge management module is deployed on-board to build a detailed knowledge base and provide rules for the on-board business modules;

[0019] The on-board intelligent algorithm management module is deployed on the satellite to provide intelligent algorithms for the on-board business modules;

[0020] The on-board business module is deployed on the satellite, and business features are extracted based on the health data of the on-board data acquisition module. According to different feature values, the corresponding rules in the on-board knowledge management module are called in the first stage to obtain rule-based results, and the rule-based results are saved in the ground data management module; the corresponding rules in the on-board knowledge management module are called in the second and third stages to obtain rule-based results, and the intelligent algorithm in the on-board intelligent algorithm management module is called to obtain the result based on the intelligent algorithm. After judging the rule-based results and the result based on the intelligent algorithm, the final result is saved in the ground data management module;

[0021] The ground data management module is deployed on the ground to manage the satellite's health data and results;

[0022] The ground algorithm training module is deployed on the ground, calls the health data and results of the ground data management module, conducts intelligent algorithm training, and injects the trained intelligent algorithm into the onboard intelligent algorithm management module;

[0023] The ground integrated management module is deployed on the ground to provide ground control personnel with more comprehensive satellite status analysis and prediction, and to perform manual judgment on scenarios where manual judgment is required by the onboard business modules.

[0024] The phased approach to satellite autonomous health management includes the following steps:

[0025] S1, in the first stage, the on-board business module manages the health of the satellite based on rules and obtains rule-based results. The ground algorithm training module uses health data and rule-based results to train the intelligent algorithm. When the intelligent algorithm reaches a certain accuracy, the system automatically or manually determines until entering the second stage. At the same time, the trained intelligent algorithm is injected into the on-board intelligent algorithm management module;

[0026] S2, in the second stage, the on-board business module manages the health of the satellite based on the rules and the intelligent algorithm trained in S1, obtains the rule-based results and the intelligent algorithm-based results, and judges the rule-based results and the intelligent algorithm-based results in S2 to obtain the final results. The ground algorithm training module uses the health data in S2 and the final results in S2 to train the intelligent algorithm trained in S1. When the intelligent algorithm reaches a certain accuracy rate, the system automatically judges or manually judges until entering the third stage. At the same time, the trained intelligent algorithm is injected into the on-board intelligent algorithm management module;

[0027] S3, the third stage on-board business module manages the health of the satellite based on the rules and the intelligent algorithm trained in S2, obtains the rule-based results and the intelligent algorithm-based results, and judges the rule-based results and the intelligent algorithm-based results in S3 to get the final result. The ground algorithm training module uses the health data in S3 and the final result in S3 to train the intelligent algorithm trained in S2, and at the same time, the trained intelligent algorithm is injected into the on-board intelligent algorithm management module.

[0028] Preferably, in S1, the ground algorithm training module uses health data and rule-based results to train the intelligent algorithm. When the intelligent algorithm reaches a certain accuracy, the system automatically judges or manually judges until entering the second stage. At the same time, the specific steps of injecting the trained intelligent algorithm into the on-board intelligent algorithm management module are:

[0029] S10301, the ground algorithm training module obtains health data and rule-based results from the ground data management module in a periodic or data-triggered manner to perform intelligent algorithm training;

[0030] S10302, the ground algorithm training module determines whether to enter the second stage through automatic judgment based on the intelligent algorithm training results, and if so, determines whether the accuracy obtained by the intelligent algorithm is better than the preset value; if not, performs manual judgment;

[0031] S10303, the ground algorithm training module determines whether the accuracy of the intelligent algorithm is better than the preset value. If yes, it enters the second stage, injects the trained intelligent algorithm into the on-board intelligent algorithm management module for intelligent algorithm update, and informs the on-board business module at the same time. The on-board business module enters the second stage and the process ends. If no, repeat S10301 to S10303;

[0032] The user manually determines whether to enter the second stage based on the intelligent algorithm training results; if yes, it enters the second stage, injects the trained intelligent algorithm into the on-board intelligent algorithm management module for intelligent algorithm update, and informs the on-board business module at the same time, the on-board business module enters the second stage, and the process ends. Otherwise, repeat S10301 to S10303.

[0033] Preferably, the specific steps of S2 are:

[0034] S201, the ground data management module obtains health data of the satellite;

[0035] S202, the on-board business module extracts business features based on the health data of the on-board data acquisition module, and respectively calls the rules corresponding to the on-board knowledge management module and the intelligent algorithm corresponding to the on-board intelligent algorithm management module to obtain rule-based results and intelligent algorithm-based results;

[0036] S203, judging the rule-based results and the intelligent algorithm-based results, sending the judged results to the ground integrated management module, and saving the final results to the ground data management module;

[0037] S204, the ground algorithm training module obtains the health data and final results of the ground data management module and performs intelligent algorithm training. When the intelligent algorithm reaches a certain accuracy rate, the system makes autonomous decisions or manual judgments until entering the third stage.

[0038] Preferably, the specific steps of S203 are:

[0039] The onboard business module determines whether the rule-based results and the intelligent algorithm-based results are consistent. If they are, the consistent results are sent to the ground user for review via telemetry. After the review is passed, the final results are saved to the ground data management module; otherwise, it determines whether manual judgment is required;

[0040] The on-board business module determines whether manual judgment is required. If so, the ground integrated management module performs manual judgment, obtains the final result after manual judgment, saves the final result to the ground data management module, and informs the on-board business module of the final result through remote control; if not, determines whether to default to the rule-based result;

[0041] The onboard business module determines whether to use the rule-based result as the default. If so, the rule-based result is sent to the ground integrated management module for review via telemetry. After the review is passed, the final result is saved to the ground data management module; if not, the onboard business module makes a decision and obtains the result after the decision, and sends the result after the decision to the ground integrated management module for review via telemetry. After the review is passed, the final result is saved to the ground data management module.

[0042] Preferably, the specific steps of S204 are:

[0043] S20401, the ground algorithm training module obtains health data and final results from the ground data management module in a periodic or data-triggered manner to perform intelligent algorithm training;

[0044] S20402, the ground algorithm training module determines whether to enter the third stage through automatic judgment based on the intelligent algorithm training results, and if so, determines whether the accuracy rate obtained by the intelligent algorithm is better than the preset value; if not, a manual judgment is performed;

[0045] S20403, the ground algorithm training module determines whether the accuracy of the intelligent algorithm is better than the preset value. If yes, it enters the third stage, injects the trained intelligent algorithm into the on-board intelligent algorithm management module for intelligent algorithm update, and informs the on-board business module at the same time. The on-board business module enters the third stage and the process ends; if not, repeat S20401 to S20403;

[0046] The user manually determines whether to enter the third stage based on the intelligent algorithm training results. If yes, the third stage is entered, the trained intelligent algorithm is injected into the on-board intelligent algorithm management module for intelligent algorithm update, and the on-board business module is informed at the same time. The on-board business module enters the third stage and the process ends; otherwise, repeat S20401 to S20403.

[0047] Preferably, the specific steps of S3 are:

[0048] S301, the ground data management module obtains health data of the satellite;

[0049] S302, the on-board business module extracts business features based on the health data of the on-board data acquisition module, and respectively calls the rules corresponding to the on-board knowledge management module and the intelligent algorithm corresponding to the on-board intelligent algorithm management module to obtain rule-based results and intelligent algorithm-based results;

[0050] S303, judging the rule-based results and the intelligent algorithm-based results, sending the judged results to the ground integrated management module, and saving the final results to the ground data management module;

[0051] S304, the ground algorithm training module obtains the health data and final results of the ground data management module, and performs intelligent algorithm training. When the intelligent algorithm reaches a certain accuracy rate, the trained intelligent algorithm is injected into the onboard intelligent algorithm management module for intelligent algorithm update.

[0052] Preferably, the specific steps of S303 are:

[0053] The onboard business module determines whether the rule-based results and the intelligent algorithm-based results are consistent. If they are, the consistent results are sent to the ground user for review via telemetry. After the review is passed, the final results are saved to the ground data management module; otherwise, it determines whether manual judgment is required;

[0054] The on-board business module determines whether manual judgment is required. If so, the ground integrated management module performs manual judgment, obtains the final result after manual judgment, saves the final result to the ground data management module, and informs the on-board business module of the final result through remote control; if not, determines whether to default to the result based on the intelligent algorithm;

[0055] The on-board business module determines whether to use the result based on the intelligent algorithm as the default. If so, the result based on the intelligent algorithm is sent to the ground integrated management module for review by telemetry. After the review is passed, the final result is saved to the ground data management module; if not, the on-board business module makes a decision and obtains the result after the decision. The result after the decision is sent to the ground integrated management module for review by telemetry. After the review is passed, the final result is saved to the ground data management module.

[0056] Preferably, the specific steps of S304 are:

[0057] S30401, the ground algorithm training module obtains health data and final results from the ground data management module in a periodic or data-triggered manner to perform intelligent algorithm training;

[0058] S30302, the ground algorithm training module determines whether to update the intelligent algorithm through automatic judgment according to the intelligent algorithm training result, and if so, determines whether the accuracy rate obtained by the intelligent algorithm is better than the preset value; if not, a manual judgment is performed;

[0059] S30303, the ground algorithm training module determines whether the accuracy of the intelligent algorithm is better than the preset value. If yes, the updated intelligent algorithm is sent to the intelligent algorithm management module on the satellite by remote control, and the process ends; if not, S30401 to S30403 are repeated;

[0060] The user manually determines whether the intelligent algorithm needs to be updated based on the intelligent algorithm training results; if so, the updated intelligent algorithm is sent to the intelligent algorithm management module on the satellite by remote control, and the process ends; if not, repeat S30401 to S30403.

[0061] The beneficial effects of the present invention are:

[0062] 1. Real-time data processing.

[0063] On-board health management allows data to be processed directly on the satellite, reducing data transmission delays between the satellite and the ground.

[0064] 2. Improve autonomous control capabilities and reliability.

[0065] Autonomous health management on board the satellite can reduce dependence on ground systems and enable independent operation under special circumstances, thereby enhancing the satellite's autonomous control and decision-making capabilities and improving system reliability.

[0066] 3. Reduce the amount of data transmitted.

[0067] If the autonomous health management system on board has developed to a mature state and its functional performance has reached expectations, then the health management function can be completed autonomously on board, and the ground can be responsible for the auxiliary and inspection tasks. In this case, the satellite can choose not to transmit the original data related to health management, and only transmit the data after feature extraction of the original data; in order to achieve the inspection effect, the satellite can send the original health management data in the telemetry data in a certain period or by sending a remote control command from the ground to request the satellite to upload, so as to check the health management on board. In either case, the amount of telemetry data uploaded and downloaded from the satellite will be greatly reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 Flowchart of the satellite autonomous health management method implemented in stages. DETAILED DESCRIPTION

[0069] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the present invention.

[0070] The ground-based satellite health management system implemented in stages is deployed on-board, enabling satellites to have the ability to manage their own health. However, considering the limited resources on board, the training of the intelligent algorithm is carried out on the ground. When the intelligent algorithm reaches a certain level of accuracy, the intelligent algorithm on board is updated on the ground through version updates. With the development of technology, when resources are no longer a limitation, the training of the intelligent algorithm can also be deployed on board, which further reduces the dependence on the ground.

[0071] The satellite autonomous health management system implemented in stages includes on-board data acquisition module, on-board knowledge management module, on-board intelligent algorithm management module, on-board business module, ground data management module, ground algorithm training module, and ground comprehensive management module, among which:

[0072] The onboard data acquisition module is deployed on the satellite to directly collect the satellite's health data.

[0073] The on-board knowledge management module is deployed on-board to build a detailed knowledge base and provide rules for the on-board business modules.

[0074] The on-board intelligent algorithm management module is deployed on the satellite to provide intelligent algorithms for the on-board business modules.

[0075] The on-board business module is deployed on-board, and business features are extracted based on the health data of the on-board data acquisition module. According to different feature values, the corresponding rules in the on-board knowledge management module are called in the first stage to obtain rule-based results, and the rule-based results are saved in the ground data management module; the second and third stages call the corresponding rules in the on-board knowledge management module to obtain rule-based results, and call the intelligent algorithm in the on-board intelligent algorithm management module to obtain results based on the intelligent algorithm. After judging the rule-based results and the intelligent algorithm-based results, the final results are saved in the ground data management module.

[0076] The ground data management module is deployed on the ground to manage the satellite's health data and results.

[0077] The ground algorithm training module is deployed on the ground, calls the health data and results of the ground data management module to perform intelligent algorithm training, and injects the trained intelligent algorithm into the onboard intelligent algorithm management module.

[0078] The ground integrated management module is deployed on the ground to provide ground control personnel with more comprehensive satellite status analysis and prediction, and to perform manual judgment on scenarios where manual judgment is required by the onboard business modules.

[0079] 1. The onboard data acquisition module is responsible for collecting real-time data from satellite components, extensions, and subsystems, and obtaining the satellite's mission plan, remote control instructions, etc. All of this information serves as the basis for health management.

[0080] 2. The on-board knowledge management module builds a detailed knowledge base (mainly rules) based on satellite design parameters and historical operation data, and defines the thresholds and conditions for normal and abnormal states.

[0081] 3. The on-board intelligent algorithm management module manages the intelligent algorithms used in the health management system and provides algorithm version management and intelligent algorithm operation and maintenance functions.

[0082] 4. The on-board business modules include on-board fault diagnosis module, on-board operation prediction module, on-board health assessment module, and on-board performance analysis module.

[0083] (1) The on-board fault diagnosis module can match and process the data collected by the on-board data acquisition module with the knowledge base of the on-board knowledge management module in real time, and quickly evaluate whether the satellite components, single machines, subsystems, and the entire satellite are faulty.

[0084] The onboard fault diagnosis module has the function of fault diagnosis based on expert knowledge base and expert rules, common fault diagnosis algorithms such as support vector machines, and intelligent algorithms such as Bayesian networks. The fault diagnosis module can also perform corresponding fault handling according to preset fault handling methods if the diagnosed fault matches the corresponding handling method. Of course, the fault diagnosis results and fault handling methods can also be pushed to the operator through telemetry data, and the next step of decision-making can be made manually.

[0085] (2) The on-board operation prediction module can match and process the data collected by the on-board data acquisition module with the knowledge base of the on-board knowledge management module in real time, and perform life and failure prediction on key equipment of the satellite platform and key equipment of the payload.

[0086] The onboard operation prediction module predicts the life and faults of key equipment on the satellite platform and key equipment on the payload. The operation prediction module can also predict fault handling according to the preset fault prediction method. If the predicted fault has reached the condition for fault handling, it can perform predictive fault handling to avoid the occurrence of the fault in advance, and make pre-decommissioning suggestions based on the operation data and prediction results; of course, the fault prediction results, fault handling suggestions, life prediction results, and pre-decommissioning suggestions can also be pushed to the operator through telemetry data, and the next step of decision-making can be made manually.

[0087] (3) The on-board health assessment module can match and process the data collected by the on-board data acquisition module with the knowledge base of the on-board knowledge management module in real time, and quickly assess the health status of each component, single machine, subsystem, and the entire satellite.

[0088] The onboard health assessment module assesses the health status of the satellite and determines whether the satellite is in a healthy, sub-healthy, unhealthy, deteriorating, or pathological state. The health assessment results are pushed to the operator through telemetry data, and the operator makes the next decision manually.

[0089] (4) The on-board performance analysis module can match and process the data collected by the on-board data acquisition module with the knowledge base of the on-board knowledge management module in real time, and quickly evaluate the performance of each component, single machine, subsystem, and the entire satellite.

[0090] The on-board performance analysis module analyzes the performance of the satellite through ADC methods, Bayesian networks and other methods. It can perform performance analysis from various dimensions of the satellite, such as components, single machines, subsystems, and the entire satellite, and push the performance analysis results to the operator through telemetry data, who will make the next decision manually.

[0091] 5. The ground data management module is responsible for storing and managing all key data in the health management system, such as health management related data, satellite mission plans, diagnostic results, knowledge bases, intelligent algorithms, etc.

[0092] 6. The ground algorithm training module has various algorithm training management functions, and completes algorithm training tasks according to existing samples, simulated samples, real satellite operation data, etc., so that the algorithm has practical application capabilities. And according to the algorithm upgrade rules and model algorithm training results, the intelligent algorithm is upgraded.

[0093] Develop machine learning algorithms suitable for satellite health management, such as Bayesian networks, support vector machines, neural networks, etc.

[0094] The artificial intelligence algorithm is trained and optimized using existing samples (relevant design documents, test data of components, single machines, subsystems, desktop joint debugging data, previous operation data of satellites on the same platform, etc.), simulation samples (simulated data combining design documents, test data, previous experience, etc.) and the operation data of the first phase of satellites to improve its ability to recognize and predict changes in satellite status.

[0095] Using the large amount of data accumulated in the first phase and the satellite operation data in the second phase, the artificial intelligence algorithm is trained and optimized to improve its ability to recognize and predict changes in satellite status.

[0096] Continuous learning and optimization mechanism (deployed on the ground): Establish a continuous learning and optimization mechanism so that the system can continuously improve the health management strategy based on new data and experience and adapt to changes in the satellite operating environment.

[0097] 7. The ground integrated management module provides result comparison and analysis functions and auxiliary decision support functions.

[0098] Establish a result comparison and analysis system to compare the evaluation results of the artificial intelligence algorithm with the results of the knowledge base, analyze the consistency and difference between the two, and provide a basis for algorithm optimization.

[0099] Auxiliary decision support: Using artificial intelligence algorithms as auxiliary tools to provide ground control personnel with more comprehensive satellite status analysis and predictions, helping them make more accurate decisions.

[0100] The satellite autonomous health management system, which is implemented in stages, uses intelligent algorithms, knowledge bases, and automated health management processes to achieve real-time monitoring of satellite status, fault diagnosis, predictive maintenance, and decision support to improve the operating efficiency and lifespan of satellites.

[0101] Intelligent algorithm-led: Based on the second phase, further optimize and improve the artificial intelligence algorithm to make it the main tool for satellite health management.

[0102] Knowledge base assistance: The knowledge base is used as an auxiliary tool, and the generated results are used for reference. In addition, it can also be used to handle special situations or boundary conditions that are difficult for intelligent algorithms to identify, ensuring the robustness of the system.

[0103] Automated health management process: fully automates satellite health management, including data collection, fault diagnosis, predictive maintenance, etc., to reduce reliance on ground control personnel.

[0104] like Figure 1 As shown, the satellite autonomous health management method implemented in stages includes the following steps:

[0105] S1, in the first stage, the on-board business module manages the health of the satellite based on rules. In the first stage, the on-board business module manages the health of the satellite based on rules and obtains rule-based results. The ground algorithm training module uses health data and rule-based results to train the intelligent algorithm. When the intelligent algorithm reaches a certain accuracy rate, the system automatically or manually judges until entering the second stage, and the trained intelligent algorithm is injected into the on-board intelligent algorithm management module.

[0106] S101, the ground algorithm training module performs intelligent algorithm training based on existing samples and simulated samples.

[0107] Users input existing samples (relevant design documents, test data of components, single machines, subsystems, desktop joint debugging data, previous operation data of satellites on the same platform, etc.) and simulation samples (simulated data combining design documents, test data, past experience, etc.) into the data management module of the satellite health management system. The intelligent algorithm management module can train the intelligent algorithm based on these sample data before obtaining the actual on-orbit operation data of the satellite.

[0108] Users can modify, add, delete, etc. the knowledge and intelligent algorithms in the ground integrated management module, update the ground data management module and the ground algorithm training module, and update the on-board knowledge management module and the on-board intelligent algorithm management module through remote control.

[0109] S102, the on-board business module extracts business features based on the satellite's health data, calls the corresponding rules of the on-board knowledge management module, obtains rule-based results, and sends the rule-based results to the ground integrated management module.

[0110] (1) The satellite sends telemetry data to the ground data management module.

[0111] After the satellite is in orbit, it will periodically send telemetry data to the satellite measurement and control system, which will forward it to the ground data management module deployed on the ground. At the same time, the ground data management module obtains the satellite's mission plan, remote control instructions, etc. provided by the satellite's related external systems to obtain the satellite's health data.

[0112] (2) The onboard data acquisition module collects the satellite's health data.

[0113] The on-board data acquisition module collects the status data of each component, single machine, and subsystem of the satellite, obtains the satellite's mission plan, remote control instructions, etc., and sends these health data to the on-board business module.

[0114] The on-board business modules include on-board fault diagnosis module, on-board operation prediction module, on-board health assessment module, and on-board performance analysis module.

[0115] (3) The onboard business module receives the health data and extracts business features. Based on different feature values, it calls the corresponding rules in the onboard knowledge management module to obtain rule-based results and processing methods. The rule-based results and processing methods are sent to the ground integrated management module and saved in the ground data management module.

[0116] The onboard fault diagnosis module obtains the fault diagnosis result, the operation prediction module obtains the fault prediction result and the life prediction result, the health assessment module obtains the health assessment result, the performance analysis module obtains the performance analysis result, and performs the next step of processing. The result and the processing method are sent to the ground integrated management module through the satellite measurement and control system for user review. The ground integrated management module performs telemetry analysis, obtains the result and the processing method, and reviews them. After the review is passed, they are saved to the ground data management module.

[0117] S103, the ground algorithm training module obtains the health data and rule-based results of the ground data management module, and continues to train the intelligent algorithm trained in S101. When the intelligent algorithm reaches a certain accuracy rate, the system automatically judges or manually determines until entering the second stage, and injects the trained intelligent algorithm into the onboard intelligent algorithm management module for intelligent algorithm update.

[0118] S10301, the ground algorithm training module obtains health data and rule-based results from the ground data management module in a periodic or data-triggered manner to perform intelligent algorithm training.

[0119] The accuracy rate is continuously improved through continuous intelligent algorithm training.

[0120] S10302, the ground algorithm training module determines whether to enter the second stage through automatic judgment based on the intelligent algorithm training results. If yes, it determines whether the accuracy obtained by the intelligent algorithm is better than the preset value; if not, it makes a manual judgment.

[0121] S10303, the ground algorithm training module determines whether the accuracy of the intelligent algorithm is better than the preset value. If so, it enters the second stage and injects the trained intelligent algorithm into the on-board intelligent algorithm management module for intelligent algorithm update. At the same time, it informs the on-board business module, and the on-board business module enters the second stage and the process ends. Otherwise, repeat S10301 to S10303.

[0122] The user manually determines whether to enter the second stage based on the intelligent algorithm training results; if yes, it enters the second stage, injects the trained intelligent algorithm into the on-board intelligent algorithm management module for intelligent algorithm update, and informs the on-board business module at the same time, the on-board business module enters the second stage, and the process ends. Otherwise, repeat S10301 to S10303.

[0123] S2, in the second stage, the on-board business module manages the health of the satellite based on the rules and the intelligent algorithm trained in S1, obtains the rule-based results and the intelligent algorithm-based results, and judges the rule-based results and the intelligent algorithm-based results in S2 to obtain the final result. The ground algorithm training module uses the health data in S2 and the final result in S2 to train the intelligent algorithm trained in S1. When the intelligent algorithm reaches a certain accuracy rate, the system automatically judges or manually judges until entering the third stage. At the same time, the trained intelligent algorithm is injected into the on-board intelligent algorithm management module.

[0124] Users can modify, add, delete, etc. the knowledge and intelligent algorithms in the ground integrated management module, and update the on-board knowledge management module and on-board intelligent algorithm management module through remote control.

[0125] S201, the ground data management module obtains health data of the satellite.

[0126] After the satellite is in orbit, it will periodically send telemetry data to the satellite measurement and control system. The satellite measurement and control system forwards it to the ground data management module deployed on the ground. At the same time, the ground data management module obtains the satellite's mission plan, remote control instructions, etc. provided by the satellite's related external system to obtain the satellite's health data.

[0127] S202, the on-board business module extracts business features based on the health data of the on-board data acquisition module, and calls the rules corresponding to the on-board knowledge management module and the intelligent algorithm corresponding to the on-board intelligent algorithm management module respectively to obtain rule-based results and intelligent algorithm-based results.

[0128] (1) The onboard data acquisition module collects the satellite's health data.

[0129] The onboard data acquisition module collects the status data of each component, single machine, and subsystem in the satellite, obtains the satellite's mission plan, remote control instructions, etc., and sends these health data to the onboard business module as a basis for health management judgment, and is stored in the database at the same time.

[0130] The on-board business modules include on-board fault diagnosis module, on-board operation prediction module, on-board health assessment module, and on-board performance analysis module.

[0131] (2) The onboard business module receives the health data and extracts business features. Based on different feature values, it calls the corresponding rules in the onboard knowledge management module and the intelligent algorithm in the onboard intelligent algorithm management module to obtain the results and processing methods under two methods.

[0132] The onboard fault diagnosis module receives health data, extracts fault features, and calls the corresponding rules in the onboard knowledge management module according to different feature values ​​to obtain rule-based fault diagnosis results. It then performs data mining and calls the corresponding intelligent algorithm in the onboard intelligent algorithm management module to obtain fault diagnosis results based on the intelligent algorithm.

[0133] The on-board operation prediction module receives the health data, extracts fault features and life features, and calls the corresponding rules in the on-board knowledge management module according to different feature values ​​to obtain rule-based fault prediction results and life prediction results. It performs data mining and calls the corresponding intelligent algorithm in the on-board intelligent algorithm management module to obtain fault prediction results and life prediction results based on the intelligent algorithm.

[0134] The on-board health assessment module receives health data, extracts health features, and calls the corresponding rules in the on-board knowledge management module according to different feature values ​​to obtain rule-based health assessment results, which are then stored in the data management module. Data mining is performed, and the corresponding intelligent algorithms in the on-board intelligent algorithm management module are called to obtain health assessment results based on the intelligent algorithms.

[0135] The on-board performance analysis module receives health data, extracts performance features, and calls the corresponding rules in the on-board knowledge management module according to different feature values ​​to obtain rule-based performance analysis results, and stores the results in the data management module. It performs data mining and calls the corresponding intelligent algorithm in the on-board intelligent algorithm management module to obtain performance analysis results based on the intelligent algorithm.

[0136] S203, judging the rule-based results and the intelligent algorithm-based results, sending the judged results to the ground integrated management module, and saving the final results to the ground data management module.

[0137] The onboard business module determines whether the rule-based results and the intelligent algorithm-based results are consistent. If so, the consistent results are sent to the ground user for review via telemetry. After the review is passed, the final results are saved to the ground data management module; otherwise, it determines whether manual judgment is required.

[0138] The onboard business module determines whether manual judgment is required. If so, the rule-based results, the intelligent algorithm-based results and the extracted business feature data are sent to the ground integrated management module via telemetry for manual judgment. The module manually decides which result is valid and obtains the final result after manual judgment. The final result is saved in the ground data management module and the onboard business module is informed of the final result via remote control. Otherwise, it is determined whether the rule-based result is valid by default.

[0139] The onboard business module determines whether to use the rule-based result as the default. If so, the rule-based result is used as the default. The rule-based result is sent to the ground integrated management module for review via telemetry. After the review is passed, the final result is saved to the ground data management module; otherwise, the onboard business module makes a decision through conflict resolution, voting mechanism and other methods to obtain the result after the decision. The result after the decision is sent to the ground integrated management module for review via telemetry. After the review is passed, the final result is saved to the ground data management module.

[0140] S204, the ground algorithm training module obtains the health data and final results of the ground data management module and performs intelligent algorithm training. When the intelligent algorithm reaches a certain accuracy rate, the system makes autonomous decisions or manual judgments until entering the third stage.

[0141] S20401, the ground algorithm training module obtains health data and final results from the ground data management module in a periodic or data-triggered manner to perform intelligent algorithm training.

[0142] The accuracy rate is continuously improved through continuous intelligent algorithm training.

[0143] S20402, the ground algorithm training module determines whether to enter the third stage through automatic judgment based on the intelligent algorithm training results. If so, it determines whether the accuracy obtained by the intelligent algorithm is better than the preset value; if not, manual judgment is performed.

[0144] S20403, the ground algorithm training module determines whether the accuracy of the intelligent algorithm is better than the preset value. If so, it enters the third stage and injects the trained intelligent algorithm into the on-board intelligent algorithm management module for intelligent algorithm update. At the same time, it informs the on-board business module, and the on-board business module enters the third stage and the process ends; otherwise, repeat S20401 to S20403.

[0145] The user manually determines whether to enter the third stage based on the intelligent algorithm training results. If yes, the third stage is entered, the trained intelligent algorithm is injected into the on-board intelligent algorithm management module for intelligent algorithm update, and the on-board business module is informed at the same time. The on-board business module enters the third stage and the process ends; otherwise, repeat S20401 to S20403.

[0146] S3, the third stage on-board business module manages the health of the satellite based on the rules and the intelligent algorithm trained in S2, obtains the rule-based results and the intelligent algorithm-based results, and judges the rule-based results and the intelligent algorithm-based results in S3 to get the final result. The ground algorithm training module uses the health data in S3 and the final result in S3 to train the intelligent algorithm trained in S2, and at the same time, the trained intelligent algorithm is injected into the on-board intelligent algorithm management module.

[0147] Users can modify, add, delete, etc. the knowledge and intelligent algorithms in the ground integrated management module, and update the on-board knowledge management module and on-board intelligent algorithm management module through remote control.

[0148] S301, the ground data management module obtains the health data of the satellite.

[0149] After the satellite is in orbit, it will periodically send telemetry data to the satellite measurement and control system. The satellite measurement and control system forwards it to the ground data management module deployed on the ground. At the same time, the ground data management module obtains the satellite's mission plan, remote control instructions, etc. provided by the satellite's related external system to obtain the satellite's health data.

[0150] S302, the on-board business module extracts business features based on the health data of the on-board data acquisition module, and calls the rules corresponding to the on-board knowledge management module and the intelligent algorithm corresponding to the on-board intelligent algorithm management module respectively to obtain rule-based results and intelligent algorithm-based results.

[0151] (1) The onboard data acquisition module collects the satellite's health data.

[0152] The onboard data acquisition module collects the status data of each component, single machine, and subsystem in the satellite, obtains the satellite's mission plan, remote control instructions, etc., and sends these health data to the onboard business module as a basis for health management judgment, and is stored in the database at the same time.

[0153] The on-board business modules include on-board fault diagnosis module, on-board operation prediction module, on-board health assessment module, and on-board performance analysis module.

[0154] (2) The onboard business module receives the health data and extracts business features. Based on different feature values, it calls the corresponding rules in the onboard knowledge management module and the intelligent algorithm in the onboard intelligent algorithm management module to obtain the results and processing methods under two methods.

[0155] The onboard fault diagnosis module receives health data, extracts fault features, and calls the corresponding rules in the onboard knowledge management module according to different feature values ​​to obtain rule-based fault diagnosis results. It then performs data mining and calls the corresponding intelligent algorithm in the onboard intelligent algorithm management module to obtain fault diagnosis results based on the intelligent algorithm.

[0156] The on-board operation prediction module receives the health data, extracts fault features and life features, and calls the corresponding rules in the on-board knowledge management module according to different feature values ​​to obtain rule-based fault prediction results and life prediction results. It performs data mining and calls the corresponding intelligent algorithm in the on-board intelligent algorithm management module to obtain fault prediction results and life prediction results based on the intelligent algorithm.

[0157] The on-board health assessment module receives health data, extracts health features, and calls the corresponding rules in the on-board knowledge management module according to different feature values ​​to obtain rule-based health assessment results, which are then stored in the data management module. Data mining is performed, and the corresponding intelligent algorithms in the on-board intelligent algorithm management module are called to obtain health assessment results based on the intelligent algorithms.

[0158] The on-board performance analysis module receives health data, extracts performance features, and calls the corresponding rules in the on-board knowledge management module according to different feature values ​​to obtain rule-based performance analysis results, and stores the results in the data management module. It performs data mining and calls the corresponding intelligent algorithm in the on-board intelligent algorithm management module to obtain performance analysis results based on the intelligent algorithm.

[0159] S303, judging the rule-based results and the intelligent algorithm-based results, sending the judged results to the ground integrated management module, and saving the final results to the ground data management module.

[0160] The onboard business module determines whether the rule-based results and the intelligent algorithm-based results are consistent. If so, the consistent results are sent to the ground user for review via telemetry. After the review is passed, the final results are saved to the ground data management module; otherwise, it determines whether manual judgment is required.

[0161] The onboard business module determines whether human judgment is required. If so, the rule-based results, the intelligent algorithm-based results and the extracted business feature data are sent to the ground integrated management module by telemetry for human judgment. The human decides which result is valid and obtains the final result after human judgment. The final result is saved in the ground data management module and the onboard business module is informed of the final result by remote control. Otherwise, it is determined whether the result based on the intelligent algorithm is valid by default.

[0162] The onboard business module determines whether to use the result based on the intelligent algorithm as the default. If so, the result based on the intelligent algorithm is used as the default. The result based on the intelligent algorithm is sent by telemetry to the ground integrated management module for review. After the review is passed, the final result is saved in the ground data management module; otherwise, the onboard business module makes a decision through conflict resolution, voting mechanism and other methods to obtain the result after the decision. The result after the decision is sent by telemetry to the ground integrated management module for review. After the review is passed, the final result is saved in the ground data management module.

[0163] S304, the ground algorithm training module obtains the health data and final results of the ground data management module, and performs intelligent algorithm training. When the intelligent algorithm reaches a certain accuracy rate, the trained intelligent algorithm is injected into the onboard intelligent algorithm management module for intelligent algorithm update.

[0164] S30401, the ground algorithm training module obtains health data and final results from the ground data management module in a periodic or data-triggered manner to perform intelligent algorithm training.

[0165] The accuracy rate is continuously improved through continuous intelligent algorithm training.

[0166] S30302, the ground algorithm training module determines whether to update the intelligent algorithm through automatic judgment based on the intelligent algorithm training results. If so, it determines whether the accuracy obtained by the intelligent algorithm is better than the preset value; if not, manual judgment is performed.

[0167] S30303, the ground algorithm training module determines whether the accuracy of the intelligent algorithm is better than the preset value. If so, the updated intelligent algorithm is sent to the intelligent algorithm management module on the satellite by remote control, and the process ends; if not, repeat S30401 to S30403.

[0168] The user manually determines whether the intelligent algorithm needs to be updated based on the intelligent algorithm training results; if so, the updated intelligent algorithm is sent to the intelligent algorithm management module on the satellite by remote control, and the process ends; if not, repeat S30401 to S30403.

Claims

1. The satellite autonomous health management system implemented in stages is characterized by: It includes on-board data acquisition module, on-board knowledge management module, on-board intelligent algorithm management module, on-board business module, ground data management module, ground algorithm training module and ground comprehensive management module, among which: The onboard data collection module is deployed onboard the satellite to directly collect the satellite's health data; The on-board knowledge management module is deployed on-board to build a detailed knowledge base and provide rules for the on-board business modules; The on-board intelligent algorithm management module is deployed on the satellite to provide intelligent algorithms for the on-board business modules; The on-board business module is deployed on the satellite, and business features are extracted based on the health data of the on-board data acquisition module. According to different feature values, the corresponding rules in the on-board knowledge management module are called in the first stage to obtain rule-based results, and the rule-based results are saved in the ground data management module; the corresponding rules in the on-board knowledge management module are called in the second and third stages to obtain rule-based results, and the intelligent algorithm in the on-board intelligent algorithm management module is called to obtain the result based on the intelligent algorithm. After judging the rule-based results and the result based on the intelligent algorithm, the final result is saved in the ground data management module; The ground data management module is deployed on the ground to manage the satellite's health data and results; The ground algorithm training module is deployed on the ground, calls the health data and results of the ground data management module, conducts intelligent algorithm training, and injects the trained intelligent algorithm into the onboard intelligent algorithm management module; The ground integrated management module is deployed on the ground to provide ground control personnel with more comprehensive satellite status analysis and prediction, and to perform manual judgment on scenarios where manual judgment is required by the onboard business modules.

2. The satellite autonomous health management method implemented in stages is characterized by: The phased implementation of the satellite autonomous health management system according to claim 1 comprises the following steps: S1, in the first stage, the on-board business module manages the health of the satellite based on rules and obtains rule-based results. The ground algorithm training module uses health data and rule-based results to train the intelligent algorithm. When the intelligent algorithm reaches a certain accuracy, the system automatically or manually determines until entering the second stage. At the same time, the trained intelligent algorithm is injected into the on-board intelligent algorithm management module; S2, in the second stage, the on-board business module manages the health of the satellite based on the rules and the intelligent algorithm trained in S1, obtains the rule-based results and the intelligent algorithm-based results, and judges the rule-based results and the intelligent algorithm-based results in S2 to obtain the final results. The ground algorithm training module uses the health data in S2 and the final results in S2 to train the intelligent algorithm trained in S1. When the intelligent algorithm reaches a certain accuracy rate, the system automatically judges or manually judges until entering the third stage. At the same time, the trained intelligent algorithm is injected into the on-board intelligent algorithm management module; S3, the third stage on-board business module manages the health of the satellite based on the rules and the intelligent algorithm trained in S2, obtains the rule-based results and the intelligent algorithm-based results, and judges the rule-based results and the intelligent algorithm-based results in S3 to get the final result. The ground algorithm training module uses the health data in S3 and the final result in S3 to train the intelligent algorithm trained in S2, and at the same time, the trained intelligent algorithm is injected into the on-board intelligent algorithm management module.

3. The satellite autonomous health management method implemented in stages according to claim 2 is characterized in that: In S1, the ground algorithm training module uses health data and rule-based results to train the intelligent algorithm. When the intelligent algorithm reaches a certain accuracy, the system automatically judges or manually judges until entering the second stage. At the same time, the specific steps of injecting the trained intelligent algorithm into the on-board intelligent algorithm management module are: S10301, the ground algorithm training module obtains health data and rule-based results from the ground data management module in a periodic or data-triggered manner to perform intelligent algorithm training; S10302, the ground algorithm training module determines whether to enter the second stage through automatic judgment based on the intelligent algorithm training results, and if so, determines whether the accuracy obtained by the intelligent algorithm is better than the preset value; if not, performs manual judgment; S10303, the ground algorithm training module determines whether the accuracy of the intelligent algorithm is better than the preset value. If yes, it enters the second stage, injects the trained intelligent algorithm into the on-board intelligent algorithm management module for intelligent algorithm update, and informs the on-board business module at the same time. The on-board business module enters the second stage and the process ends. If no, repeat S10301 to S10303; The user manually determines whether to enter the second stage based on the results of intelligent algorithm training; If yes, it enters the second stage, injects the trained intelligent algorithm into the on-board intelligent algorithm management module for intelligent algorithm update, and informs the on-board business module at the same time. The on-board business module enters the second stage and the process ends. If no, repeat S10301 to S10303.

4. The satellite autonomous health management method implemented in stages according to claim 2 is characterized in that: The specific steps of S2 are: S201, the ground data management module obtains health data of the satellite; S202, the on-board business module extracts business features based on the health data of the on-board data acquisition module, and respectively calls the rules corresponding to the on-board knowledge management module and the intelligent algorithm corresponding to the on-board intelligent algorithm management module to obtain rule-based results and intelligent algorithm-based results; S203, judging the rule-based results and the intelligent algorithm-based results, sending the judged results to the ground integrated management module, and saving the final results to the ground data management module; S204, the ground algorithm training module obtains the health data and final results of the ground data management module and performs intelligent algorithm training. When the intelligent algorithm reaches a certain accuracy rate, the system makes autonomous decisions or manual judgments until entering the third stage.

5. The satellite autonomous health management method implemented in stages according to claim 4 is characterized in that: The specific steps of S203 are: The onboard business module determines whether the rule-based results and the intelligent algorithm-based results are consistent. If so, the consistent results are sent to the ground user for review via telemetry. After the review is passed, the final results are saved to the ground data management module; If otherwise, whether human judgment is required; The on-board business module determines whether manual judgment is required. If so, the ground integrated management module performs manual judgment, obtains the final result after manual judgment, saves the final result to the ground data management module, and informs the on-board business module of the final result through remote control; If otherwise, the judgment is whether to default to the rule-based result; The on-board business module determines whether the rule-based result is used by default. If so, the rule-based result is sent to the ground integrated management module for review by telemetry. After the review is passed, the final result is saved to the ground data management module; Otherwise, the on-board business module makes a decision and obtains the result, which is then sent to the ground integrated management module for review via telemetry. After the review is passed, the final result is saved in the ground data management module.

6. The satellite autonomous health management method implemented in stages according to claim 4 is characterized in that: The specific steps of S204 are: S20401, the ground algorithm training module obtains health data and final results from the ground data management module in a periodic or data-triggered manner to perform intelligent algorithm training; S20402, the ground algorithm training module determines whether to enter the third stage through automatic judgment based on the intelligent algorithm training results, and if so, determines whether the accuracy rate obtained by the intelligent algorithm is better than the preset value; if not, a manual judgment is performed; S20403, the ground algorithm training module determines whether the accuracy of the intelligent algorithm is better than the preset value. If yes, it enters the third stage, injects the trained intelligent algorithm into the on-board intelligent algorithm management module for intelligent algorithm update, and informs the on-board business module at the same time. The on-board business module enters the third stage and the process ends; if not, repeat S20401 to S20403; The user manually determines whether to enter the third stage based on the intelligent algorithm training results. If yes, the third stage is entered, the trained intelligent algorithm is injected into the on-board intelligent algorithm management module for intelligent algorithm update, and the on-board business module is informed at the same time. The on-board business module enters the third stage and the process ends; otherwise, repeat S20401 to S20403.

7. The satellite autonomous health management method implemented in stages according to claim 2 is characterized in that: The specific steps of S3 are: S301, the ground data management module obtains health data of the satellite; S302, the on-board business module extracts business features based on the health data of the on-board data acquisition module, and respectively calls the rules corresponding to the on-board knowledge management module and the intelligent algorithm corresponding to the on-board intelligent algorithm management module to obtain rule-based results and intelligent algorithm-based results; S303, judging the rule-based results and the intelligent algorithm-based results, sending the judged results to the ground integrated management module, and saving the final results to the ground data management module; S304, the ground algorithm training module obtains the health data and final results of the ground data management module, and performs intelligent algorithm training. When the intelligent algorithm reaches a certain accuracy rate, the trained intelligent algorithm is injected into the onboard intelligent algorithm management module for intelligent algorithm update.

8. The satellite autonomous health management method implemented in stages according to claim 7 is characterized in that: The specific steps of S303 are: The onboard business module determines whether the rule-based results and the intelligent algorithm-based results are consistent. If so, the consistent results are sent to the ground user for review via telemetry. After the review is passed, the final results are saved to the ground data management module; If otherwise, whether human judgment is required; The on-board business module determines whether manual judgment is required. If so, the ground integrated management module performs manual judgment, obtains the final result after manual judgment, saves the final result to the ground data management module, and informs the on-board business module of the final result through remote control; If otherwise, the judgment is whether to default to the result based on the intelligent algorithm; The on-board business module determines whether the result based on the intelligent algorithm is used by default. If so, the result based on the intelligent algorithm is sent to the ground integrated management module for review by telemetry. After the review is passed, the final result is saved to the ground data management module; Otherwise, the on-board business module makes a decision and obtains the result, which is then sent to the ground integrated management module for review via telemetry. After the review is passed, the final result is saved in the ground data management module.

9. The satellite autonomous health management method implemented in stages according to claim 7, characterized in that: The specific steps of S304 are: S30401, the ground algorithm training module obtains health data and final results from the ground data management module in a periodic or data-triggered manner to perform intelligent algorithm training; S30302, the ground algorithm training module determines whether to update the intelligent algorithm through automatic judgment according to the intelligent algorithm training result, and if so, determines whether the accuracy rate obtained by the intelligent algorithm is better than the preset value; if not, a manual judgment is performed; S30303, the ground algorithm training module determines whether the accuracy of the intelligent algorithm is better than the preset value. If so, the updated intelligent algorithm is sent to the intelligent algorithm management module on the satellite by remote control, and the process ends; If otherwise, repeat S30401 to S30403; The user manually determines whether the intelligent algorithm needs to be updated based on the intelligent algorithm training results; if so, the updated intelligent algorithm is sent to the intelligent algorithm management module on the satellite by remote control, and the process ends; If otherwise, repeat S30401 to S30403.

Images