A method for evaluating the health of spacecraft on-orbit

Abstract

The present invention relates to the technical field of spacecraft health assessment, and in particular to a spacecraft on-orbit health assessment method, comprising step S1, collecting real-time operation data of the spacecraft; step S2, predicting pre-data according to the real-time operation data of the spacecraft; step S3, making digital twin predictions of the on-orbit pre-status of the spacecraft; step S4, monitoring the real-time on-orbit state of the spacecraft; step S5, adjusting the edge computing mode of the spacecraft; step S6, collecting and fusing multimodal data, and evaluating the on-orbit health of the spacecraft; step S7, making autonomous decisions based on the on-orbit health assessment results of the spacecraft, and repairing the spacecraft; step S8, calculating the error rate of the on-orbit health assessment results of the spacecraft, and adjusting the pre-data prediction process; step S9, optimizing the adjustment process of the pre-data prediction process. The present invention improves the accuracy of the on-orbit health assessment of the spacecraft.

Description

Technical Field

[0001] The present invention relates to the technical field of spacecraft health assessment, and in particular to a spacecraft on-orbit health assessment method. Background Art

[0002] The on-orbit health assessment of spacecraft is crucial to ensure its long-term stable operation. However, this assessment is currently inaccurate. Due to the relatively small amount of failure data, it is difficult to build a comprehensive and accurate prediction model. The performance and reliability of the system often rely on the experience and qualitative descriptions of experts, which may be subjective and ambiguous, increasing the uncertainty of the assessment. In addition, the operating status of the spacecraft cannot be simply divided into "normal" and "faulty". In view of these challenges, it is particularly important to develop a more accurate on-orbit health assessment system, which can not only improve the operating efficiency of the spacecraft and reduce the risk of unexpected failures, but also effectively extend the service life of the spacecraft and ensure the successful completion of the space mission.

[0003] Chinese patent publication number: CN103425874A discloses a spacecraft health assessment method based on rate modulus reliability theory, which has seven steps: Step 1: Spacecraft operation state space division; Step 2: Design spacecraft state transition probability matrix calculation method; Step 3: Design spacecraft steady-state probability calculation method; Step 4: Design spacecraft rate modulus reliability estimation method; Step 5: Design spacecraft health level determination method; Step 6: Design spacecraft average fuzzy failure time calculation method; Step 7: Enter the design end stage. However, the accuracy of spacecraft health assessment of this scheme is still insufficient, and the spacecraft is not repaired according to the spacecraft health assessment results. Summary of the invention

[0004] To this end, the present invention provides a method for assessing the on-orbit health of a spacecraft, so as to overcome the problems in the prior art that there is no pre-prediction of the on-orbit health of the spacecraft, no digital twin pre-production and multimodal data fusion processing, and the on-orbit health status of the spacecraft cannot be predicted, and the accuracy of the on-orbit health assessment of the spacecraft is low. Edge computing backup prevention and autonomous decision-making are not performed based on the results of the on-orbit health assessment of the spacecraft, and the on-orbit failure of the spacecraft cannot be resolved in time, resulting in poor effect of the on-orbit health assessment of the spacecraft.

[0005] To achieve the above object, the present invention provides a method for evaluating the health of a spacecraft on-orbit operation, comprising:

[0006] Step S1, collecting real-time operation data of the spacecraft;

[0007] Step S2, predicting the pre-set data according to the real-time operation data of the spacecraft;

[0008] Step S3, performing digital twin prediction of the spacecraft's on-orbit pre-positioned state based on the pre-positioned data;

[0009] Step S4, monitoring the real-time status of the spacecraft on orbit according to the real-time operation data of the spacecraft;

[0010] Step S5, adjusting the spacecraft edge computing method according to the spacecraft on-orbit pre-conditioning state and the spacecraft on-orbit real-time state;

[0011] Step S6, collecting and fusing multimodal data according to the pre-orbit state and real-time state of the spacecraft on orbit, and evaluating the on-orbit health of the spacecraft to obtain the on-orbit health evaluation result of the spacecraft;

[0012] Step S7, making an autonomous decision based on the on-orbit health assessment result of the spacecraft, and repairing the spacecraft based on the autonomous decision result;

[0013] Step S8, calculating the error rate of the spacecraft on-orbit health assessment result, and adjusting the pre-data prediction process according to the error rate of the spacecraft on-orbit health assessment result;

[0014] Step S9, optimizing the adjustment process of the preceding data prediction process according to the preceding data prediction duration.

[0015] Furthermore, in step S3, when performing digital twin prediction on the on-orbit pre-state of the spacecraft, the pre-state data is input into the twin prediction model, and the twin prediction model outputs the on-orbit pre-state of the spacecraft. When the twin prediction model outputs that the on-orbit pre-state of the spacecraft is an abnormal pre-state, the twin prediction model marks the time point when the spacecraft has an abnormal on-orbit operation state within the pre-time as an abnormal pre-time point.

[0016] Furthermore, in step S4, when monitoring the real-time state of the spacecraft on orbit, the real-time operation data of the spacecraft is compared with the preset spacecraft standard operation data threshold, and the real-time state of the spacecraft on orbit is monitored according to the comparison result, wherein:

[0017] When all the real-time operation data of the spacecraft fall within the preset spacecraft standard operation data threshold, it is determined that the real-time status of the spacecraft on orbit is normal;

[0018] When there is real-time spacecraft operation data that falls outside the preset spacecraft standard operation data threshold, the real-time state of the spacecraft on orbit is judged to be abnormal, and the real-time spacecraft operation data that falls outside the preset spacecraft standard operation data threshold is used as key abnormal data.

[0019] Furthermore, in step S5, when the spacecraft edge computing mode is adjusted, the adjustment of the spacecraft edge computing mode is judged according to the spacecraft on-orbit pre-conditioning state and the spacecraft on-orbit real-time state, wherein:

[0020] When the spacecraft on-orbit pre-conditioning state is an abnormal pre-conditioning state and the spacecraft on-orbit real-time state is normal, it is determined that the priority of the data computing task of the spacecraft edge computing method is adjusted;

[0021] When the on-orbit pre-conditioning state of the spacecraft is an abnormal pre-conditioning state, and the on-orbit real-time state of the spacecraft is abnormal, it is determined that the priority of the data computing tasks of the spacecraft edge computing method is adjusted;

[0022] When the spacecraft on-orbit pre-conditioning state is a normal pre-conditioning state and the spacecraft on-orbit real-time state is normal, it is determined that the priority of the data computing task of the spacecraft edge computing mode is not adjusted;

[0023] When the on-orbit pre-conditioning state of the spacecraft is a normal pre-conditioning state and the on-orbit real-time state of the spacecraft is abnormal, it is determined to adjust the priority of the data computing tasks of the spacecraft edge computing method.

[0024] Furthermore, in step S6, when evaluating the on-orbit health of the spacecraft, the collection of multimodal data is judged according to the pre-orbit state of the spacecraft and the real-time state of the spacecraft, wherein:

[0025] When the on-orbit pre-conditioning state of the spacecraft is an abnormal pre-conditioning state, and the on-orbit real-time state of the spacecraft is normal, it is determined to collect multimodal data, and the collected multimodal data is integrated with the real-time operation data and pre-conditioning data of the spacecraft to obtain evaluation data;

[0026] When the on-orbit pre-conditioning state of the spacecraft is an abnormal pre-conditioning state, and the on-orbit real-time state of the spacecraft is abnormal, it is determined to collect multimodal data, and the collected multimodal data is fused with the real-time operation data and pre-conditioning data of the spacecraft to obtain evaluation data;

[0027] When the on-orbit pre-conditioning state of the spacecraft is a normal pre-conditioning state, and the on-orbit real-time state of the spacecraft is normal, it is determined that the multimodal data will not be collected, and the real-time operation data of the spacecraft and the pre-conditioning data are fused to obtain evaluation data;

[0028] When the on-orbit pre-conditioning state of the spacecraft is a normal pre-conditioning state, and the on-orbit real-time state of the spacecraft is abnormal, it is determined to collect multimodal data, and the collected multimodal data is integrated with the real-time operation data and pre-conditioning data of the spacecraft to obtain evaluation data;

[0029] The evaluation data is input into each modal spacecraft evaluation model, and the evaluation coefficients output by each modal spacecraft evaluation model are obtained. Each evaluation coefficient is input into the neural network model, and the spacecraft in-orbit health assessment result is output through the neural network model.

[0030] Furthermore, in step S7, when making autonomous decisions, autonomous decision-making judgments are made according to the on-orbit health assessment results of the spacecraft, wherein:

[0031] When the result of the on-orbit health assessment of the spacecraft is that the spacecraft is healthy on-orbit, no autonomous decision is made;

[0032] When the result of the on-orbit health assessment of the spacecraft is that the on-orbit operation of the spacecraft is unhealthy, an autonomous decision is made, the assessment data is input into the autonomous decision-making model, and the autonomous decision result is output.

[0033] Furthermore, in step S7, when repairing the spacecraft, the spacecraft repair method is determined according to the autonomous decision result, wherein:

[0034] When the autonomous decision result is a repair execution instruction, it is determined that the execution instruction will be sent to the aircraft repair controller, and the aircraft repair controller executes the repair execution instruction;

[0035] When the autonomous decision result is a blank execution instruction, it is determined that the aircraft repair controller executes the fuzzy control algorithm.

[0036] Further, in step S8, when calculating the error rate of the spacecraft on-orbit health assessment result, the number of errors in the spacecraft on-orbit health assessment result N1, the total number of spacecraft on-orbit health assessments N10, and the number of unhealthy and unevaluated spacecraft on-orbit operations N2 are obtained, and the error rate r of the spacecraft on-orbit health assessment result is calculated, and r=a×N1 / N10+b×N2 / N20 is set, where N20 is the preset number of unevaluated times, a is the wrong evaluation weight parameter, b is the missed evaluation weight parameter, a+b=1, and the values ​​of a and b are set according to the values ​​of N1 / N10 and N2 / N20, wherein:

[0037] When N1 / N10<0.5 and N2 / N20<1.7, set a=0.5, b=0.5;

[0038] When N1 / N10≥0.5 and N2 / N20<1.7, set a=0.7, b=0.3;

[0039] When N1 / N10<0.5 and N2 / N20≥1.7, set a=0.3, b=0.7;

[0040] When N1 / N10≥0.5 and N2 / N20<1.7, set a=0.5, b=0.5.

[0041] Furthermore, in step S8, when the pre-data prediction process is adjusted, the error rate r of the spacecraft on-orbit health assessment result is compared with the preset error rate r0 of the spacecraft on-orbit health assessment result, and the adjustment of the pre-data prediction process is judged according to the comparison result, wherein:

[0042] When r<r0, it is determined that no adjustment is made to the pre-data prediction process;

[0043] When r≥r0, it is determined that the pre-data prediction process is adjusted and the pre-adjustment coefficient u is set, u=1+e -(r0-r) , adjust the lead time T according to the lead adjustment coefficient u, the adjusted lead time is TU, and set TU=u×T.

[0044] Furthermore, in step S9, the pre-data prediction duration Th is obtained, and the pre-data prediction duration Th is compared with the preset pre-data prediction duration Th0, and the optimization of the adjustment process of the pre-data prediction process is judged according to the comparison result, wherein:

[0045] When Th≤Th0, it is determined that the adjustment process of the pre-data prediction process is not optimized;

[0046] When Th>Th0, it is determined that the adjustment process of the pre-data prediction process is optimized, and the optimization coefficient h is set, 0.75<h<0.8, the preferred value is h=0.78, and the pre-adjustment coefficient u is optimized according to the optimization coefficient h. The optimized pre-adjustment coefficient is uh, and uh=h×u is set. When uh≤1, the value of uh is 1.

[0047] Compared with the prior art, the beneficial effects of the present invention are that, the method collects the real-time operation data of the spacecraft through step S1, ensures that all key parameters in the spacecraft operation process are obtained, and provides a comprehensive and accurate data basis for subsequent analysis; the method predicts the pre-data according to the real-time operation data of the spacecraft through step S2, and predicts the operation state of the spacecraft in a future period of time based on the real-time data, thereby improving the reliability of the prediction; the method performs digital twin prediction of the pre-state of the spacecraft on orbit according to the pre-data through step S3, and simulates the pre-state of the spacecraft in a virtual environment through digital twin technology to help understand the behavior of complex systems; the method monitors the real-time state of the spacecraft on orbit according to the real-time operation data of the spacecraft through step S4, analyzes and displays the operation state of the spacecraft in real time, and ensures safe and stable operation; the method adjusts the edge computing mode of the spacecraft according to the pre-state of the spacecraft on orbit and the real-time state of the spacecraft on orbit through step S5, and dynamically adjusts the edge computing strategy to better adapt to the operation requirements of the spacecraft, optimize the allocation of computing resources, and improve the processing efficiency; the method performs multi-mode prediction according to the pre-state of the spacecraft on orbit and the real-time state of the spacecraft on orbit through step S6 The method collects and integrates the multimodal data, evaluates the on-orbit health of the spacecraft, obtains the on-orbit health evaluation result of the spacecraft, combines the multimodal data, comprehensively evaluates the on-orbit health status of the spacecraft, improves the accuracy and reliability of the evaluation, and provides strong support for decision-making. The method makes an autonomous decision according to the on-orbit health evaluation result of the spacecraft through step S7, and repairs the spacecraft according to the autonomous decision result. Based on the evaluation result, a repair plan is formulated and executed to ensure the continuous and stable operation of the spacecraft, realize autonomous decision-making and repair of the spacecraft, reduce human intervention, and improve the efficiency and accuracy of repair. To ensure that the spacecraft resumes normal operation as soon as possible, the method calculates the error rate of the spacecraft on-orbit health assessment result through step S8, and adjusts the pre-data prediction process according to the error rate of the spacecraft on-orbit health assessment result, and continuously optimizes the pre-data prediction process through feedback of the assessment result error rate, thereby improving the prediction accuracy and the overall performance of the system. The method optimizes the adjustment process of the pre-data prediction process according to the pre-data prediction time through step S9, adjusts the prediction time according to actual needs, improves the prediction efficiency and practicality, and makes the prediction process more flexible and adaptable to different scenarios and needs.

[0048] In particular, in step S6, when either the pre-status or the real-time status of the spacecraft is abnormal, the system determines that multimodal data needs to be collected to ensure that more comprehensive data can be obtained for in-depth evaluation when there are potential problems or actual problems. When both the pre-status and real-time status of the spacecraft are normal, the system determines that additional multimodal data does not need to be collected, and only real-time operation data and pre-status data are used for evaluation, thereby reducing unnecessary data collection and processing work and improving work efficiency.

[0049] In particular, in step S8, key data are first collected, including the number of evaluation result errors, the total number of evaluations, the number of unhealthy unassessed times, and the preset number of unassessed times. Based on these data, a comprehensive error rate r is calculated, which not only takes into account the wrong evaluation, but also takes into account the necessary evaluation that has not been performed. The weighting is performed by the wrong evaluation weight parameter a and the missed evaluation weight parameter b. The weight parameters a and b are dynamically adjusted according to the severity of the current error situation to ensure the sensitivity of the system to different error types, and the calculated error rate r is compared with the preset error rate threshold r0. If r<r0, it means that The accuracy of the current evaluation system is within an acceptable range, so the lead data prediction process is not adjusted. If r ≥ r0, it means that the accuracy of the evaluation system is insufficient, and the lead data prediction process needs to be adjusted to improve its accuracy. When adjustment is needed, the system calculates a lead adjustment coefficient u, which is inversely proportional to the difference between the error rate r and the preset error rate r0, that is, the higher the error rate, the greater the adjustment range. The lead time T is adjusted using the lead adjustment coefficient u to obtain the adjusted lead time TU, which helps the system predict the status of the spacecraft more accurately in the future, thereby discovering and solving problems in advance.

[0050] In particular, in the step S9, the current leading data prediction duration Th is obtained and compared with the preset leading data prediction duration Th0. Through this comparison, the system can evaluate whether the current prediction duration is too long or too short, so as to determine whether the adjustment process needs to be further optimized. When Th≤Th0, it means that the current leading data prediction duration is within a reasonable range and no further optimization of the adjustment process is required. When Th>Th0, it means that the prediction duration is too long, which may lead to waste of resources or low prediction efficiency. Therefore, the adjustment process needs to be optimized. At this time, the system sets an optimization coefficient h, where 0.75<h<0.8, and the preferred value is h=0.78, which is used to adjust the leading adjustment coefficient u. The optimization coefficient h is used to optimize the leading adjustment coefficient u to obtain the optimized leading adjustment coefficient uh, where uh=h×u. If the optimized coefficient uh≤1, the value of uh is directly taken as 1 to avoid excessive adjustment leading to too short a leading time. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 Schematic diagram of the flow chart of the spacecraft on-orbit health assessment method of this embodiment. DETAILED DESCRIPTION

[0052] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0053] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0054] See also Figure 1 As shown, it is a flow chart of the spacecraft on-orbit health assessment method of this embodiment, and the method includes:

[0055] Step S1, collecting real-time operation data of the spacecraft;

[0056] Step S2, predicting the pre-set data according to the real-time operation data of the spacecraft;

[0057] Step S3, performing digital twin prediction of the spacecraft's on-orbit pre-positioned state based on the pre-positioned data;

[0058] Step S4, monitoring the real-time status of the spacecraft on orbit according to the real-time operation data of the spacecraft;

[0059] Step S5, adjusting the spacecraft edge computing method according to the spacecraft on-orbit pre-conditioning state and the spacecraft on-orbit real-time state;

[0060] Step S6, collecting and fusing multimodal data according to the pre-orbit state and real-time state of the spacecraft on orbit, and evaluating the on-orbit health of the spacecraft to obtain the on-orbit health evaluation result of the spacecraft;

[0061] Step S7, making an autonomous decision based on the on-orbit health assessment result of the spacecraft, and repairing the spacecraft based on the autonomous decision result;

[0062] Step S8, calculating the error rate of the spacecraft on-orbit health assessment result, and adjusting the pre-data prediction process according to the error rate of the spacecraft on-orbit health assessment result;

[0063] Step S9, optimizing the adjustment process of the preceding data prediction process according to the preceding data prediction duration.

[0064] Specifically, the method is applied to the field of spacecraft health assessment technology. The method collects the real-time operation data of the spacecraft through step S1 to ensure that all key parameters in the spacecraft operation process are obtained, providing a comprehensive and accurate data basis for subsequent analysis. The method predicts the pre-data according to the real-time operation data of the spacecraft through step S2, and predicts the operation status of the spacecraft in a future period of time based on the real-time data to improve the reliability of the prediction. The method performs digital twin prediction of the spacecraft on-orbit pre-state according to the pre-data through step S3, and simulates the pre-state of the spacecraft in a virtual environment through digital twin technology to help understand the behavior of complex systems. The method monitors the real-time state of the spacecraft on-orbit according to the real-time operation data of the spacecraft through step S4, analyzes and displays the operation status of the spacecraft in real time to ensure safe and stable operation. The method adjusts the spacecraft edge computing mode according to the pre-state of the spacecraft on-orbit and the real-time state of the spacecraft on-orbit through step S5, and dynamically adjusts the edge computing strategy to better adapt to the operation needs of the spacecraft, optimize the allocation of computing resources, and improve the processing efficiency. The method performs digital twin prediction of the spacecraft on-orbit pre-state according to the pre-state of the spacecraft on-orbit and the real-time state of the spacecraft on-orbit through step S6 The multimodal data are collected and integrated, and the on-orbit health of the spacecraft is evaluated to obtain the on-orbit health evaluation result of the spacecraft. The on-orbit health status of the spacecraft is comprehensively evaluated by combining the multimodal data, the accuracy and reliability of the evaluation are improved, and strong support is provided for decision-making. The method makes an autonomous decision according to the on-orbit health evaluation result of the spacecraft through step S7, and repairs the spacecraft according to the autonomous decision result. Based on the evaluation result, a repair plan is formulated and executed to ensure the continuous and stable operation of the spacecraft, realize autonomous decision-making and repair of the spacecraft, reduce human intervention, and improve the efficiency and accuracy of repair. The method calculates the error rate of the spacecraft on-orbit health assessment result through step S8, and adjusts the pre-data prediction process according to the error rate of the spacecraft on-orbit health assessment result. The pre-data prediction process is continuously optimized through feedback of the assessment result error rate, and the prediction accuracy and overall system performance are improved. The method optimizes the adjustment process of the pre-data prediction process according to the pre-data prediction time through step S9, adjusts the prediction time according to actual needs, improves prediction efficiency and practicality, and makes the prediction process more flexible and adaptable to different scenarios and needs.

[0065] Specifically, in step S1, when the real-time operation data of the spacecraft is collected, the real-time operation data of the spacecraft includes the spacecraft operation temperature, the output of the aircraft battery, the pressure difference between the inside and outside of the aircraft cabin, the spacecraft operation acceleration, the vibration frequency of the mechanical equipment inside the aircraft, the aircraft communication signal strength, the aircraft position, the aircraft operation orbit, the energy output of the solar panel, the energy consumption of the aircraft, the cosmic radiation, the impact force of micrometeoroids and the impact force of space debris. The spacecraft operation temperature refers to the operation temperature of the key parts inside the aircraft, such as the temperature of the controller and the radiator of the aircraft. The aircraft battery output refers to the power consumption of the aircraft battery. The pressure difference between the inside and outside of the aircraft cabin refers to the pressure difference between the inside and outside of the aircraft cabin. the pressure difference, the spacecraft operation acceleration refers to the value of the acceleration of the spacecraft when it is controlled to operate, the vibration frequency of the aircraft's internal mechanical equipment refers to the frequency of vibration of the aircraft's internal mechanical equipment due to operation or the external environment, the aircraft communication signal strength refers to the signal strength of the aircraft's communication with the outside, the outside refers to an external system that transmits information to the aircraft, such as a ground station, the aircraft position refers to the three-dimensional coordinates of the aircraft's position, the aircraft operation orbit refers to the current space orbit position of the aircraft, the solar panel energy output refers to the energy consumed by the aircraft's solar panels to the aircraft's operation, the aircraft energy consumption refers to the total energy consumption consumed by the aircraft's operation, and the space The cosmic radiation refers to the radiation of the cosmic environment in which the aircraft is located. The micrometeoroid impact force refers to the pressure on the aircraft when it is hit by a micrometeoroid. The space debris impact force refers to the pressure on the aircraft when it is hit by space debris. The step S1 collects the spacecraft operating temperature through a temperature sensor. This embodiment does not limit the collection position of the spacecraft operating temperature. Those skilled in the art can freely set it according to actual conditions, as long as the monitoring requirements of the temperature at key positions of the spacecraft are met. For example, the temperature sensor can be set at the spacecraft operation controller to collect the temperature of the spacecraft operation controller. The step S1 collects the aircraft battery output through a voltage and current sensor. The step S1 collects the pressure difference between the inside and outside of the aircraft cabin through the pressure sensor, the step S1 collects the acceleration of the spacecraft through the accelerometer, the step S1 collects the vibration frequency of the mechanical equipment inside the aircraft through the vibration sensor, the step S1 collects the aircraft communication signal strength through the signal demodulator, the step S1 collects the aircraft position and the aircraft orbit through the GPS receiver, the step S1 collects the solar panel energy output and the aircraft energy consumption through the energy monitoring sensor, the step S1 collects the cosmic radiation through the radiation monitor, and the step S1 collects the micrometeoroid impact force and the space debris impact force through the impact sensor.

[0066] Specifically, in step S2, when predicting the preceding data, the historical operation data of the spacecraft is obtained, and a preceding data prediction model is constructed based on the historical operation data of the spacecraft, the real-time operation data of the spacecraft is input into the preceding data prediction model, and the preceding data prediction model outputs the preceding data within the preceding time length.

[0067] Specifically, the spacecraft historical operation data refers to various real-time operation data of the spacecraft collected during its entire operation cycle. The pre-data prediction model refers to a convolutional neural network model that is trained based on the spacecraft historical operation data and meets the prediction accuracy. This embodiment does not limit the training method of the convolutional neural network model. Relevant technical personnel in this field can freely set it according to actual needs, and only need to meet the need for predicting the pre-data. For example, the training method of the convolutional neural network model can be set to divide the spacecraft historical operation data into 75% pre-training set and 25% pre-validation set. The training set is input into the convolutional neural network model to train the convolutional neural network model, and the pre-validation set is input into the trained convolutional neural network model to iteratively optimize the parameters of the convolutional neural network model until the prediction accuracy of the convolutional neural network model reaches 98%, and the convolutional neural network model is output as the pre-data prediction model. The convolutional neural network model refers to a neural network model based on deep learning, and the lead time refers to the time interval from the current time node of the spacecraft to the preset time node. For example, if the current time node is 12:00 and the preset time node is 7:00, the lead time is 5 hours.

[0068] Specifically, in step S3, when performing digital twin prediction on the on-orbit pre-state of the spacecraft, the pre-state data is input into the twin prediction model, and the twin prediction model outputs the on-orbit pre-state of the spacecraft. When the twin prediction model outputs that the on-orbit pre-state of the spacecraft is an abnormal pre-state, the twin prediction model marks the time point when the spacecraft has an abnormal on-orbit operation state within the pre-time as an abnormal pre-time point.

[0069] Specifically, the spacecraft on-orbit pre-state includes a normal pre-state and an abnormal pre-state. The normal pre-state means that the spacecraft is in a normal on-orbit operation state within the pre-time, and the abnormal pre-state means that the spacecraft is in an abnormal on-orbit operation state within the pre-time. The twin prediction model refers to a linear regression model that is trained based on historical pre-data and meets the twin accuracy. This embodiment does not limit the training method of the linear regression model. Relevant technical personnel in this field can freely set it according to actual needs, and only need to meet the need to predict the on-orbit pre-state of the spacecraft. For example, the training method of the linear regression model can be set to divide the historical pre-data into It is divided into 75% twin training set and 25% twin validation set. The twin training set is input into the linear regression model to train the linear regression model, and the twin validation set is input into the trained linear regression model to gradually optimize the parameters of the linear regression model until the prediction accuracy of the linear regression model reaches 95%. The linear regression model is output as a twin prediction model. The linear regression model refers to a model for analyzing and modeling the linear relationship between one or more independent variables and dependent variables. The historical prerequisite data refers to the prerequisite data obtained during the training cycle of the twin prediction model. For example, if the training cycle of the twin prediction model is 2 months, the prerequisite data within two months is obtained.

[0070] Specifically, in step S4, when monitoring the real-time state of the spacecraft on orbit, the real-time operation data of the spacecraft is compared with the preset spacecraft standard operation data threshold, and the real-time state of the spacecraft on orbit is monitored according to the comparison result, wherein:

[0071] When all the real-time operation data of the spacecraft fall within the preset spacecraft standard operation data threshold, it is determined that the real-time status of the spacecraft on orbit is normal;

[0072] When there is real-time spacecraft operation data that falls outside the preset spacecraft standard operation data threshold, the real-time state of the spacecraft on orbit is judged to be abnormal, and the real-time spacecraft operation data that falls outside the preset spacecraft standard operation data threshold is used as key abnormal data.

[0073] Specifically, the preset spacecraft standard operating data threshold refers to the preset spacecraft standard operating data threshold for monitoring the real-time status of the spacecraft on orbit. The present embodiment does not limit the value range of the preset spacecraft standard operating data threshold. Relevant technical personnel in the field can freely set it according to actual needs, and it is only necessary to meet the needs of monitoring the real-time status of the spacecraft on orbit. For example, the preset spacecraft standard operating data threshold can be set to a spacecraft operating temperature of -55°C to +125°C, an aircraft battery output of 11.1V to 12.6V, a pressure difference between the inside and outside of the aircraft cabin of 24.5kPa to 78.4kPa, a spacecraft operating acceleration of 0 to 4g, and a vibration frequency of mechanical equipment inside the aircraft of 1mHz to 10Hz.

[0074] Specifically, in step S5, when the spacecraft edge computing mode is adjusted, the adjustment of the spacecraft edge computing mode is judged according to the spacecraft on-orbit pre-conditioning state and the spacecraft on-orbit real-time state, wherein:

[0075] When the on-orbit pre-conditioning state of the spacecraft is an abnormal pre-conditioning state and the on-orbit real-time state of the spacecraft is normal, it is determined that the priority of the data computing tasks of the spacecraft edge computing method is adjusted;

[0076] When the on-orbit pre-conditioning state of the spacecraft is an abnormal pre-conditioning state and the on-orbit real-time state of the spacecraft is abnormal, it is determined that the priority of the data computing tasks of the spacecraft edge computing method is adjusted;

[0077] When the spacecraft on-orbit pre-conditioning state is a normal pre-conditioning state and the spacecraft on-orbit real-time state is normal, it is determined that the priority of the data computing task of the spacecraft edge computing mode is not adjusted;

[0078] When the on-orbit pre-conditioning state of the spacecraft is a normal pre-conditioning state and the on-orbit real-time state of the spacecraft is abnormal, it is determined to adjust the priority of the data computing tasks of the spacecraft edge computing method.

[0079] Specifically, the data computing tasks of the spacecraft edge computing method include data backup, real-time data computing and edge node data computing. The present embodiment does not limit the method for adjusting the priority of the data computing tasks of the spacecraft edge computing method. Relevant technical personnel in this field can freely set it according to actual needs. It is only necessary to meet the needs of adjusting the priority of the data computing tasks of the spacecraft edge computing method. For example, the method for adjusting the priority of the data computing tasks of the spacecraft edge computing method can be set to adjust the data backup to the first priority, and close the real-time data computing and edge node data computing tasks. The data backup refers to the process of storing the real-time operation data of the spacecraft, the real-time data computing refers to the process of instantly processing and analyzing the data generated in real time on the spacecraft, and the edge node data computing refers to the process of data processing and analysis performed on the edge computing nodes deployed on the spacecraft.

[0080] Specifically, in step S6, when evaluating the on-orbit health of the spacecraft, the collection of multimodal data is judged according to the pre-orbit state of the spacecraft and the real-time state of the spacecraft on-orbit, wherein:

[0081] When the on-orbit pre-conditioning state of the spacecraft is an abnormal pre-conditioning state, and the on-orbit real-time state of the spacecraft is normal, it is determined to collect multimodal data, and the collected multimodal data is integrated with the real-time operation data and pre-conditioning data of the spacecraft to obtain evaluation data;

[0082] When the on-orbit pre-conditioning state of the spacecraft is an abnormal pre-conditioning state, and the on-orbit real-time state of the spacecraft is abnormal, it is determined to collect multimodal data, and the collected multimodal data is fused with the real-time operation data and pre-conditioning data of the spacecraft to obtain evaluation data;

[0083] When the on-orbit pre-conditioning state of the spacecraft is a normal pre-conditioning state, and the on-orbit real-time state of the spacecraft is normal, it is determined that the multimodal data will not be collected, and the real-time operation data of the spacecraft and the pre-conditioning data are fused to obtain evaluation data;

[0084] When the on-orbit pre-conditioning state of the spacecraft is a normal pre-conditioning state, and the on-orbit real-time state of the spacecraft is abnormal, it is determined to collect multimodal data, and the collected multimodal data is integrated with the real-time operation data and pre-conditioning data of the spacecraft to obtain evaluation data;

[0085] The evaluation data is input into each modal spacecraft evaluation model, and the evaluation coefficients output by each modal spacecraft evaluation model are obtained. Each evaluation coefficient is input into the neural network model, and the spacecraft in-orbit health assessment result is output through the neural network model.

[0086] Specifically, the multimodal data refers to data containing multiple different modes or types, and each mode can represent different types of data, such as text, images, audio, video, etc. The present embodiment does not limit the manner in which the multimodal data is integrated with the real-time operation data and pre-data of the spacecraft. Relevant technical personnel in this field can freely set it according to actual needs, and only need to meet the need to integrate the collected multimodal data with the real-time operation data and pre-data of the spacecraft into evaluation data. For example, it can be set to integrate the collected multimodal data with the real-time operation data and pre-data of the spacecraft into a data editing tool to obtain evaluation data. The spacecraft evaluation models of each mode refer to models used to evaluate the performance of different modes of a spacecraft according to the evaluation data, such as structural modal evaluation models, dynamic evaluation models, thermal evaluation models, electromagnetic evaluation models and comprehensive evaluation models.

[0087] Specifically, in step S7, when making autonomous decisions, autonomous decision-making judgments are made based on the on-orbit health assessment results of the spacecraft, wherein:

[0088] When the result of the on-orbit health assessment of the spacecraft is that the spacecraft is healthy on-orbit, no autonomous decision is made;

[0089] When the result of the on-orbit health assessment of the spacecraft is that the on-orbit operation of the spacecraft is unhealthy, an autonomous decision is made, the assessment data is input into the autonomous decision-making model, and the autonomous decision result is output.

[0090] Specifically, in step S7, when repairing the spacecraft, the spacecraft repair method is determined according to the autonomous decision result, wherein:

[0091] When the autonomous decision result is a repair execution instruction, it is determined that the execution instruction will be sent to the aircraft repair controller, and the aircraft repair controller executes the repair execution instruction;

[0092] When the autonomous decision result is a blank execution instruction, it is determined that the aircraft repair controller executes the fuzzy control algorithm.

[0093] Specifically, the fuzzy control algorithm refers to a computer digital control technology based on fuzzy set theory, fuzzy linguistic variables and fuzzy logic reasoning. The repair execution instruction refers to an instruction used to repair errors and defects in the aircraft. The blank execution instruction refers to an instruction that does not contain any actual operation. This embodiment does not limit the content of the repair execution instruction. Relevant technical personnel can freely set it according to actual needs, and it only needs to meet the needs of repairing the spacecraft. For example, the repair execution instruction can be set to be edited by the developer.

[0094] Specifically, in step S8, when calculating the error rate of the spacecraft on-orbit health assessment result, the number of errors in the spacecraft on-orbit health assessment result N1, the total number of spacecraft on-orbit health assessments N10, and the number of unhealthy and unevaluated spacecraft on-orbit operations N2 are obtained, and the error rate r of the spacecraft on-orbit health assessment result is calculated, and r=a×N1 / N10+b×N2 / N20 is set, where N20 is the preset number of unevaluated times, a is the wrong assessment weight parameter, b is the missed assessment weight parameter, a+b=1, and the values ​​of a and b are set according to the values ​​of N1 / N10 and N2 / N20, wherein:

[0095] When N1 / N10<0.5 and N2 / N20<1.7, set a=0.5, b=0.5;

[0096] When N1 / N10≥0.5 and N2 / N20<1.7, set a=0.7, b=0.3;

[0097] When N1 / N10<0.5 and N2 / N20≥1.7, set a=0.3, b=0.7;

[0098] When N1 / N10≥0.5 and N2 / N20<1.7, set a=0.5, b=0.5.

[0099] Specifically, in step S8, when the pre-data prediction process is adjusted, the error rate r of the spacecraft on-orbit health assessment result is compared with the preset error rate r0 of the spacecraft on-orbit health assessment result, and the adjustment of the pre-data prediction process is judged according to the comparison result, wherein:

[0100] When r<r0, it is determined that no adjustment is made to the pre-data prediction process;

[0101] When r≥r0, it is determined that the pre-data prediction process is adjusted and the pre-adjustment coefficient u is set, u=1+e -(r0-r) , adjust the lead time T according to the lead adjustment coefficient u, the adjusted lead time is TU, and set TU=u×T.

[0102] Specifically, the number of errors in the on-orbit health assessment results of the spacecraft refers to the number of times that the assessment results do not match the actual health status of the spacecraft during the process of health assessment of the spacecraft, the total number of on-orbit health assessments of the spacecraft refers to the total number of health assessments of the spacecraft, the number of unhealthy and unevaluated on-orbit operations of the spacecraft refers to the number of times that the health assessment is not performed due to some reason (such as failure of the assessment system and missing data, etc.) when the spacecraft is actually in an unhealthy state, the preset number of unevaluated times refers to the preset value of the number of unhealthy and unevaluated on-orbit operations of the spacecraft used to calculate the error rate of the on-orbit health assessment results of the spacecraft, the wrong assessment weight parameter refers to the parameter set to quantify the severity or impact of the wrong assessment, and the missed assessment weight parameter refers to the parameter set to quantify the severity or impact of the missed assessment.

[0103] Specifically, in step S9, the pre-data prediction duration Th is obtained, and the pre-data prediction duration Th is compared with the preset pre-data prediction duration Th0, and the optimization of the adjustment process of the pre-data prediction process is judged according to the comparison result, wherein:

[0104] When Th≤Th0, it is determined that the adjustment process of the pre-data prediction process is not optimized;

[0105] When Th>Th0, it is determined that the adjustment process of the pre-data prediction process is optimized, and the optimization coefficient h is set, 0.75<h<0.8, the preferred value is h=0.78, and the pre-adjustment coefficient u is optimized according to the optimization coefficient h. The optimized pre-adjustment coefficient is uh, and uh=h×u is set. When uh≤1, the value of uh is 1.

[0106] Specifically, the pre-data prediction duration refers to the time interval from the time node when the pre-data is input to the twin prediction model to the time node when the twin prediction model outputs the on-orbit pre-state of the spacecraft. The preset pre-data prediction duration refers to the preset time interval for judging the optimization status of the adjustment process of the pre-data prediction process. This embodiment does not limit the preset pre-data prediction duration. Relevant technical personnel in this field can freely set it according to actual needs, and it only needs to meet the need to judge the optimization status of the adjustment process of the pre-data prediction process. For example, the preset pre-data prediction duration Th0=2s can be set.

[0107] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. A method for evaluating the health of a spacecraft on-orbit operation, characterized in that: include: Step S1, collecting real-time operation data of the spacecraft; Step S2, predicting the pre-set data according to the real-time operation data of the spacecraft; Step S3, performing digital twin prediction of the spacecraft's on-orbit pre-positioned state based on the pre-positioned data; Step S4, monitoring the real-time status of the spacecraft on orbit according to the real-time operation data of the spacecraft; Step S5, adjusting the spacecraft edge computing method according to the spacecraft on-orbit pre-conditioning state and the spacecraft on-orbit real-time state; Step S6, collecting and fusing multimodal data according to the pre-orbit state and real-time state of the spacecraft on orbit, and evaluating the on-orbit health of the spacecraft to obtain the on-orbit health evaluation result of the spacecraft; Step S7, making an autonomous decision based on the on-orbit health assessment result of the spacecraft, and repairing the spacecraft based on the autonomous decision result; Step S8, calculating the error rate of the spacecraft on-orbit health assessment result, and adjusting the pre-data prediction process according to the error rate of the spacecraft on-orbit health assessment result; Step S9, optimizing the adjustment process of the pre-data prediction process according to the pre-data prediction duration; In step S3, when performing digital twin prediction on the on-orbit pre-condition of the spacecraft, the pre-condition data is input into the twin prediction model, and the twin prediction model outputs the on-orbit pre-condition of the spacecraft. When the twin prediction model outputs that the on-orbit pre-condition of the spacecraft is an abnormal pre-condition, the twin prediction model marks the time point at which the spacecraft has an abnormal on-orbit operation state within the pre-condition time as an abnormal pre-condition time point. In step S8, when calculating the error rate of the spacecraft on-orbit health assessment result, the number of errors in the spacecraft on-orbit health assessment result N1, the total number of spacecraft on-orbit health assessments N10, and the number of unhealthy and unevaluated spacecraft on-orbit operations N2 are obtained, and the error rate r of the spacecraft on-orbit health assessment result is calculated, and r=a×N1 / N10+b×N2 / N20 is set, where N20 is the preset number of unevaluated times, a is the wrong evaluation weight parameter, b is the missed evaluation weight parameter, a+b=1, and the values ​​of a and b are set according to the values ​​of N1 / N10 and N2 / N20, wherein: When N1 / N10<0.5 and N2 / N20<1.7, set a=0.5, b=0.5; When N1 / N10≥0.5 and N2 / N20<1.7, set a=0.7, b=0.3; When N1 / N10<0.5 and N2 / N20≥1.7, set a=0.3, b=0.7; When N1 / N10≥0.5 and N2 / N20<1.7, set a=0.5, b=0.5; In step S8, when the pre-data prediction process is adjusted, the error rate r of the spacecraft on-orbit health assessment result is compared with the preset error rate r0 of the spacecraft on-orbit health assessment result, and the adjustment of the pre-data prediction process is judged according to the comparison result, wherein: When r<r0, it is determined that no adjustment is made to the pre-data prediction process; When r≥r0, it is determined that the pre-data prediction process needs to be adjusted and the pre-adjustment coefficient u is set, u=1+e -（r0-r） , adjust the lead time T according to the lead adjustment coefficient u, the adjusted lead time is TU, and set TU=u×T; The lead time refers to the time interval from the current time node of the spacecraft to the preset time node.

2. The spacecraft on-orbit health assessment method according to claim 1, characterized in that: In step S4, when monitoring the real-time state of the spacecraft on orbit, the real-time operation data of the spacecraft is compared with the preset spacecraft standard operation data threshold, and the real-time state of the spacecraft on orbit is monitored according to the comparison result, wherein: When all the real-time operation data of the spacecraft fall within the preset spacecraft standard operation data threshold, it is determined that the real-time status of the spacecraft on orbit is normal; When there is real-time spacecraft operation data that falls outside the preset spacecraft standard operation data threshold, the real-time state of the spacecraft on orbit is judged to be abnormal, and the real-time spacecraft operation data that falls outside the preset spacecraft standard operation data threshold is used as key abnormal data.

3. The spacecraft on-orbit health assessment method according to claim 1, characterized in that: In step S5, when the spacecraft edge computing mode is adjusted, the adjustment of the spacecraft edge computing mode is judged according to the spacecraft on-orbit pre-conditioning state and the spacecraft on-orbit real-time state, wherein: When the spacecraft on-orbit pre-conditioning state is an abnormal pre-conditioning state and the spacecraft on-orbit real-time state is normal, it is determined that the priority of the data computing task of the spacecraft edge computing method is adjusted; When the on-orbit pre-conditioning state of the spacecraft is an abnormal pre-conditioning state, and the on-orbit real-time state of the spacecraft is abnormal, it is determined that the priority of the data computing tasks of the spacecraft edge computing method is adjusted; When the spacecraft on-orbit pre-conditioning state is a normal pre-conditioning state and the spacecraft on-orbit real-time state is normal, it is determined that the priority of the data computing task of the spacecraft edge computing mode is not adjusted; When the on-orbit pre-conditioning state of the spacecraft is a normal pre-conditioning state and the on-orbit real-time state of the spacecraft is abnormal, it is determined to adjust the priority of the data computing tasks of the spacecraft edge computing method.

4. The spacecraft on-orbit health assessment method according to claim 1, characterized in that: In step S6, when evaluating the on-orbit health of the spacecraft, the collection of multimodal data is judged according to the pre-orbit state of the spacecraft and the real-time state of the spacecraft on-orbit, wherein: When the on-orbit pre-conditioning state of the spacecraft is an abnormal pre-conditioning state, and the on-orbit real-time state of the spacecraft is normal, it is determined to collect multimodal data, and the collected multimodal data is fused with the real-time operation data and pre-conditioning data of the spacecraft to obtain evaluation data; When the on-orbit pre-conditioning state of the spacecraft is an abnormal pre-conditioning state, and the on-orbit real-time state of the spacecraft is abnormal, it is determined to collect multimodal data, and the collected multimodal data is fused with the real-time operation data and pre-conditioning data of the spacecraft to obtain evaluation data; When the on-orbit pre-conditioning state of the spacecraft is a normal pre-conditioning state, and the on-orbit real-time state of the spacecraft is normal, it is determined that the multimodal data will not be collected, and the real-time operation data of the spacecraft and the pre-conditioning data are fused to obtain evaluation data; When the on-orbit pre-conditioning state of the spacecraft is a normal pre-conditioning state, and the on-orbit real-time state of the spacecraft is abnormal, it is determined to collect multimodal data, and the collected multimodal data is integrated with the real-time operation data and pre-conditioning data of the spacecraft to obtain evaluation data; The evaluation data is input into each modal spacecraft evaluation model, and the evaluation coefficients output by each modal spacecraft evaluation model are obtained. Each evaluation coefficient is input into the neural network model, and the spacecraft in-orbit health assessment result is output through the neural network model.

5. The spacecraft on-orbit health assessment method according to claim 1, characterized in that: In step S7, when making autonomous decisions, autonomous decision-making judgments are made based on the on-orbit health assessment results of the spacecraft, wherein: When the result of the on-orbit health assessment of the spacecraft is that the spacecraft is healthy on-orbit, no autonomous decision is made; When the result of the on-orbit health assessment of the spacecraft is that the on-orbit operation of the spacecraft is unhealthy, an autonomous decision is made, the assessment data is input into the autonomous decision-making model, and the autonomous decision result is output.

6. The spacecraft on-orbit health assessment method according to claim 5, characterized in that: In step S7, when repairing the spacecraft, the spacecraft repair method is determined according to the autonomous decision result, wherein: When the autonomous decision result is a repair execution instruction, it is determined that the execution instruction will be sent to the aircraft repair controller, and the aircraft repair controller executes the repair execution instruction; When the autonomous decision result is a blank execution instruction, it is determined that the aircraft repair controller executes the fuzzy control algorithm.

7. The spacecraft on-orbit health assessment method according to claim 1, characterized in that: In the step S9, the pre-data prediction duration Th is obtained, and the pre-data prediction duration Th is compared with the preset pre-data prediction duration Th0, and the optimization of the adjustment process of the pre-data prediction process is judged according to the comparison result, wherein: When Th≤Th0, it is determined that the adjustment process of the pre-data prediction process is not optimized; When Th>Th0, it is determined that the adjustment process of the pre-data prediction process is optimized, and the optimization coefficient h is set, 0.75<h<0.8, and the pre-adjustment coefficient u is optimized according to the optimization coefficient h. The optimized pre-adjustment coefficient is uh, and uh=h×u is set. When uh≤1, the value of uh is 1.

Images