CN112631240A

CN112631240A - Spacecraft fault active detection method and device

Info

Publication number: CN112631240A
Application number: CN202011416177.1A
Authority: CN
Inventors: 周倜; 苗毅; 欧余军; 唐卿; 刘晓辉; 欧阳柳; 王磊; 张祖丽; 王尧
Original assignee: Unit 63920 Of Pla
Current assignee: Unit 63920 Of Pla
Priority date: 2020-12-07
Filing date: 2020-12-07
Publication date: 2021-04-09

Abstract

The embodiment of the application provides a spacecraft fault active detection method and a spacecraft fault active detection device, wherein the method comprises the following steps: acquiring abnormal scores corresponding to normal mode expected data respectively corresponding to each telemetering parameter of the current time of the spacecraft by using a preset parameter normal trend model and the current actual telemetering data of the spacecraft, wherein the parameter normal trend model is obtained by training normal historical telemetering data of the spacecraft in a preset time period; and determining whether the current actual telemetering data of the spacecraft contains the fault telemetering parameters or not based on the abnormal values and preset abnormal threshold values respectively corresponding to the telemetering parameters, and if so, outputting the abnormal data corresponding to the fault telemetering parameters. According to the method and the device, the downlink normal data can be utilized to the maximum extent under the condition that a spacecraft fault sample is lacked, the correct change model of the data is learned, and active detection of the spacecraft fault is achieved.

Description

Spacecraft fault active detection method and device

Technical Field

The application relates to the technical field of spacecraft detection, in particular to a spacecraft fault active detection method and device.

Background

In the implementation process of the space mission, the requirement on the cooperative flight control capability of the multiple spacecrafts is higher. In the operation stage, the operation and control center needs to complete uplink control on a plurality of spacecrafts at the same time, the control process is more complex, and the timeliness requirement is higher. The center needs to organize the in-orbit spacecraft platform, the passenger equipment and the extravehicular space suit according to a failure plan, and to judge and dispose the failure of the load platform equipment and the main load. When part of the spacecraft runs for 10 to 15 years in orbit, functional parts of a spacecraft platform are attenuated along with the time, and the probability of emergency and fault occurrence is increased.

At present, a fault detection software system usually takes a fault diagnosis mode based on rules and models as a main mode, and under the condition of no fault positive case or lack of fault samples, because a spacecraft fault model is difficult to obtain, and the mode belongs to passive monitoring, the existing spacecraft fault diagnosis mode cannot actively carry out fault judgment and early warning on the state of a spacecraft, the automation level of a fault handling process is low, the personnel dependence degree is high, and the requirement of intelligent operation and maintenance of fault diagnosis of a space mission cannot be met. Meanwhile, fault diagnosis based on data analysis is still in a research stage, and a set of complete design framework is not provided.

Disclosure of Invention

Aiming at the problems in the prior art, the method and the device for actively detecting the spacecraft fault can maximally utilize downlink normal data and learn a correct change model of the data under the condition of lacking a spacecraft fault sample, realize the active detection of the spacecraft fault, effectively improve the real-time performance, the automation degree and the efficiency of the spacecraft fault detection, and effectively improve the prediction accuracy and the reliability of the spacecraft fault.

In order to solve the technical problem, the application provides the following technical scheme:

in a first aspect, the present application provides a method for actively detecting a spacecraft fault, including:

acquiring abnormal scores corresponding to normal mode expected data respectively corresponding to each telemetering parameter of the current time of the spacecraft by using a preset parameter normal trend model and the current actual telemetering data of the spacecraft, wherein the parameter normal trend model is obtained by training normal historical telemetering data of the spacecraft in a preset time period;

and determining whether the current actual telemetering data of the spacecraft contains the fault telemetering parameters or not based on the abnormal values and preset abnormal threshold values respectively corresponding to the telemetering parameters, and if so, outputting the abnormal data corresponding to the fault telemetering parameters.

Further, the parametric normal trend model includes: a single parameter model;

correspondingly, before the obtaining of the abnormal score corresponding to the normal mode expected data respectively corresponding to each telemetry parameter of the spacecraft at the current time, the method further comprises the following steps:

acquiring normal historical telemetry data of the spacecraft in a preset time period;

preprocessing the normal historical telemetry data to form standardized data meeting a preset standard format;

and generating a single-parameter model corresponding to each telemetering parameter according to the standardized data, wherein the single-parameter model is used for outputting normal mode expected data of the telemetering parameter at the current time according to the input standardized data corresponding to the telemetering parameter, and is also used for outputting an abnormal score of the telemetering parameter.

Further, the parametric normal trend model further comprises: a correlation model;

correspondingly, after the forming of the standardized data conforming to the preset standard format, the method further includes:

performing relevance processing on the standardized data by using a preset data relevance analysis mode to obtain a relevance relation among the remote sensing parameters;

analyzing and processing the standard data to obtain a corresponding working condition characteristic group;

training to obtain a relevant model corresponding to the incidence relation of each remote sensing parameter according to the single parameter model corresponding to each remote sensing parameter, the relevant relation among the remote sensing parameters and the working condition characteristic group; the correlation model is used for outputting normal mode expected data of the corresponding remote sensing parameters at the current time according to the input standardized data of the corresponding remote sensing parameters and is also used for outputting abnormal scores of the corresponding remote sensing parameters.

Further, before the obtaining of the abnormal score corresponding to the normal mode expected data respectively corresponding to each telemetry parameter of the current time of the spacecraft, the method further includes:

and automatically selecting one of the abnormal scores corresponding to the telemetry parameters as an abnormal threshold value based on a preset threshold value selection mode.

Further, the applying a preset data correlation analysis mode to perform correlation processing on the standardized data to obtain a correlation relationship between the remote sensing parameters includes:

dividing each telemetering parameter in the normal historical telemetering data into a parameter group corresponding to each control subsystem of the spacecraft respectively;

setting a corresponding correlation degree threshold value according to the correlation degree between the telemetry parameters in each parameter group;

and constructing a knowledge graph for representing the correlation among the remote sensing parameters based on the correlation degree threshold among the remote sensing parameters in each parameter group, and establishing a tree structure of fault association relations among the remote sensing parameters to determine the correlation relations among the remote sensing parameters.

Further, still include:

according to the normal historical telemetering data of the spacecraft in a preset time period, applying the single-parameter model and the related model by a set step length to sequentially predict normal mode expected data of each telemetering parameter and abnormal scores corresponding to each telemetering parameter at each extrapolation time point;

judging whether the time point of the spacecraft in each extrapolation time point contains a time point of telemetering parameter prediction failure or not based on the abnormal score and a preset abnormal threshold value respectively corresponding to each telemetering parameter, and if yes, determining the time point as the failure prediction occurrence time;

and outputting abnormal data corresponding to the failure prediction occurrence time.

Further, the applying a preset parameter normal trend model and the current actual telemetry data of the spacecraft to obtain the abnormal scores corresponding to the normal mode expected data respectively corresponding to each telemetry parameter of the current time of the spacecraft includes:

inputting the standardized data into the parameter normal trend model, and determining normal mode expected data corresponding to each telemetering parameter of the spacecraft at the current time according to the output of the parameter normal trend model;

and determining the current abnormal score of each telemetry parameter according to the comparison result of the current actual telemetry data of the spacecraft and the expected data of the normal mode.

Further, the preprocessing the normal historical telemetry data comprises:

and performing at least one of wild value elimination, vacancy value filling and normalization processing on the normal historical telemetry data.

In a second aspect, the present application provides an active spacecraft fault detection apparatus, including:

the real-time detection module is used for applying a preset parameter normal trend model and current actual telemetering data of the spacecraft to obtain abnormal scores corresponding to normal mode expected data respectively corresponding to each telemetering parameter of the current time of the spacecraft, wherein the parameter normal trend model is obtained by applying normal historical telemetering data of the spacecraft in a preset time period for training;

and the abnormal result calculation module is used for determining whether the current actual telemetering data of the spacecraft contains the fault telemetering parameters or not based on the abnormal scores and the preset abnormal threshold values respectively corresponding to the telemetering parameters, and if so, outputting the abnormal data corresponding to the fault telemetering parameters.

Further, the parametric normal trend model includes: a single parameter model;

correspondingly, the active spacecraft fault detection device further comprises:

a data preprocessing module for executing the following:

and the model training module is used for generating a single-parameter model corresponding to each telemetering parameter according to the standardized data, wherein the single-parameter model is used for outputting normal mode expected data of the current time of the telemetering parameter according to the input standardized data corresponding to the telemetering parameter and outputting an abnormal score of the telemetering parameter.

the correlation analysis module is used for performing correlation processing on the standardized data by applying a preset data correlation analysis mode to obtain a correlation relation among the remote sensing parameters;

the working condition analysis module is used for carrying out working condition analysis processing on the standardized data to obtain a corresponding working condition characteristic group;

the model training module is also used for training to obtain a relevant model corresponding to the incidence relation of each remote sensing parameter according to the single parameter model corresponding to each remote sensing parameter, the relevant relation among the remote sensing parameters and the working condition characteristic group; the correlation model is used for outputting normal mode expected data of the corresponding remote sensing parameters at the current time according to the input standardized data of the corresponding remote sensing parameters and is also used for outputting abnormal scores of the corresponding remote sensing parameters.

Further, still include:

and the threshold selection module is used for automatically selecting one of the abnormal scores corresponding to the telemetry parameters as an abnormal threshold based on a preset threshold selection mode.

Further, the correlation analysis module is specifically configured to perform the following:

Further, still include:

a failure prediction module to perform the following:

Further, the real-time detection module is specifically configured to execute the following:

Further, the data preprocessing module is specifically configured to perform at least one of outlier rejection, vacancy filling, and normalization processing on the normal historical telemetry data.

In a third aspect, the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor implements the active spacecraft failure detection method when executing the program.

In a fourth aspect, the present application provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method for active detection of spacecraft faults as described.

According to the technical scheme, the spacecraft fault active detection method and device provided by the application comprise the following steps: acquiring abnormal scores corresponding to normal mode expected data respectively corresponding to each telemetering parameter of the current time of the spacecraft by using a preset parameter normal trend model and the current actual telemetering data of the spacecraft, wherein the parameter normal trend model is obtained by training normal historical telemetering data of the spacecraft in a preset time period; and determining whether the current actual telemetering data of the spacecraft contains the fault telemetering parameters or not based on the abnormal values and the preset abnormal threshold values respectively corresponding to the telemetering parameters, if so, outputting the abnormal data corresponding to the fault telemetering parameters, and learning a correct change model of data by maximally utilizing downlink normal data under the condition of lacking a spacecraft fault sample, so that the active detection of the spacecraft fault is realized, the real-time performance, the automation degree and the efficiency of the spacecraft fault detection can be effectively improved, and the prediction accuracy and the reliability of the spacecraft fault can be effectively improved.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

Fig. 1 is a first flowchart of an active spacecraft fault detection method in an embodiment of the present application.

Fig. 2 is a schematic flow chart of steps 010 to 030 in the active spacecraft failure detection method in the embodiment of the present application.

Fig. 3 is a schematic flowchart of steps 010 to 060 in the active detection method for spacecraft failure in the embodiment of the present application.

Fig. 4 is a schematic flow chart of steps 010 to 070 in the active spacecraft fault detection method in the embodiment of the present application.

Fig. 5 is a detailed flowchart of step 040 in the active spacecraft failure detection method in the embodiment of the present application.

Fig. 6 is a second flowchart of an active spacecraft fault detection method in an embodiment of the present application.

Fig. 7 is a detailed flowchart of step 100 in the active detection method for spacecraft failure in the embodiment of the present application.

Fig. 8 is a first structural schematic diagram of an active spacecraft failure detection apparatus in an embodiment of the present application.

Fig. 9 is a second structural schematic diagram of an active spacecraft failure detection apparatus in an embodiment of the present application.

Fig. 10 is a third structural schematic diagram of an active spacecraft failure detection apparatus in an embodiment of the present application.

Fig. 11 is a fourth structural schematic diagram of an active spacecraft failure detection apparatus in an embodiment of the present application.

Fig. 12 is a fifth structural schematic diagram of an active spacecraft failure detection apparatus in an embodiment of the present application.

Fig. 13 is a data flow diagram of an active detection method for spacecraft fault according to an application example of the present application.

FIG. 14 is a diagram illustrating a detailed process of model generation provided by an application example of the present application.

FIG. 15 is a schematic representation of a knowledge map of telemetry data correlations provided in an example application of the present application.

Fig. 16 is a schematic structural diagram of an electronic device in the embodiment of the present application.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are some embodiments of the present application, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

The method aims at the problems that the existing software system mainly adopts a rule-based and model-based fault diagnosis mode, the model is difficult to obtain, the passive monitoring is adopted, the fault judgment and early warning cannot be actively carried out on the state of the spacecraft, the automation level of the fault handling process is low, the personnel dependence degree is high, and the requirement of intelligent operation and maintenance of the fault diagnosis of the space mission cannot be met. The method comprises the steps of analyzing a fault of a spacecraft, and judging whether the fault of the spacecraft is correct or not according to the fault sample, wherein the fault sample is a failure sample, and if the fault sample is not correct, the fault sample is acquired and then the failure sample is analyzed.

The detailed description is provided in the following embodiments and application examples.

In order to realize active detection of a spacecraft fault, the application provides an embodiment of an active detection method for a spacecraft fault, and referring to fig. 1, the active detection method for a spacecraft fault is composed of an online part and an offline part, wherein the online part specifically includes the following contents:

step 100: and acquiring abnormal scores corresponding to the normal mode expected data respectively corresponding to each telemetering parameter of the current time of the spacecraft by using a preset parameter normal trend model and the current actual telemetering data of the spacecraft, wherein the parameter normal trend model is obtained by training normal historical telemetering data of the spacecraft in a preset time period.

Step 200: and determining whether the current actual telemetering data of the spacecraft contains the fault telemetering parameters or not based on the abnormal values and preset abnormal threshold values respectively corresponding to the telemetering parameters, and if so, outputting the abnormal data corresponding to the fault telemetering parameters.

It is to be understood that reference to normal in one or more embodiments of the present application refers to a non-fault condition.

Specifically, the current parameter normal trend model may be applied to calculate the normal mode value at the current time, and the normal mode value may be compared with the actual data to obtain the abnormal score of each parameter. And judging the abnormal score according to the trained abnormal threshold a, if the abnormal score of the parameter is lower than the abnormal threshold, declaring the parameter to be abnormal, and otherwise, declaring the parameter to be normal. When the parameter is determined to be abnormal, the result is written into the database, and the abnormal record is issued to the client.

As can be seen from the above description, the spacecraft fault active detection method provided in the embodiment of the present application can maximally utilize downlink normal data and learn a correct change model of data in the absence of a spacecraft fault sample, can actively detect a spacecraft fault, can effectively improve the real-time performance, the automation degree and the efficiency of spacecraft fault detection, and can effectively improve the prediction accuracy and the reliability of the spacecraft fault.

In order to improve the application accuracy and reliability of the parameter normal trend model, in an embodiment of the active detection method for spacecraft fault, the parameter normal trend model includes: a single parameter model; referring to fig. 2, before step 100 of the active spacecraft fault detection method, the following offline contents are also specifically included:

step 010: and acquiring normal historical telemetry data of the spacecraft in a preset time period.

Step 020: and preprocessing the normal historical telemetry data to form standardized data which accords with a preset standard format.

Step 030: and generating a single-parameter model corresponding to each telemetering parameter according to the standardized data, wherein the single-parameter model is used for outputting normal mode expected data of the telemetering parameter at the current time according to the input standardized data corresponding to the telemetering parameter, and is also used for outputting an abnormal score of the telemetering parameter.

Specifically, telemetering data of the spacecraft stored in a database in advance in a period of time is subjected to data preprocessing, and standardized data in accordance with a preset standard format are obtained. And performing autonomous learning on the standardized data to form a single-parameter model corresponding to each single parameter in the telemetering data.

In one example, the single parameter model may employ a machine learning model such as an exponentially weighted average (EWMA) or a deep learning algorithm LSTM (long-short term).

In order to further improve the application comprehensiveness and accuracy of the parameter normal trend model, in an embodiment of the active spacecraft fault detection method, the parameter normal trend model further includes: a correlation model; referring to fig. 3, before step 100 of the active detection method for spacecraft failure, the following offline partial contents executed after step 020 are also specifically included:

step 040: and performing relevance processing on the standardized data by using a preset data relevance analysis mode to obtain the relevance relation among the remote sensing parameters.

Step 050: and analyzing and processing the standard data to obtain a corresponding working condition characteristic group.

Step 060: training to obtain a relevant model corresponding to the incidence relation of each remote sensing parameter according to the single parameter model corresponding to each remote sensing parameter, the relevant relation among the remote sensing parameters and the working condition characteristic group; the correlation model is used for outputting normal mode expected data of the corresponding remote sensing parameters at the current time according to the input standardized data of the corresponding remote sensing parameters and is also used for outputting abnormal scores of the corresponding remote sensing parameters.

Specifically, correlation analysis processing is performed on the standardized data to obtain correlation relations among the parameters.

In one example, the correlation model may be a machine learning model such as a maximum Information correlation coefficient mic (maximum Information coefficient), an isolated forest algorithm, a Pearson algorithm, or a Spearman algorithm.

In order to further improve the efficiency and reliability of the active detection of the spacecraft fault, in an embodiment of the active detection method of the spacecraft fault, referring to fig. 4, before step 100 in the active detection method of the spacecraft fault, the following offline contents executed after step 020 are further specifically included:

step 070: and automatically selecting one of the abnormal scores corresponding to the telemetry parameters as an abnormal threshold value based on a preset threshold value selection mode.

In an example of an implementation of step 040, referring to fig. 5, step 040 includes the following steps:

step 041: and respectively dividing each telemetry parameter in the normal historical telemetry data into a parameter group corresponding to each control subsystem of the spacecraft.

Step 042: and setting a corresponding correlation degree threshold according to the correlation degree between the telemetry parameters in each parameter group.

Step 043: and constructing a knowledge graph for representing the correlation among the remote sensing parameters based on the correlation degree threshold among the remote sensing parameters in each parameter group, and establishing a tree structure of fault association relations among the remote sensing parameters to determine the correlation relations among the remote sensing parameters.

In particular, due to the complexity of the equipment on the spacecraft, the parameters cannot be represented by simple stacking or even linear relationships with weights. Therefore, in the data relationship, it is necessary to obtain the coefficient in the multi-parameter fusion through the correlation analysis of the data in the time domain or the frequency domain. The universality and the timeliness of a plurality of mining algorithms are tested, and the universality and the timeliness of none of the algorithms are excellent. Considering that the correlation of telemetry data is relatively complex, most of the telemetry data is strong nonlinear and uncertain correlation, in order to comprehensively discover the correlation among parameters, an algorithm with good universality should be selected, and therefore the correlation between data is calculated by selecting the maximum information correlation coefficient (MIC) with good universality but poor timeliness. Although the correlation relationship of the test is complex, the task data is not used for testing, and the task data is used for testing the universality and the timeliness of the test.

It can be understood that, each parameter in the telemetering data is divided according to a preset subsystem, and then the correlation among the parameters in each subsystem is mined; setting corresponding correlation degree threshold values according to the correlation degrees of all parameters in all subsystems to construct a knowledge graph for expressing the correlation among all parameters in the telemetering data and establish a tree structure of fault association relations among all parameters; determining a navigation fault reason of the spacecraft based on the tree structure of the fault association relation among the parameters; and analyzing and processing the standard data to obtain a corresponding working condition characteristic group.

In order to further realize spacecraft fault prediction on the basis of realizing active detection of spacecraft fault, in an embodiment of an active detection method for spacecraft fault, referring to fig. 6, the active detection method for spacecraft fault further specifically includes the following offline execution contents:

step 300: according to the normal historical telemetering data of the spacecraft in a preset time period, applying the single-parameter model and the related model by a set step length to sequentially predict normal mode expected data of each telemetering parameter and abnormal scores corresponding to each telemetering parameter at each extrapolation time point;

step 400: judging whether the time point of the spacecraft in each extrapolation time point contains a time point of telemetering parameter prediction failure or not based on the abnormal score and a preset abnormal threshold value respectively corresponding to each telemetering parameter, and if yes, determining the time point as the failure prediction occurrence time;

step 500: and outputting abnormal data corresponding to the failure prediction occurrence time.

In order to further improve the efficiency and reliability of the active detection of the spacecraft fault, in an embodiment of the active detection method of the spacecraft fault, referring to fig. 7, step 100 of the active detection method of the spacecraft fault specifically includes the following online execution contents:

step 110: and acquiring normal historical telemetry data of the spacecraft in a preset time period.

Step 120: and preprocessing the normal historical telemetry data to form standardized data which accords with a preset standard format.

Step 130: and inputting the standardized data into the parameter normal trend model, and determining normal mode expected data corresponding to each telemetering parameter of the spacecraft at the current time according to the output of the parameter normal trend model.

Step 140: and determining the current abnormal score of each telemetry parameter according to the comparison result of the current actual telemetry data of the spacecraft and the expected data of the normal mode.

Specifically, the client may send an instruction to drive the failure prediction service, where the prediction process is to perform time extrapolation according to the training model y ═ F (t, a) and the historical data, and calculate whether the parameter value y0 in the process of reaching the extrapolation point t0 satisfies the failure state (e.g., exceeds the valid range, satisfies the failure rule, or exceeds the threshold value in deviation from the normal model, etc.). If the fault rule (or the deviation from the normal model) is detected, the parameter value y needs to be gradually extrapolated by a certain step length, whether the extrapolation result meets the rule (or the deviation exceeds a threshold value) or not is calculated, if yes, the time t generated by y is the fault occurrence time, and the early warning of the fault rule (or the deviation from the normal model) is realized. If y is a monotonic function, then the method can be simplified to determine if the parameter value exceeds the valid range y0>a (or y0<a) Can be obtained by inverting the function F^-1Inverse solution time ta ═ F^-1(a) Thereby, the expected occurrence time ta of the fault is given, and the fault is predicted.

In order to further improve the accuracy of the data basis of the active detection of the spacecraft fault, in an embodiment of the active detection method of the spacecraft fault, step 020 of the active detection method of the spacecraft fault specifically includes the following offline execution contents:

In terms of software, in order to realize active detection of a spacecraft fault, the present application provides an embodiment of an active detection device for a spacecraft fault, which is used for realizing all or part of contents in the active detection method for a spacecraft fault, and referring to fig. 8, the active detection device for a spacecraft fault specifically includes the following contents:

the real-time detection module 10 is configured to apply a preset parameter normal trend model and current actual telemetry data of the spacecraft to obtain abnormal scores corresponding to normal mode expected data respectively corresponding to each telemetry parameter of the current time of the spacecraft, where the parameter normal trend model is obtained by training normal historical telemetry data of the spacecraft in a preset time period;

and the abnormal result calculating module 20 is configured to determine whether the current actual telemetry data of the spacecraft includes the fault telemetry parameter based on the abnormal score and the preset abnormal threshold value respectively corresponding to each telemetry parameter, and if yes, output the abnormal data corresponding to the fault telemetry parameter.

The embodiment of the active detection device for spacecraft fault provided by the application can be specifically used for executing the processing flow of the embodiment of the active detection method for spacecraft fault in the above embodiment, and the functions of the active detection device for spacecraft fault are not described herein again, and reference can be made to the detailed description of the embodiment of the method.

As can be seen from the above description, the active detection device for spacecraft fault provided in the embodiment of the present application can maximally utilize downlink normal data and learn a correct change model of data in the absence of a spacecraft fault sample, can actively detect spacecraft fault, can effectively improve the real-time performance, the automation degree and the efficiency of spacecraft fault detection, and can effectively improve the accuracy and the reliability of prediction of spacecraft fault.

In order to improve the application accuracy and reliability of the parameter normal trend model, in an embodiment of the active spacecraft failure detection apparatus, the parameter normal trend model includes: a single parameter model; referring to fig. 9, the active spacecraft fault detection apparatus further specifically includes the following offline partial contents:

a data preprocessing module 01, configured to perform the following:

And the model training module 02 is used for generating a single-parameter model corresponding to each telemetering parameter according to the standardized data, wherein the single-parameter model is used for outputting normal mode expected data of the telemetering parameter at the current time according to the input standardized data corresponding to the telemetering parameter, and is also used for outputting an abnormal score of the telemetering parameter.

In order to further improve the application comprehensiveness and accuracy of the parameter normal trend model, in an embodiment of the active spacecraft fault detection apparatus, the parameter normal trend model further includes: a correlation model; referring to fig. 10, the active spacecraft failure detection apparatus further specifically includes the following offline part contents:

the correlation analysis module 03 is configured to apply a preset data correlation analysis mode to perform correlation processing on the standardized data to obtain a correlation relationship between the remote sensing parameters;

the working condition analysis module 04 is used for carrying out working condition analysis processing on the standardized data to obtain a corresponding working condition characteristic group;

the model training module 02 is further configured to train to obtain a correlation model corresponding to the correlation relationship of each remote sensing parameter according to the single-parameter model corresponding to each remote sensing parameter, the correlation relationship among the remote sensing parameters, and the working condition feature set; the correlation model is used for outputting normal mode expected data of the corresponding remote sensing parameters at the current time according to the input standardized data of the corresponding remote sensing parameters and is also used for outputting abnormal scores of the corresponding remote sensing parameters.

In order to further improve the efficiency and reliability of the active detection of the spacecraft fault, in an embodiment of an active detection device of the spacecraft fault, referring to fig. 11, the active detection device of the spacecraft fault further includes the following offline contents:

and the threshold selection module 05 is configured to automatically select one of the multiple abnormality scores corresponding to the telemetry parameters as an abnormality threshold based on a preset threshold selection manner.

In a specific application example of the correlation analysis module 03, the correlation analysis module 03 is specifically configured to perform the following:

It can be understood that the correlation analysis module 03 divides each parameter in the telemetry data according to the preset subsystem, and then mines the correlation between the parameters in each subsystem; setting corresponding correlation degree threshold values according to the correlation degrees of all parameters in all subsystems to construct a knowledge graph for expressing the correlation among all parameters in the telemetering data and establish a tree structure of fault association relations among all parameters; determining a navigation fault reason of the spacecraft based on the tree structure of the fault association relation among the parameters; and analyzing and processing the standard data to obtain a corresponding working condition characteristic group.

In order to further realize spacecraft fault prediction based on the realization of spacecraft fault active detection, in an embodiment of a spacecraft fault active detection device, referring to fig. 12, the spacecraft fault active detection device further specifically includes the following offline execution contents:

a failure prediction module 30 configured to perform the following:

In order to further improve the efficiency and reliability of the active detection of a spacecraft fault, in an embodiment of the active detection device of a spacecraft fault, the real-time detection module 10 of the active detection device of a spacecraft fault is specifically configured to perform the following:

Specifically, the client may send an instruction to drive the failure prediction service, where the prediction process is to perform time extrapolation according to the training model y ═ F (t, a) and the historical data, and calculate whether the parameter value y0 in the process of reaching the extrapolation point t0 satisfies the failure state (e.g., exceeds the valid range, satisfies the failure rule, or exceeds the threshold value in deviation from the normal model, etc.). If the fault rule (or the deviation from the normal model) is detected, the parameter value y needs to be gradually extrapolated by a certain step length, whether the extrapolation result meets the rule (or the deviation exceeds a threshold value) or not is calculated, if yes, the time t generated by y is the fault occurrence time, and the early warning of the fault rule (or the deviation from the normal model) is realized. If y is a monotonic function, F (t, a), the computing device can be simplified to determine if the parameter value exceeds the valid range y0>a (or y0<a) Can be obtained by inverting the function F^-1Inverse solution time ta ═ F^-1(a)，Therefore, the expected occurrence time ta of the fault is given, and the fault is predicted.

In order to further improve the accuracy of the data basis of the active detection of the spacecraft fault, in an embodiment of the active detection device of the spacecraft fault, the data preprocessing module 01 is specifically configured to perform at least one of outlier elimination, vacancy filling and normalization processing on the normal historical telemetry data.

To further illustrate the solution, the present application further provides a specific application example of implementing the spacecraft fault active detection method by using the spacecraft fault active detection apparatus, and referring to fig. 13, the spacecraft fault active detection method specifically includes:

from the perspective of data analysis, spacecraft faults are mainly classified into two categories: single parameter faults and multi-parameter correlated faults. The single-parameter fault analysis comprises abnormal detection of large fluctuation of parameter values, whether the parameter values exceed a threshold range, whether a mode is suddenly changed, a sudden change state monitoring function based on the mode and a fault prediction function based on a trend. The multi-parameter correlation fault comprises an incidence relation construction function, an incidence state detection function and a fault correlation analysis function based on a tree structure. From the technical implementation point of view, the system is divided into an off-line part (model training and fault prediction) and an on-line part (real-time anomaly detection). The model training comprises the learning of a single-parameter normal change model, the calculation of an abnormal threshold value and the mining of a multi-parameter association relation. The fault prediction uses a learned model to extrapolate whether a fault occurs within a certain time. The on-line part utilizes the training model to perform anomaly detection.

The data preprocessing module provides standardized data for the function, the model training module in the function reads data in a period of time from the database and then performs autonomous learning to form a telemetering parameter model for a single parameter and generate a relevant model for a multi-parameter incidence relation, and the detailed process of the model generation is shown in fig. 14. The model captures the normal pattern of the time series and outputs an anomaly score for each observation. The threshold selection module uses these anomaly scores to automatically select an anomaly threshold in a non-parametric manner. Such an offline training procedure may be performed periodically, for example, weekly or monthly.

The correlation analysis module divides the telemetry parameters according to the subsystems and then mines the correlation of the parameters in each subsystem. Due to the complexity of the equipment on the spacecraft, the parameters cannot be represented by simple stacking or even linear relationships with weights. Therefore, in the data relationship, it is necessary to obtain the coefficient in the multi-parameter fusion through the correlation analysis of the data in the time domain or the frequency domain. The universality and the timeliness of a plurality of mining algorithms are tested, and the universality and the timeliness of none of the algorithms are excellent. Considering that the correlation of telemetry data is relatively complex, most of the telemetry data is strong nonlinear and uncertain correlation, in order to comprehensively discover the correlation among parameters, an algorithm with good universality should be selected, and therefore the correlation between data is calculated by selecting the maximum information correlation coefficient (MIC) with good universality but poor timeliness. Although the correlation relationship of the test is complex, the task data is not used for testing, and the task data is used for testing the universality and the timeliness of the test.

And the correlation analysis module calculates the correlation degree among the parameters, sets a threshold value according to the correlation degree and groups the parameters. Referring to fig. 15, a knowledge graph of telemetry data correlation is constructed, a tree structure of a fault association relation is established, further, deep-level reason analysis of the fault is navigated, the goal of fault analysis is improved, and the diagnosis blindness is reduced.

The online detection part mainly comprises a real-time detection module and an abnormal result calculation module, wherein the real-time detection module calculates an expected value of a current time parameter according to a training model and then compares the expected value with actual data to obtain an abnormal score. And the abnormal result calculation module judges the abnormal score according to the trained abnormal threshold a, if the abnormal score of the parameter is lower than the abnormal threshold, the parameter is declared to be abnormal, and if not, the parameter is normal. When the parameter is determined to be abnormal, the result is written into the database, and the abnormal record is issued to the client.

The fault prediction function module sends an instruction to drive a fault prediction service by a client, and the prediction process is carried out according to a training model of y, F (t, A) andand (3) carrying out time extrapolation on the historical data, and calculating whether the parameter value y0 meets the fault state (such as exceeding a valid range, meeting a fault rule, or exceeding a threshold value in deviation from a normal model) in the process of reaching an extrapolation point t 0. If the fault rule (or the deviation from the normal model) is detected, the parameter value y needs to be gradually extrapolated by a certain step length, whether the extrapolation result meets the rule (or the deviation exceeds a threshold value) or not is calculated, if yes, the time t generated by y is the fault occurrence time, and the early warning of the fault rule (or the deviation from the normal model) is realized. If y is a monotonic function, then the method can be simplified to determine if the parameter value exceeds the valid range y0>a (or y0<a) Can be obtained by inverting the function F^-1Inverse solution time ta ═ F^-1(a) Thereby, the expected occurrence time ta of the fault is given, and the fault is predicted.

In a specific application example, the specific application content of the spacecraft fault active detection method is as follows:

(1) the method comprises the steps of firstly preprocessing the telemetering data of the spacecraft, firstly cleaning the telemetering historical data, interpolating missing values, eliminating abnormal data, converting the missing values into a standard format which can be input by a model, sending results to a real-time detection and database, and storing the preprocessed data in the database.

(2) The off-line training link can be set once every two weeks or adjusted according to the actual running condition of the satellite. At the moment, feature extraction work is carried out on the standardized historical data, and the original data is output in combination with the category feature, the correlation feature and the working condition feature. And respectively establishing models suitable for single-parameter and multi-parameter anomaly detection, and outputting optimal model parameters and anomaly threshold values after verification and parameter adjustment. And (4) sending the original data and the extracted features into a model, adjusting parameters, verifying and outputting optimal model parameters. And saving the model to a model library.

(3) And respectively establishing a parameter prediction model and a parameter normal trend model according to the historical task data of the key equipment. And if the current input data is judged to be abnormal by the abnormality detection module, removing the abnormal values, supplementing by adopting an interpolation method, and then carrying out extrapolation on the normal trend. The input to the parametric prediction model is the current actual value, and is retained even if there is an anomaly. And accumulating the real-time data into a sliding window, sending the real-time data into a prediction model, and predicting the future state of the spacecraft by taking the current spacecraft use state as a starting point.

(4) Unless an extreme condition is met, one point cannot be detected to be abnormal, so that real-time telemetry data is accumulated into a sliding window of about ten minutes, and the size of the window can be adjusted according to an actual scene. The real-time data is standardized by the data processing module. And then, sequencing the data according to the satellite time to form positive sequence real-time data. And matching the characteristics of the sorted real-time data, and sending the matched characteristics to an anomaly detection model for detection. In the abnormality detection stage, an abnormality score is given, and whether a potential abnormality exists is determined according to the difference between the abnormality score and an abnormality threshold value. Setting detection sensitivity, such as setting abnormality to 1, 3, 5, etc. points in succession; and then carrying out abnormity early warning.

As can be seen from the above description, the method and apparatus for actively detecting a spacecraft fault provided by the application example of the present application can preprocess space data, including: picking out outliers in the data; filling the vacancy value; data were normalized to max-min. And learning and training through historical normal data to obtain a parameter change model. The correlation analysis module divides the telemetry parameters according to the subsystems and then mines the correlation of the parameters in each subsystem. The fault prediction function module performs time extrapolation according to the training model y, F (t, A) and historical data, calculates whether a parameter value y0 in the process of reaching an extrapolation point t0 meets a fault state (if the parameter value y exceeds an effective range, meets a fault rule, or has a deviation from a normal model exceeding a threshold value and the like), and further enables the spacecraft fault active detection method and device provided by the application to use an LSTM algorithm to perform learning training of normal data, identify the normal state characteristics of a spacecraft, establish a normal state model and then perform fault detection in real time. And obtaining a coefficient in multi-parameter fusion through the correlation analysis of the data in a time domain or a frequency domain. Impending failures can be efficiently predicted.

In terms of hardware, the present application provides an embodiment of an electronic device for implementing all or part of the contents in the active detection method for spacecraft fault, where the electronic device specifically includes the following contents:

fig. 16 is a schematic block diagram of a device configuration of an electronic device 9600 according to an embodiment of the present application. As shown in fig. 16, the electronic device 9600 can include a central processor 9100 and a memory 9140; the memory 9140 is coupled to the central processor 9100. Notably, this fig. 16 is exemplary; other types of structures may also be used in addition to or in place of the structure to implement telecommunications or other functions.

In an embodiment, the spacecraft fault active detection functionality may be integrated into the central processor. Wherein the central processor may be configured to control:

From the above description, the electronic device provided in the embodiment of the application can maximally utilize the downlink normal data and learn the correct change model of the data in the absence of the spacecraft fault sample, can actively detect the spacecraft fault, can effectively improve the real-time performance, the automation degree and the efficiency of spacecraft fault detection, and can effectively improve the prediction accuracy and the reliability of the spacecraft fault.

In another embodiment, the active detection device for spacecraft fault may be configured separately from the central processor 9100, for example, the active detection device for spacecraft fault may be configured as a chip connected to the central processor 9100, and the active detection function for spacecraft fault is realized by the control of the central processor.

As shown in fig. 16, the electronic device 9600 may further include: a communication module 9110, an input unit 9120, an audio processor 9130, a display 9160, and a power supply 9170. It is noted that the electronic device 9600 also does not necessarily include all of the components shown in fig. 16; further, the electronic device 9600 may further include components not shown in fig. 16, which can be referred to in the related art.

As shown in fig. 16, a central processor 9100, sometimes referred to as a controller or operational control, can include a microprocessor or other processor device and/or logic device, which central processor 9100 receives input and controls the operation of the various components of the electronic device 9600.

The memory 9140 can be, for example, one or more of a buffer, a flash memory, a hard drive, a removable media, a volatile memory, a non-volatile memory, or other suitable device. The information relating to the failure may be stored, and a program for executing the information may be stored. And the central processing unit 9100 can execute the program stored in the memory 9140 to realize information storage or processing, or the like.

The input unit 9120 provides input to the central processor 9100. The input unit 9120 is, for example, a key or a touch input device. Power supply 9170 is used to provide power to electronic device 9600. The display 9160 is used for displaying display objects such as images and characters. The display may be, for example, an LCD display, but is not limited thereto.

The memory 9140 can be a solid state memory, e.g., Read Only Memory (ROM), Random Access Memory (RAM), a SIM card, or the like. There may also be a memory that holds information even when power is off, can be selectively erased, and is provided with more data, an example of which is sometimes called an EPROM or the like. The memory 9140 could also be some other type of device. Memory 9140 includes a buffer memory 9141 (sometimes referred to as a buffer). The memory 9140 may include an application/function storage portion 9142, the application/function storage portion 9142 being used for storing application programs and function programs or for executing a flow of operations of the electronic device 9600 by the central processor 9100.

The memory 9140 can also include a data store 9143, the data store 9143 being used to store data, such as contacts, digital data, pictures, sounds, and/or any other data used by an electronic device. The driver storage portion 9144 of the memory 9140 may include various drivers for the electronic device for communication functions and/or for performing other functions of the electronic device (e.g., messaging applications, contact book applications, etc.).

The communication module 9110 is a transmitter/receiver 9110 that transmits and receives signals via an antenna 9111. The communication module (transmitter/receiver) 9110 is coupled to the central processor 9100 to provide input signals and receive output signals, which may be the same as in the case of a conventional mobile communication terminal.

Based on different communication technologies, a plurality of communication modules 9110, such as a cellular network module, a bluetooth module, and/or a wireless local area network module, may be provided in the same electronic device. The communication module (transmitter/receiver) 9110 is also coupled to a speaker 9131 and a microphone 9132 via an audio processor 9130 to provide audio output via the speaker 9131 and receive audio input from the microphone 9132, thereby implementing ordinary telecommunications functions. The audio processor 9130 may include any suitable buffers, decoders, amplifiers and so forth. In addition, the audio processor 9130 is also coupled to the central processor 9100, thereby enabling recording locally through the microphone 9132 and enabling locally stored sounds to be played through the speaker 9131.

An embodiment of the present application further provides a computer-readable storage medium capable of implementing all the steps in the spacecraft failure active detection method in the foregoing embodiment, where the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the computer program implements all the steps of the spacecraft failure active detection method in the foregoing embodiment, where an execution subject of the computer program is a server or a client, for example, when the processor executes the computer program, the processor implements the following steps:

As can be seen from the above description, the computer-usable storage medium provided in the embodiment of the present application can maximally utilize downlink normal data and learn a correct variation model of data in the absence of a spacecraft fault sample, can actively detect a spacecraft fault, can effectively improve the real-time performance, the automation degree and the efficiency of spacecraft fault detection, and can effectively improve the accuracy and the reliability of prediction of a spacecraft fault.

As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, apparatus, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (devices), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

The principle and the implementation mode of the invention are explained by applying specific embodiments in the invention, and the description of the embodiments is only used for helping to understand the method and the core idea of the invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.

Claims

1. An active detection method for spacecraft fault is characterized by comprising the following steps:

2. A spacecraft fault active detection method according to claim 1, wherein the parametric normal trend model comprises: a single parameter model;

3. A spacecraft fault active detection method according to claim 2, wherein the parametric normal trend model further comprises: a correlation model;

4. A spacecraft fault active detection method according to claim 3, wherein before obtaining the abnormal score corresponding to the normal mode expected data corresponding to each telemetry parameter of the spacecraft at the current time, the method further comprises:

5. The active spacecraft fault detection method according to claim 3, wherein the applying a preset data correlation analysis mode to perform correlation processing on the standardized data to obtain correlation relations among the remote sensing parameters comprises:

6. A spacecraft fault active detection method according to claim 4, further comprising:

7. The active spacecraft fault detection method according to claim 1, wherein the step of obtaining abnormal scores corresponding to normal mode expected data respectively corresponding to each telemetry parameter of the current time of the spacecraft by applying a preset parameter normal trend model and the current actual telemetry data of the spacecraft comprises the steps of:

8. A method as claimed in claim 2, wherein said preprocessing of said normal historical telemetry data comprises:

9. An active detection device for spacecraft fault, comprising:

10. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method for active detection of spacecraft fault according to any of claims 1 to 8 when executing the program.