Minor fault diagnosis method and system based on fault trend ratio and feature selection
By eliminating trend items and selecting feature of the monitoring data of the spacecraft control system, efficient detection and separation of micro faults is achieved, the problem of difficulty in detecting micro faults is solved, and the reliability and life of the spacecraft is improved.
Patent Information
- Application Number
- CN202010547795.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2020-06-16
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2040-06-16
AI Technical Summary
The prior art is difficult to effectively detect and separate tiny failures in spacecraft, resulting in failure events and affecting the reliability and life of the spacecraft.
Monitoring data is obtained by analyzing the kinematics and dynamics of the spacecraft control system, trend terms of normal data are eliminated, feature selection is used using asymmetric importance factors, and suboptimal visual mapping is obtained to separate fault data.
It improves the detection rate and separation rate of micro faults, provides the best view of fault information, enhances the credibility of micro fault diagnosis results, improves the spacecraft fault library, and improves the on-orbit fault diagnosis and processing capabilities.
Smart Images

Figure CN111783286B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of spacecraft data processing, and in particular to a method and system for diagnosing minor faults based on fault trend ratio and feature selection. Background Art
[0002] Significant faults often evolve from early minor faults. If minor faults can be detected and isolated in time within the controllable range of system operation, failure events will be effectively avoided. Therefore, the diagnosis of minor faults plays an important role in achieving high reliability and long life of the entire satellite. Summary of the invention
[0003] The embodiments of the present invention provide a method and system for diagnosing minor faults based on fault trend ratio and feature selection, which improves the performance of minor fault diagnosis based on fault trend ratio and feature selection by eliminating normal data and performing suboptimal visualization mapping of data feature selection. This is of great significance to improving the reliability of on-orbit spacecraft and extending the life of spacecraft.
[0004] To achieve the above object, on the one hand, an embodiment of the present invention provides a method for diagnosing minor faults based on fault trend ratio and feature selection, the method comprising:
[0005] Obtaining monitoring data of the control system by kinematic and dynamic analysis of the spacecraft control system;
[0006] Eliminating trend items of normal data in the monitoring data according to prior knowledge;
[0007] Based on the asymmetric importance factors between different faults, feature selection is performed on the monitoring data with normal data trend items removed to obtain the corresponding suboptimal visualization mapping;
[0008] Faulty data in the monitoring data is separated according to the suboptimal visualization map.
[0009] On the other hand, an embodiment of the present invention provides a minor fault diagnosis system based on fault trend ratio and feature selection, the system comprising:
[0010] A data acquisition unit, used for obtaining monitoring data of the control system by kinematic and dynamic analysis of the control system of the spacecraft;
[0011] A data elimination unit, used for eliminating trend items of normal data in the monitoring data according to prior knowledge;
[0012] A mapping unit, used for performing feature selection on the monitoring data from which normal data trend items are removed based on asymmetric importance factors between different faults to obtain corresponding suboptimal visualization mapping;
[0013] A fault separation unit is used to separate fault data in the monitoring data according to the suboptimal visualization mapping.
[0014] The above technical solution has the following beneficial effects:
[0015] The technical solution of the present invention utilizes a minor fault detection method based on elimination of normal data and a minor fault separation method based on visual mapping technology to overcome the difficulties in minor fault detection, improve the detection rate of minor faults, provide an optimal fault information view, improve the fault separation rate, enhance the credibility of minor fault diagnosis results, improve the minor fault diagnosis theory, and complete the fault library of on-orbit spacecraft control systems, providing theoretical and technical support for improving the on-orbit fault diagnosis and processing capabilities of spacecraft. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0017] Figure 1 is a flow chart of a minor fault diagnosis method based on fault trend ratio and feature selection according to an embodiment of the present invention;
[0018] Figure 2 This is a schematic diagram of minor fault data in an embodiment of the present invention;
[0019] Figure 3 It is a schematic diagram of the structure of a minor fault diagnosis system based on fault trend ratio and feature selection according to an embodiment of the present invention. DETAILED DESCRIPTION
[0020] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0021] like Figure 1 FIG. 1 is a flow chart of a method for diagnosing minor faults based on fault trend ratio and feature selection according to an embodiment of the present invention, wherein the method comprises:
[0022] S101: Obtain monitoring data of the control system by kinematic and dynamic analysis of the spacecraft control system; preferably, the monitoring data includes operation mechanism data, model structure data and model parameter data of the control system.
[0023] Through mechanism analysis and state acquisition, the information source of subsequent minor fault diagnosis methods is obtained. The information source of typical cases on orbit mainly includes two parts, namely mechanism model and monitoring data. Through the kinematic and dynamic analysis of the spacecraft control system, the operation mechanism, model structure and model parameters of the control system can be obtained. This information is not only the model information source of the minor fault diagnosis method, but also the simulation basis of Matlab's Simulink control and S function module. The computer can store actuator monitoring data (momentum wheel / thruster / sailboard drive mechanism, etc.) and sensor monitoring data (gyroscope / infrared sensor / star sensor / sun sensor, etc.) in real time. This information is not only the data information source of the minor fault diagnosis method, but also the entry information of the Qt Widgets form application.
[0024] S102: Eliminate trend items of normal data in the monitoring data according to prior knowledge.
[0025] like Figure 2 As shown in Figure 1, the monitoring data y of the system can be decomposed into three parts: noise e, trend t and fault f. The trend term is relatively smooth, but affected by noise, the data fluctuates between the upper and lower envelopes. If the trend is non-stationary and the variance of the noise is large, then minor faults are easily masked by the trend and noise. Figure 2 In the monitoring data between the beginning and the end of the fault, there is a minor fault, but it does not significantly break through the upper and lower envelopes, so it is difficult to visually detect the changes caused by the fault. The monitoring data of the gyroscope under the minor drift fault of the spacecraft control system has a similar phenomenon. One of the core issues of minor fault detection is to weaken the "trend drowning fault effect", that is, to improve the fault trend ratio FTR (Fault-trend ratio). If R(t) represents the expected sphere radius of the trend space, then the fault trend ratio is FTR = ||f|| / R(t). The R(t) of the non-stationary system is very large, and the normal trend needs to be extracted and eliminated. If I represents the unit matrix, Ψ(m,p) represents the m-order design operator that describes the trend, and p is the delay factor, for example, Φ(2,p) = p*p+p+1, then Φ(2,p)f(k) = f(k-2)+f(k-1)+f(k); the signal after trend removal is recorded as (I-Ψ(m,p))y, then the designed FTR is
[0026] FTR(Ψ,m)=||f|| / ||(I-Ψ(m,p))y|| (1)
[0027] Whether it is data characteristics such as monotonicity, periodicity, and inflection points, or the system structure, model parameters, and operating mechanism of the monitored system, domain experts have corresponding prior knowledge. Therefore, these prior knowledge can be used to extract the trend term Ψ(m,p)y of normal data. When most of the trend terms are removed from the monitoring data, the fault trend ratio can be improved, thereby weakening or even eliminating the "trend-overwhelmed fault effect", and ultimately improving the fault detection rate and reducing the fault detection delay time. Existing studies have shown that if the monitored object satisfies the linear structure, then conventional multivariate statistical dimensionality reduction methods and dynamic dimensionality reduction methods, such as principal component analysis and dynamic principal component analysis, can significantly weaken the "trend-overwhelmed fault effect". However, the spacecraft control system is a typical nonlinear system, and the conventional trend removal method is not effective. At this time, the prior knowledge of the control system model structure is particularly important. If used represents the m-order trend operator family under the constraints of system structure, model parameters, operation mechanism and prior knowledge, then the above research route can be described as
[0028]
[0029] Preferably, by analyzing the measurement and control mechanism, transfer characteristics and trend characteristics of the spacecraft control system; the trend items of normal data in the monitoring data are eliminated based on prior knowledge, including: using system modeling means and parameter estimation means to extract nonlinear and non-stationary trend data of the monitoring data; eliminating the nonlinear and non-stationary trend data of the monitoring data.
[0030] S103: performing feature selection on the monitoring data from which normal data trend items are removed based on asymmetric importance factors between different faults to obtain corresponding suboptimal visualization mapping;
[0031] After dimension reduction, the separation information of any two faults among q faults is Iso(r i ,r j )=||Mr i -sign(i,j)Mr j || 2 (i,j=1,…,q), where if the direction angle is acute, then sign(i,j)=1, otherwise sign(i,j)=-1. Symmetric optimal visualization mapping assumes that the importance of different types of faults is the same, so the feature extraction method can be simply transformed into a principal component analysis process. However, in spacecraft control systems, the importance of different fault separation information should actually be treated differently. For example, if the angle between two feature directions is very small, a slight loss of separation information may make it impossible to separate the two faults; on the contrary, if the angle between the two feature directions is large, a slight loss of separation information will not have much effect on the fault separation result.
[0032] Find the suboptimal visualization mapping based on feature selection. The more separation information a feature has, the more it should be retained. The process is refined into three steps: First, determine the required feature set, such as retaining features with large amplitude changes and prone to failure. Second, determine the must-have feature set, such as eliminating a small number of features with small amplitude changes and not prone to failure. Third, given the correlation measure between the feature and the separation information, the optimization criterion is obtained, and the feature subset with the most separation information is searched in the alternative feature set.
[0033] Preferably, the feature selection of the monitoring data excluding normal data trend items based on the asymmetric importance factors between different faults to obtain the corresponding suboptimal visualization mapping is implemented by the following formula:
[0034]
[0035] Among them, w ij (i,j=1,…,q) is the asymmetric importance factor, q is the number of faults after dimensionality reduction;
[0036] Iso(r i ,r j ) is the characteristic vector r of any two faults i ,r j The separation information is expressed as Iso(r i ,r j )=[Mr i -sign(i,j)Mr j || 2 (i,j=1,…,q), if the characteristic vectors of two faults r i ,r j If the direction angle is acute, then sign(i,j)=1, otherwise sign(i,j)=-1;
[0037] M represents the feature selection sparse matrix and I is the identity matrix.
[0038] S104: Separating fault data from the monitoring data according to the suboptimal visualization mapping.
[0039] Further preferably, for the separation of fault noise in this scheme, a test toolbox is set up, and the toolbox is divided into three layers. Input layer: Simulink controls and S-function modules are used to simulate the satellite attitude control system, and the QFileDialog library of the QtWidgets designer is used as the data import port of the satellite attitude control system. Algorithm layer: M files are used to implement micro-fault amplification detection algorithms, micro-fault visualization separation algorithms, data graphics export functions, and so on. Output layer: The toolbox GUI interface is completed with the Qt designer. Use simulation data and real satellite control system monitoring data to test the availability, computational efficiency, and diagnostic performance of the toolbox and micro-fault diagnosis method. The test feedback results can be used as a basis for improving micro-fault diagnosis methods.
[0040] Corresponding to the above method, such as Figure 3 FIG. 1 is a schematic diagram of a structure of a minor fault diagnosis system based on fault trend ratio and feature selection according to an embodiment of the present invention. The system includes:
[0041] A data acquisition unit 21, used for obtaining monitoring data of the control system by kinematic and dynamic analysis of the spacecraft control system;
[0042] A data elimination unit 22, configured to eliminate trend items of normal data in the monitoring data according to prior knowledge;
[0043] A mapping unit 23, configured to perform feature selection on the monitoring data from which normal data trend items are removed based on asymmetric importance factors between different faults to obtain a corresponding suboptimal visualization mapping;
[0044] The fault separation unit 24 is configured to separate fault data in the monitoring data according to the suboptimal visualization mapping.
[0045] Preferably, the monitoring data includes operation mechanism data, model structure data and model parameter data of the control system.
[0046] Preferably, the data elimination unit 23 is specifically used for:
[0047] Extracting data of nonlinear and non-stationary trends of the monitoring data by using system modeling means and parameter estimation means;
[0048] Eliminate the data with nonlinear and non-stationary trends in the monitoring data.
[0049] Preferably, the system modeling means include hybrid polynomial approximation, high-dimensional spline fitting, time series modeling and multi-frequency signal decomposition.
[0050] Preferably, the mapping unit 23 implements suboptimal visual mapping of monitoring data by the following formula:
[0051]
[0052] Among them, w ij (i,j=1,…,p) is the asymmetric importance factor, and p is the number of faults after dimensionality reduction;
[0053] Iso(r i ,r j ) is the characteristic vector r of any two faults i ,r j The separation information is expressed as Iso(r i ,r j )=||Mr i -sign(i,j)Mr j || 2 (i,j=1,…,p), if the characteristic vectors of two faults r i ,r j If the direction angle is acute, then sign(i,j)=1, otherwise sign(i,j)=-1;
[0054] M represents the feature selection sparse matrix and I is the identity matrix.
[0055] It should be understood that the specific order or hierarchy of steps in the disclosed process is an example of an exemplary method. Based on design preferences, it should be understood that the specific order or hierarchy of steps in the process can be rearranged without departing from the scope of protection of the present disclosure. The attached method claims present the elements of the various steps in an exemplary order and are not intended to be limited to the specific order or hierarchy described.
[0056] In the above detailed description, various features are grouped together in a single embodiment to simplify the disclosure. This method of disclosure should not be interpreted as reflecting an intention that the embodiments of the claimed subject matter require more features than are clearly stated in each claim. On the contrary, as reflected in the appended claims, the invention is in a state of having less than all the features of the disclosed individual embodiments. Therefore, the appended claims are hereby expressly incorporated into the detailed description, with each claim standing on its own as a separate preferred embodiment of the invention.
[0057] The disclosed embodiments are described above to enable any person skilled in the art to implement or use the present invention. Various modifications of these embodiments are obvious to those skilled in the art, and the general principles defined herein may also be applied to other embodiments without departing from the spirit and scope of the present disclosure. Therefore, the present disclosure is not limited to the embodiments given herein, but is consistent with the broadest scope of the principles and novel features disclosed in this application.
[0058] The above description includes examples of one or more embodiments. Of course, it is impossible to describe all possible combinations of components or methods for the purpose of describing the above embodiments, but it should be recognized by those skilled in the art that the various embodiments may be further combined and arranged. Therefore, the embodiments described herein are intended to cover all such changes, modifications and variations that fall within the scope of protection of the appended claims. In addition, with respect to the term "comprising" used in the specification or claims, the word is covered in a manner similar to the term "including", just as "including," is explained as a transitional word in the claims. In addition, any term "or" used in the specification of the claims is intended to mean "non-exclusive or".
[0059] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.
Claims
1. A method for diagnosing minor faults based on fault trend ratio and feature selection, characterized in that: include: Obtaining monitoring data of the control system by kinematic and dynamic analysis of the spacecraft control system; Eliminating trend items of normal data in the monitoring data according to prior knowledge; Based on the asymmetric importance factors between different faults, feature selection is performed on the monitoring data with normal data trend items removed to obtain the corresponding suboptimal visualization mapping; separating faulty data in the monitoring data according to the suboptimal visualization mapping; Wherein, the monitoring data includes the operation mechanism data, model structure data and model parameter data of the control system; The removing of trend items of normal data in the monitoring data according to prior knowledge includes: Extracting data of nonlinear and non-stationary trends of the monitoring data by using system modeling means and parameter estimation means; Eliminate data with nonlinear and non-stationary trends in the monitoring data; The system modeling means include mixed polynomial approximation, high-dimensional spline fitting, time series modeling and multi-frequency signal decomposition; The feature selection of the monitoring data excluding normal data trend items based on the asymmetric importance factors between different faults to obtain the corresponding suboptimal visualization mapping is implemented by the following formula: Among them, w ij (i,j=1,…,q) is the asymmetric importance factor, q is the number of faults after dimensionality reduction; Iso(r i ,r j ) is the characteristic vector r of any two faults i ,r j The separation information is expressed as Iso(r i ,r j )=||Mr i -sign(i,j)Mr j || 2 (i,j=1,…,q), if the characteristic vectors of two faults r i ,r j If the direction angle is acute, then sign(i,j)=1, otherwise sign(i,j)=-1; M represents the feature selection sparse matrix and I is the identity matrix.
2. A minor fault diagnosis system based on fault trend ratio and feature selection, characterized in that: include: A data acquisition unit, used for obtaining monitoring data of the control system by kinematic and dynamic analysis of the control system of the spacecraft; A data elimination unit, used for eliminating trend items of normal data in the monitoring data according to prior knowledge; A mapping unit, used for performing feature selection on the monitoring data from which normal data trend items are removed based on asymmetric importance factors between different faults to obtain corresponding suboptimal visualization mapping; a fault separation unit, configured to separate fault data in the monitoring data according to the suboptimal visualization mapping; Wherein, the monitoring data includes the operation mechanism data, model structure data and model parameter data of the control system; The data elimination unit is specifically used for: Extracting data of nonlinear and non-stationary trends of the monitoring data by using system modeling means and parameter estimation means; Eliminate data with nonlinear and non-stationary trends in the monitoring data; The system modeling means include mixed polynomial approximation, high-dimensional spline fitting, time series modeling and multi-frequency signal decomposition; The mapping unit implements suboptimal visual mapping of monitoring data through the following formula: Among them, w ij (i,j=1,…,q) is the asymmetric importance factor, q is the number of faults after dimensionality reduction; Iso(r i ,r j ) is the characteristic vector r of any two faults i ,r j The separation information is expressed as Iso(r i ,r j )=||Mr i -sign(i,j)Mr j || 2 (i,j=1,…,q), if the characteristic vectors of two faults r i ,r j If the direction angle is acute, then sign(i,j)=1, otherwise sign(i,j)=-1; M represents the feature selection sparse matrix and I is the identity matrix.
