Equipment fault diagnosis method based on avionics bus data

Abstract

The invention discloses an avionics bus data-based equipment fault diagnosis method, which comprises the following steps of S1, collecting M types of avionics bus data in a work period, and constructing an avionics bus data matrix; s2, giving K working environments of the target equipment; s3, for the fault-free target equipment, constructing an avionics bus data matrix in a working environment, and calculating a reference avionics bus data matrix; s4, obtaining an avionics data matrix of the target equipment of various fault types in each working environment; s5, constructing a sample set corresponding to the fault-free target equipment and the target equipment of multiple fault types; s6, adopting a machine learning method to construct a classifier model, and training to obtain a maturely trained classifier; and S7, performing fault diagnosis on the to-be-detected target equipment based on the classifier which is maturely trained, and judging whether the to-be-detected target equipment has a fault or not. Based on the avionics bus data of each period, accurate fault diagnosis can be realized.

Description

Technical Field

[0001] The invention relates to equipment fault diagnosis, and in particular to an equipment fault diagnosis method based on avionics bus data. Background Art

[0002] Avionics bus technology is a standard for data transmission between electronic systems of aircraft and other aircraft, which allows different electronic devices to communicate and exchange data with each other. This technology is crucial for modern aircraft because it facilitates information sharing and coordination between various subsystems, thereby improving the efficiency and reliability of the entire system;

[0003] In an avionics system, there are often a variety of different avionics equipment. When testing them, professional maintenance personnel are often required to perform testing and maintenance in sequence. During regular maintenance, this type of testing and maintenance is the most comprehensive. However, in actual practical use, it is often difficult to perform fault monitoring in real time. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and provide an equipment fault diagnosis method based on avionics bus data, which can achieve accurate fault diagnosis based on the avionics bus data of each cycle and has stronger real-time performance.

[0005] The object of the present invention is achieved through the following technical solution: A device fault diagnosis method based on avionics bus data, comprising the following steps:

[0006] S1. Given M types of avionics bus data related to a target device, collect the M types of avionics bus data within a working cycle, and construct an avionics bus data matrix within a working cycle;

[0007] S2. K working environments of the given target device;

[0008] S3. For the target device without faults, repeat step S1 under K kinds of working environments respectively, obtain the avionics bus data matrix under K kinds of working environments, and take the average as the reference avionics bus data matrix;

[0009] S4. For target devices of various fault types, repeat step S1 for each target device of each fault type under K working environments to obtain an avionics data matrix of target devices of various fault types under each working environment;

[0010] S5. Construct sample sets corresponding to target devices without faults and target devices with various fault types;

[0011] S6. Use machine learning methods to build a classifier model, and use the sample set to train the classifier to obtain a mature classifier;

[0012] S7. Perform fault diagnosis on the target device to be detected based on the maturely trained classifier to determine whether the target device to be detected has a fault.

[0013] The beneficial effects of the present invention are as follows: the present invention can construct sample features of fault-free target equipment and target equipment of known fault types through the avionics bus data of each cycle, so that the training classifier can realize accurate fault diagnosis with stronger real-time performance; at the same time, the present invention takes into account the conditions of the equipment in different working environments, and the trained classifier can complete fault diagnosis in various working environments more accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0015] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings, but the protection scope of the present invention is not limited to the following.

[0016] like Figure 1 As shown, a device fault diagnosis method based on avionics bus data includes the following steps:

[0017] S1. Given M types of avionics bus data related to a target device, collect the M types of avionics bus data within a working cycle, and construct an avionics bus data matrix within a working cycle;

[0018] The step S1 comprises:

[0019] S101. Within a working cycle T, M types of avionics bus data related to the target device are collected N times at a fixed sampling interval; let the Mth type of avionics bus data obtained by the nth collection be recorded as x nm , n=1,2,...,N; m=1,2,...,M;

[0020] Then the avionics bus data matrix within a working cycle is recorded as X:

[0021]

[0022] S2. K working environments of the given target device;

[0023] In step S2, the working environment refers to the electromagnetic environment where the target device works.

[0024] S3. For the target device without faults, repeat step S1 under K kinds of working environments respectively, obtain the avionics bus data matrix under K kinds of working environments, and take the average as the reference avionics bus data matrix;

[0025] In step S3, the avionics bus data matrix under K working environments is recorded as X 01 ,X 02 ,...,X 0K , the reference avionics bus data matrix is ​​recorded as

[0026]

[0027] S4. For target devices of various fault types, repeat step S1 for each target device of each fault type under K working environments to obtain an avionics data matrix of target devices of various fault types under each working environment;

[0028] In step S4, it is assumed that there are a total of P types of target devices with faults, wherein the avionics bus data matrix of the target device with the pth type of fault under K working environments is recorded as X p1 ,X p2 ,...,X pK ,p=1,2,...,P.

[0029] S5. Construct sample sets corresponding to target devices without faults and target devices with various fault types;

[0030] The step S5 comprises:

[0031] S501. For the target device without fault, the avionics bus data matrix X under K working environments is 01 ,X 02 ,...,X 0K , minus the baseline avionics bus data matrix The sample features of K fault-free target devices are obtained, recorded as:

[0032]

[0033] For each sample feature of a non-faulty target device, add label 0;

[0034] S502. For the target device of the pth fault type:

[0035] The avionics bus data matrix X under K working environments p1 ,X p2 ,...,X pK , minus the baseline avionics bus data matrix The sample features of K fault-free target devices are obtained, recorded as:

[0036]

[0037] For each sample feature of the target device with the pth fault type, add label p;

[0038] S503. When p=1, 2, ..., P, repeat step S502 to obtain sample features and labels of target devices of each fault type;

[0039] S504. Divide the results obtained from S501 to S504 and add them to the same set to form a sample set.

[0040] S6. Use machine learning methods to build a classifier model, and use the sample set to train the classifier to obtain a mature classifier;

[0041] The machine learning method includes a KNN algorithm, a BP neural network algorithm or an AdaBoost classifier method.

[0042] In step S6, when the classifier is trained using the sample set, the sample features in the training set are used as the classifier input, and the labels corresponding to the sample features are used as the expected output. When all sample features are trained, it is considered that a mature classifier model is obtained.

[0043] S7. Perform fault diagnosis on the target device to be detected based on the maturely trained classifier to determine whether the target device to be detected has a fault.

[0044] In step S7, when the target device to be detected is in working state, the avionics bus data matrix of the target device to be detected within one working cycle is constructed according to step S101, and the reference avionics bus data matrix is ​​subtracted and recorded as X to obtain the sample features of the target device to be detected, which are then sent to a maturely trained classifier model. If the classifier model outputs 0, it is considered that the device to be detected has no fault. If the classifier model outputs p, p=1, 2, ..., P, it is considered that the device to be detected has the pth fault.

[0045] The above is a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in other combinations, modifications and environments, and can be modified within the scope of the concept described herein through the above teachings or the technology or knowledge of the relevant field. The changes and modifications made by those skilled in the art do not depart from the spirit and scope of the present invention, and should be within the scope of protection of the claims attached to the present invention.

Claims

1. A device fault diagnosis method based on avionics bus data, characterized in that: The following steps are involved: S1. Given M types of avionics bus data related to a target device, collect the M types of avionics bus data within a working cycle, and construct an avionics bus data matrix within a working cycle; S2. K working environments of the given target device; S3. For the target device without faults, repeat step S1 under K working environments respectively to obtain the avionics bus data matrix under K working environments, and take the average as the reference avionics bus data matrix; S4. For target devices of various fault types, repeat step S1 for each target device of each fault type under K working environments to obtain an avionics data matrix of target devices of various fault types under each working environment; S5. Construct sample sets corresponding to target devices without faults and target devices with various fault types; S6. Use machine learning methods to build a classifier model, and use the sample set to train the classifier to obtain a mature classifier; S7. Perform fault diagnosis on the target device to be detected based on the maturely trained classifier to determine whether the target device to be detected has a fault.

2. The device fault diagnosis method based on avionics bus data according to claim 1 is characterized in that: The step S1 comprises: S101. Within a working cycle T, M types of avionics bus data related to the target device are collected N times at a fixed sampling interval; let the Mth type of avionics bus data obtained by the nth collection be recorded as x nm , n=1,2,...,N; m=1,2,...,M; Then the avionics bus data matrix within a working cycle is recorded as X:

3. The device fault diagnosis method based on avionics bus data according to claim 1 is characterized in that: In step S2, the working environment refers to the electromagnetic environment where the target device works.

4. The device fault diagnosis method based on avionics bus data according to claim 1 is characterized in that: In step S3, the avionics bus data matrix under K working environments is recorded as X 01 ,X 02 ,...,X 0K , the reference avionics bus data matrix is ​​recorded as X:

5. The device fault diagnosis method based on avionics bus data according to claim 1 is characterized in that: In step S4, it is assumed that there are a total of P types of target devices with faults, wherein the avionics bus data matrix of the target device with the pth type of fault under K working environments is recorded as X p1 ,X p2 ,...,X pK ,p=1,2,...,P.

6. The device fault diagnosis method based on avionics bus data according to claim 1 is characterized by: The step S5 comprises: S501. For the target device without fault, the avionics bus data matrix X under K working environments is 01 ,X 02 ,...,X 0K , minus the baseline avionics bus data matrix The sample features of K fault-free target devices are obtained, recorded as: For each sample feature of a non-faulty target device, add label 0; S502. For the target device of the pth fault type: The avionics bus data matrix X under K working environments p1 ,X p2 ,...,X pK , minus the reference avionics bus data matrix X, and obtain the sample features of K fault-free target devices, recorded as: For each sample feature of the target device with the pth fault type, add label p; S503. When p=1, 2, ..., P, repeat step S502 to obtain sample features and labels of target devices of each fault type; S504. Divide the results obtained from S501 to S504 and add them to the same set to form a sample set.

7. The device fault diagnosis method based on avionics bus data according to claim 1 is characterized by: The machine learning method includes a KNN algorithm, a BP neural network algorithm or an AdaBoost classifier method.

8. The device fault diagnosis method based on avionics bus data according to claim 1 is characterized by: In step S6, when the classifier is trained using the sample set, the sample features in the training set are used as the classifier input, and the labels corresponding to the sample features are used as the expected output. When all sample features are trained, it is considered that a mature classifier model is obtained.

9. The device fault diagnosis method based on avionics bus data according to claim 1 is characterized by: In step S7, when the target device to be detected is in working state, the avionics bus data matrix within one working cycle of the target device to be detected is constructed according to step S101, and the reference avionics bus data matrix is ​​subtracted and recorded as The sample features of the target device to be tested are obtained and then sent to a maturely trained classifier model. If the classifier model outputs 0, it is considered that the device to be tested has no fault. If the classifier model outputs p, p = 1, 2, ..., P, it is considered that the device to be tested has the pth fault.

Images