A multi-source fault diagnosis and identification method for complex avionics equipment

Abstract

A multi-source fault diagnosis and identification method for complex avionics equipment based on a multi-source fault diagnosis test device for complex avionics equipment, comprising: A1) performing information preprocessing; A2) using a time domain (206) and time-frequency (207) analysis method to construct different state feature vector spaces and corresponding state type spaces; A3) performing RVM-based single information source state identification; and A4) performing state attribute decision fusion (210) based on DSmT. Through the method, a connection type structure feature representation technology based on feature index parameter measurement and analysis is established, providing a bridge and link for the research and determination of the mapping relationship between fault sensitive structure features and fault detection signal responses.

Description

Technical Field

[0001] This invention relates to a method for diagnosing and identifying multi-source faults in complex avionics equipment. Background Technology

[0002] System health status assessment is the core of aircraft health management strategy. By assessing the current health status of complex systems, the causes and sources of failures can be effectively identified, leading to a series of maintenance and support recommendations and decisions. Aircraft health management strategies are widely recognized and applied in the aerospace industry, providing significant assurance for the safety and reliability of spacecraft and becoming an inevitable trend in the development of the aerospace industry.

[0003] However, traditional machine learning algorithms in system health management and state discrimination lack the ability to extract features from high-dimensional data, which limits their classification accuracy, classification speed, and other performance indicators in system health management and state discrimination problems. Summary of the Invention

[0004] To address the limitations of existing multi-source fault diagnosis methods for complex avionics equipment, this invention provides a method for multi-source fault diagnosis and identification of complex avionics equipment. This method effectively solves problems such as shallow feature extraction, gradient vanishing, and single feature scale, and significantly improves the accuracy of system health status classification and identification.

[0005] This invention establishes a connected structural feature characterization technology based on the measurement and analysis of characteristic index parameters of typical faulty components and their response characteristics and laws under different states (environment, load), providing a bridge and link for the study and determination of the mapping relationship between fault-sensitive structural features and fault detection signal response.

[0006] The main manifestation of intermittent connection failure is an increase in contact resistance exceeding a certain threshold. Therefore, this invention uses connector contact resistance as the most important characteristic parameter of the faulty component. Based on the needs of the project research, a contact resistance monitoring device was developed to monitor electrical connectors under excitation conditions at a sampling frequency of up to 250kHz. This device has been used to study the contact resistance response characteristics and laws under different excitation conditions (environment, load), and the test results have also been used for connection structure characteristic analysis. Attached Figure Description

[0007] Figure 1 A configuration diagram of a multi-source fault diagnosis and testing apparatus for complex avionics equipment according to an embodiment of the present invention is shown.

[0008] Figure 2 A flowchart of a method for multi-source fault diagnosis and identification of complex avionics equipment according to an embodiment of the present invention is shown.

[0009] Figure 3 A flowchart of a DSmT-based electronic equipment health status assessment method according to an embodiment of the present invention is shown. Detailed Implementation

[0010] The method for multi-source fault diagnosis and identification of complex avionics equipment according to an embodiment of the present invention is based on a multi-source fault diagnosis and testing device for complex avionics equipment, such as... Figure 1 As shown, it includes:

[0011] This invention uses a four-wire method to measure contact resistance. The measurement circuit revolves around a data acquisition card, and the analog output of the data acquisition card controls a voltage-controlled current source to generate a constant current with adjustable range. The current flowing through the pin and socket contact pairs generates a contact voltage, which the data acquisition card acquires after low-noise amplification and low-pass filtering. The data acquisition card (102) communicates with the PC host computer software (101) to complete the setting of the current range and the uploading of acquired data. The data acquisition card is a PCI1716 high-resolution multi-functional data acquisition card. The constant current source (103) is built using an operational amplifier OP27, a high-power Darlington transistor MJ11022, and a precision power resistor. The 3.2kHz active low-pass filter is built using an operational amplifier OP27 and several RC components. The voltage signals acquired by the data acquisition card are all processed by a follower circuit and a low-pass filter circuit (105). The low-noise amplifier circuit (106) is the core of the conditioning section. It needs to amplify a weak signal of at least 1μV by 100 times while minimizing the adverse effects of circuit noise and external interference on the signal. An RFI filter is designed at the conditioning input to eliminate noise and filter out high-frequency interference. To reduce the impact of amplifier circuit noise on the useful signal, a two-stage 100x amplification circuit is designed using the low-noise differential instrumentation amplifier AD620, with each stage amplifying by 10 times. The system measures contact resistance within a range of 0~1Ω with a resolution of 0.244mΩ. The host computer software for the test system is written using the LabVIEW platform and is mainly used for setting parameters such as test current and sampling rate, as well as for acquiring, filtering, displaying in real time, storing, and processing contact resistance data. The test software consists of a contact resistance acquisition subroutine and a data processing subroutine.

[0012] A flowchart of a multi-source fault diagnosis and identification method for complex avionics equipment according to an embodiment of the present invention is shown below. Figure 2 As shown, it includes:

[0013] (1) Information preprocessing

[0014] For the performance status information of electronic equipment (202) at different stages of operation, the status information of the equipment is collected by setting up various sensing devices (203, 204) to realize the feature extraction of status information (205). The time domain (206) and time frequency (207) analysis methods are used to construct different status feature vector spaces and their corresponding status type spaces.

[0015] (2) Single information source state recognition based on Relevance Vector Machine (RVM)

[0016] The RVM model is constructed for all state feature vector spaces and corresponding state type spaces, and then model training is carried out based on relevant sample data (208). The trained RVM models are tested using state information collected from electronic equipment and preprocessed, and the training time and number of training sessions of the RVM models are effectively recorded to provide parameter support for the next step of decision fusion (209).

[0017] (3) State attribute decision fusion based on DSmT (210)

[0018] The necessary evidence space for DSmT based on RVM identification results is constructed, thus different identification conclusions for the equipment's state become evidence elements in the evidence theory. The necessary identification space for the evidence theory is constructed by comparing all state types of electronic equipment. Then, the probability of the equipment being in each state type is calculated using the generalized basic confidence assignment function. The DSmT fusion rule is applied to fuse the possible state type probabilities of each evidence element, further obtaining the overall probability distribution of the state type. Finally, decision fusion is completed, and the state type of the equipment is determined.

[0019] A flowchart of a DSmT-based electronic equipment health status assessment method according to an embodiment of the present invention is shown below. Figure 3 As shown, it includes:

[0020] To conduct a health status assessment of electronic equipment, the monitoring information reflecting the health status of the equipment is first collected (301), and a status vector is obtained through data processing and feature extraction (302). Then, the similarity correlation degree between the current single-source status vector and the normal and fault status vectors is calculated (303) to assess the similarity between the current health status and the normal and fault states. The similarity is then normalized to obtain the current equipment health index (304). It is then determined whether there is a serious over-limit (305). If there is no serious over-limit, a multi-source health index is obtained by processing multiple single-source health indices (306). If there is a serious over-limit, the cause of the over-limit is analyzed and the equipment problem is solved (307), thereby realizing the health assessment of the electronic equipment (308).

[0021] The advantages and beneficial effects of this invention include:

[0022] 1) The health status assessment of electronic equipment takes into account both monitoring information from a single information source and monitoring information from multiple information sources.

[0023] 2) Compared with traditional classification methods, this method first makes a preliminary judgment on the health index of a single information source when acquiring monitoring information of electronic equipment. If the index value is seriously out of range, it analyzes the cause and predicts the fault, thus avoiding the situation where information from multiple information sources obscures individual important abnormal information.

[0024] 3) Based on multiple single-source health indices, conduct a comprehensive assessment of the health status of electronic equipment, laying the foundation for further prediction of the health index of electronic equipment.

Claims

1. A complex avionics equipment multi-source fault diagnosis identification method based on a complex avionics equipment multi-source fault diagnosis test device, The complex avionics equipment multi-source fault diagnosis test device measures the contact resistance by using the four-wire method, controls the voltage-controlled current source through the analog output of the data acquisition card, generates a constant current with adjustable gears, and comprises: a PC host computer, a data acquisition card in communication with the PC host computer to complete the setting of the current gear and the uploading of the collected data, a constant current source, a 3.2 kHz active low-pass filter, a following circuit through which the voltage signal collected by the data acquisition card is processed, a low-noise amplification circuit which is the core of the conditioning part, an RFI filter designed at the input end of the conditioning part to remove noise and high-frequency interference, The complex avionics equipment multi-source fault diagnosis identification method is characterized in that it comprises: A1) information preprocessing, including collecting the state information of the equipment by setting various types of sensing devices for the performance state information of the electronic equipment at different stages of operation, and realizing feature extraction of the state information, A2) using time domain and time-frequency analysis methods to construct different state feature vector spaces and their corresponding state type spaces, A3) RVM-based single information source state recognition, including: A31) constructing RVM models for all state feature vector spaces and corresponding state type spaces, A32) then implementing model training based on relevant sample data, A33) testing the trained RVM models using the state information collected from the electronic equipment and subjected to relevant preprocessing, and recording the time and number of RVM model training, thereby providing parameter support for the next decision fusion step, A4) state attribute decision fusion based on DSmT, including: A41) constructing the evidence space required by DSmT based on the RVM recognition results, so that different recognition conclusions of the state of the equipment become evidence elements in the evidence theory, A42) constructing the necessary recognition space of the evidence theory by comparing all state types of the electronic equipment, A43) then calculating the probability of the equipment being in each state type by using the generalized basic belief assignment function, A44) using the fusion rules of DSmT to fuse the possible state type probabilities of each evidence element, further obtaining the total probability distribution of the state type, and finally completing the decision fusion and determining the state type of the equipment.

2. The complex avionics equipment multi-source fault diagnosis identification method according to claim 1, wherein: The data acquisition card is a PCI1716 high-resolution multifunctional data acquisition card.

3. The complex avionics equipment multi-source fault diagnosis identification method according to claim 1, wherein: The constant current source is built with an operational amplifier OP27, a high-power Darlington transistor MJ11022, and a precision power resistor.

4. The complex avionics equipment multi-source fault diagnosis identification method according to claim 1, wherein: 3.2 kHz active low-pass filter is built with operational amplifier OP27 and several resistors and capacitors.

5. An electronic equipment health status assessment method, characterized by Comprise: B1) first, using the complex avionics equipment multi-source fault diagnosis identification method according to one of claims 1-4, the monitoring information reflecting the health state of the device is collected, B2) after data processing and feature extraction, a state vector is obtained, B3) then, the similarity correlation degree of the current single information source state vector and the normal state vector and the fault state vector is calculated, B4) according to the result of similarity correlation degree calculation, the similarity degree between the current health state and the normal and fault state is evaluated, and the similarity degree is normalized to obtain the health index of the current device, B5) judge whether there is serious overrun.

6. The electronic equipment health state evaluation method according to claim 5, characterized in that: when it is determined in step B5 that there is no serious overrun, the multi-information source health index is obtained by processing the multiple single information source health indexes.

Images