CN-122017148-A - Gas sensor fault diagnosis method and device, electronic equipment and vehicle

CN122017148ACN 122017148 ACN122017148 ACN 122017148ACN-122017148-A

Abstract

The embodiment of the application provides a fault diagnosis method, a fault diagnosis device, electronic equipment and a vehicle of a gas sensor, which are used for denoising initial gas data according to environmental data to obtain target gas data, wherein the target gas data are used for updating material parameters, temperature data, material parameters and target gas data in the environmental data are input into a fault diagnosis model which is trained in advance to obtain a fault gas sensor in the gas sensor output by the fault diagnosis model, wherein the fault diagnosis model comprises a graph convolution network, the graph convolution network is used for determining edge weights of edges according to the temperature data and the material parameters of the gas sensor, the edge weights are used for determining the fault gas sensor in the gas sensor, so that the complex fault of the gas sensor is identified, and the fault diagnosis accuracy of the gas sensor is remarkably improved.

Inventors

ZHONG JIQIN
JIANG QIONG
LI MINGYI
FU ZENGKUN
HUANG CANLIN

Assignees

广州汽车集团股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260303

Claims (11)

1. A fault diagnosis method of a gas sensor, comprising: acquiring material parameters of the gas sensor, environmental data of the environment in which the gas sensor is positioned and initial gas data collected by the gas sensor, wherein the material parameters are physical parameters reflecting material characteristics of the gas sensor; denoising the initial gas data according to the environmental data to obtain target gas data, wherein the target gas data is used for updating the material parameters; Inputting temperature data, the material parameters and the target gas data in the environment data into a failure diagnosis model which is trained in advance, and determining a failure gas sensor in the gas sensor; The fault diagnosis model comprises a graph rolling network, wherein the graph rolling network comprises nodes representing the gas sensors and edges connecting the nodes, the graph rolling network is used for determining edge weights of the edges according to the temperature data and material parameters of the gas sensors, and the edge weights are used for determining fault gas sensors in the gas sensors.
2. The method for diagnosing a fault of a gas sensor according to claim 1, wherein the environmental data includes temperature data, humidity data and air pressure data, the denoising the initial gas data according to the environmental data to obtain target gas data includes: Carrying out wavelet packet decomposition on the initial gas data to obtain a gas sub-signal; determining a sensitive sub-signal from the gas sub-signals according to the temperature data, the humidity data and the air pressure data by adopting a correlation coefficient function, wherein the sensitive sub-signal is a gas sub-signal sensitive to the changes of the temperature data, the humidity data and the air pressure data; an adaptive filtering algorithm is adopted, and an environmental interference signal in the sensitive sub-signal is determined according to the temperature data, the humidity data and the air pressure data; removing an environmental interference signal in the sensitive sub-signal from the gas sub-signal to obtain a denoised gas sub-signal; Carrying out wavelet packet reconstruction on the denoised gas sub-signals to obtain denoised gas data; and carrying out baseline compensation on the denoised gas data to obtain the target gas data.
3. The method for diagnosing a fault in a gas sensor according to claim 2, wherein the performing baseline compensation on the denoised gas data to obtain the target gas data includes: determining a baseline drift rate factor from the material parameter; and carrying out baseline compensation on the denoised gas data according to the baseline drift rate factor by adopting an exponential weighted moving average model to obtain the target gas data.
4. The method according to claim 1, wherein the material parameter includes an activation energy parameter of a gas sensor material, and the inputting the temperature data in the environmental data, the material parameter, and the target gas data into a failure diagnosis model trained in advance, to obtain a failure gas sensor in the gas sensor and a failure type of the gas sensor output from the failure diagnosis model, includes: Inputting the material parameter and the temperature data into the graph rolling network, so that the graph rolling network takes the material parameter as a node attribute of the node, and determining the edge weight between the nodes according to the temperature data and the activation energy parameter of the gas sensor material; and determining a fault gas sensor in the gas sensors according to the edge weights.
5. The method of claim 1, wherein the fault diagnosis model further comprises a residual time series convolution network, and the inputting the temperature data, the material parameters and the target gas data in the environmental data into the pre-trained fault diagnosis model comprises: and inputting the target gas data into the residual time sequence convolution network, so that the residual time sequence convolution network determines the fault evolution characteristics of the gas sensor from a time scale according to the target gas data, and the fault type of the gas sensor is obtained.
6. The method of claim 5, wherein the residual time series convolution network comprises convolution paths of different time scales, and wherein the inputting the target gas data into the residual time series convolution network comprises: Inputting the target gas data into the residual time sequence convolution network, so that the residual time sequence convolution network adopts convolution paths of different time scales to respectively process the target gas data to obtain residual signals of different scales; Carrying out fusion processing on the residual signals with different scales through the residual time sequence convolution network to obtain fusion residual; and comparing the fusion residual error with preset healthy gas sensor data to obtain the fault type of the gas sensor.
7. The method of diagnosing a malfunction of a gas sensor according to claim 1, wherein the material parameter includes an adsorption parameter of a gas sensor material, the method comprising: obtaining a fault diagnosis model to be trained, wherein the fault diagnosis model to be trained comprises a graph convolution network to be trained; based on a Langmuir adsorption equation, constructing a physical constraint term according to the adsorption parameters of the gas sensor material; and training the fault diagnosis model to be trained according to the physical constraint item to obtain the fault diagnosis model which is subjected to training in advance, wherein the physical constraint item is used for guiding model parameters of the graph rolling network to be trained to converge towards the direction conforming to gas adsorption dynamics in training.
8. The fault diagnosis method of a gas sensor according to claim 1, characterized in that the method comprises: updating the material parameters of the gas sensor according to the target gas data and the environment data to obtain locally updated material parameters; Uploading the material parameter increment determined by each gas sensor to a cloud end, so that the cloud end carries out weighted average polymerization on the locally updated material parameter of the gas sensor according to the material parameter increment to obtain a target material parameter, wherein the material parameter increment is determined according to the locally updated material parameter and the material parameter; And updating the material parameter after the local updating of the gas sensor into the target material parameter.
9. A fault diagnosis device of a gas sensor, the device comprising: The system comprises a parameter acquisition module, a gas sensor and a gas sensor, wherein the parameter acquisition module is used for acquiring material parameters of the gas sensor, environmental data of the environment where the gas sensor is located and initial gas data collected by the gas sensor, and the material parameters are physical parameters reflecting material characteristics of the gas sensor; The parameter processing module is used for denoising the initial gas data according to the environmental data to obtain target gas data, wherein the target gas data is used for updating the material parameters; The fault diagnosis module is used for inputting temperature data, the material parameters and the target gas data in the environment data into a pre-trained fault diagnosis model and determining a fault gas sensor in the gas sensor; The fault diagnosis model comprises a graph rolling network, wherein the graph rolling network comprises nodes representing the gas sensors and edges connecting the nodes, the graph rolling network is used for determining edge weights of the edges according to the temperature data and material parameters of the gas sensors, and the edge weights are used for determining fault gas sensors in the gas sensors.
10. An electronic device comprising a processor and a memory, wherein A memory for storing a computer program; A processor for executing a program stored in the memory to realize the failure diagnosis method of the gas sensor according to any one of claims 1 to 8.
11. A vehicle comprising the electronic device of claim 10.

Description

Gas sensor fault diagnosis method and device, electronic equipment and vehicle Technical Field The embodiment of the application relates to the technical field of sensors, in particular to a fault diagnosis method and device of a gas sensor, electronic equipment and a vehicle. Background The gas sensor is widely applied to the fields of industrial safety (such as chemical leakage monitoring), environmental monitoring (such as air quality detection), medical diagnosis (such as expired gas disease marker analysis) and the like. Taking a nitrogen-oxygen sensor as an example, it has a key role in an automobile exhaust gas control system, and if the sensor fails, emissions may be out of standard or engine performance may be reduced. However, the existing gas sensor fault diagnosis methods still have certain limitations that on one hand, nonlinear relations in sensor response are difficult to effectively process, the recognition rate of complex faults such as aging and pollution is low, and on the other hand, most methods rely on manual feature extraction, and generally rely on field knowledge, so that dynamic changes of response signals of the gas sensor are difficult to automatically capture. Therefore, how to automatically capture the dynamic changes of the response signals of the gas sensor, so as to identify the complex faults in the gas sensor is a problem to be solved in the field. Disclosure of Invention The embodiment of the application provides a fault diagnosis method and device of a gas sensor, electronic equipment and a vehicle, and aims to solve the problem of how to automatically capture dynamic changes of response signals of the gas sensor so as to identify complex faults in the gas sensor. In order to solve the above problems, an embodiment of the present application discloses a fault diagnosis method for a gas sensor, including: acquiring material parameters of the gas sensor, environmental data of the environment in which the gas sensor is positioned and initial gas data collected by the gas sensor, wherein the material parameters are physical parameters reflecting material characteristics of the gas sensor; denoising the initial gas data according to the environmental data to obtain target gas data, wherein the target gas data is used for updating the material parameters; Inputting temperature data, the material parameters and the target gas data in the environment data into a failure diagnosis model which is trained in advance, and determining a failure gas sensor in the gas sensor; The fault diagnosis model comprises a graph rolling network, wherein the graph rolling network comprises nodes representing the gas sensors and edges connecting the nodes, the graph rolling network is used for determining edge weights of the edges according to the temperature data and material parameters of the gas sensors, and the edge weights are used for determining fault gas sensors in the gas sensors. According to the embodiment of the application, the edge weight of the graph rolling network is dynamically determined according to the temperature data and the material parameters of the gas sensor, so that the embedding of the material science principle into the graph rolling network is realized. When the gas sensor is subjected to faults such as aging, the material parameters of the gas sensor are changed, so that the edge weight is correspondingly changed, and a traceable fault propagation path is formed in the characteristic propagation process of the graph rolling network, so that the fault diagnosis model can effectively identify the fault gas sensor, the complex fault of the gas sensor is identified, the diagnosable fault type is increased, and the fault diagnosis accuracy of the gas sensor is remarkably improved. Optionally, the environmental data includes temperature data, humidity data and air pressure data, and denoising the initial air data according to the environmental data to obtain target air data, including: Carrying out wavelet packet decomposition on the initial gas data to obtain a gas sub-signal; determining a sensitive sub-signal from the gas sub-signals according to the temperature data, the humidity data and the air pressure data by adopting a correlation coefficient function, wherein the sensitive sub-signal is a gas sub-signal sensitive to the changes of the temperature data, the humidity data and the air pressure data; an adaptive filtering algorithm is adopted, and an environmental interference signal in the sensitive sub-signal is determined according to the temperature data, the humidity data and the air pressure data; removing an environmental interference signal in the sensitive sub-signal from the gas sub-signal to obtain a denoised gas sub-signal; Carrying out wavelet packet reconstruction on the denoised gas sub-signals to obtain denoised gas data; and carrying out baseline compensation on the denoised gas data to obtain the target gas data. The embodiment of the application p