Aircraft fault diagnosis model generation method based on multi-sensor data driving

Abstract

The invention belongs to the technical field of aircraft fault diagnosis, and discloses an aircraft fault diagnosis model generation method based on multi-sensor data driving. The method comprises the steps of 1, selecting a sensor group capable of being used for aircraft fault diagnosis, and collecting historical acquisition data of the sensor group corresponding to various fault modes; preprocessing the collected data of the sensor group corresponding to each type of fault mode in one time period, and storing the preprocessed data as a two-dimensional multi-sensor data matrix, thereby obtaining T * Q two-dimensional multi-sensor data matrixes of the T types of fault modes in Q time periods; 2, randomly selecting the two-dimensional multi-sensor data matrixes and the corresponding fault modes as training samples, and taking the rest two-dimensional multi-sensor data matrixes and the corresponding fault modes as test samples; 3, establishing a convolutional neural network model; 4, training the convolutional neural network by using the training sample through a small-batch gradient descent method until a training condition is met; and 5, verifying a training result.

Description

technical field [0001] The invention belongs to the technical field of aircraft fault diagnosis, and in particular relates to an aircraft fault diagnosis model generation method driven by multi-sensor data. Background technique [0002] In order to improve flight safety and mission reliability, a large number of advanced technologies are adopted in the design of modern fighter jet systems, and the complexity and integration of the system are increasing day by day, which makes the fault diagnosis of fighter jet systems face more severe challenges. Once a fighter system fails, if it cannot be detected and dealt with in time, it will pose a serious threat to flight safety and mission reliability. [0003] In order to improve the health monitoring capability of fighter jets, a large number of sensors are arranged in the system design of modern fighter jets to monitor the working status of the system, and a large amount of monitoring data will be generated during the operation of...

AI Technical Summary

Problems solved by technology

[0007] Purpose of the present invention: In order to solve the complex fault diagnosis problem of the fighter system, according to the multi-dimensional data composed of data collected by multiple sensors of the fighter system, the present invention provides a method for generating an aircraft fault diagnosis model driven by multi-sensor data. Convolutional Neural Networks Extract Fault Features from Sensor Data

Images