Helicopter transmission system bearing detection method driven by fault sensitive coding network

Abstract

The invention discloses a helicopter transmission system bearing detection method driven by a fault sensitive coding network, and the method comprises the steps of collecting a healthy helicopter transmission system bearing noise signal as a training signal, carrying out the dimensionality reduction of the training signal through an encoder, obtaining a low-dimensional manifold, reconstructing a signal through a decoder based on the low-dimensional manifold, wherein the generative adversarial network comprises a generator and a discriminator, the generator generates a to-be-detected signal, and the discriminator discriminates whether the to-be-detected signal is from a training signal or a reconstruction signal; calculating a difference value between the training signal and the reconstruction signal to obtain a reconstruction error, and enabling the generative adversarial network to carry out adversarial training alternative optimization to achieve Nash equilibrium; obtaining the distribution of the health state signals through fitting; inputting a test signal to an encoder to reduce dimensions to obtain a low-dimensional manifold; enabling the decoder to reconstruct the signal based on the low-dimensional manifold; inputting a generative adversarial network for adversarial training and fitting to obtain test signal distribution; and if the distribution of the test signals does not conform to the distribution of the health state signals, determining that a fault exists.

Description

technical field [0001] The invention belongs to the technical field of bearing detection of a helicopter transmission system, in particular to a detection method for a bearing of a helicopter transmission system driven by a fault-sensitive coding network. Background technique [0002] As a rotary-wing aircraft, helicopters have the characteristics of vertical take-off and landing, flexible flight direction change, and hovering in the air. They have been widely used in military fields such as ground attack, air landing, logistics support, and battlefield rescue, as well as medical rescue, disaster relief and fire fighting. , short-distance transportation, hoisting equipment and other civil fields. However, compared with fixed-wing aircraft, helicopter accidents are more frequent, and the proportion of serious accidents is large, which puts forward high requirements for the safety and reliability of helicopters. While pursuing the high performance of helicopters, military pow...

AI Technical Summary

Problems solved by technology

[0004] However, there are still three shortcomings in the existing methods: 1) The process of feature extraction often depends on human experience, and the extracted features have information loss and are not reconfigurable

2) Some methods for condition monitoring do not adopt a classification strategy, and are not suitable for condition monitoring application scenarios where only health status data exists

3) A large number of anomaly monitoring methods are designed for the characteristics of the image and are not suitable for rotating machinery

Images