A planetary gear box intelligent diagnosis method based on a multi-flow fusion model

Abstract

A planetary gear box intelligent diagnosis method based on a multi-flow fusion model comprises the following steps: firstly, acquiring an encoder signal, smoothing the encoder signal to obtain a smooth signal, and performing first-order and second-order difference on the smooth signal to obtain an angular velocity signal and an angular acceleration signal; Secondly, dividing the smooth signal, theangular velocity signal and the angular acceleration signal into a training set, a verification set and a test set respectively to obtain a smooth signal data set, a harmonic signal data set and an impact signal data set; Aiming at the smooth signal, the angular velocity signal and the angular acceleration signal, respectively constructing corresponding convolutional neural network models, carrying out differential training, fusing high-level abstract features in an independent frequency domain learned by the three convolutional neural network models, and finally obtaining a multi-flow fusionmodel; optimizing High-level parameters of the multi-stream fusion model to obtain a final model, carrying out fault classification, and use the t-SNE technology to visualize the classification results of the test.; According to the invention, different types of faults can be distinguished, and high-precision fault diagnosis is realized.

Description

technical field [0001] The invention relates to the technical field of planetary gearbox fault diagnosis, in particular to an intelligent diagnosis method for a planetary gearbox based on a multi-flow fusion model. Background technique [0002] Planetary gearboxes have the advantages of small size, light weight, compact structure, high transmission efficiency, and large load-carrying capacity. They are important components for energy transmission in industrial machinery. is widely used. With the increasing requirements of modern machinery for planetary gear transmission, planetary gears not only need to be able to work under conditions such as high speed, heavy load, and special media, but also need to have good working performance such as high stability, high reliability, and compact structure. Ensure the safe and normal operation of machinery and equipment. It can be seen that the failure of the planetary gear equipment will cause the equipment to shut down, directly cau...

AI Technical Summary

Problems solved by technology

The current fault diagnosis methods for planetary gearboxes use vibration signals as the information source for fault diagnosis, and most of them have the following deficiencies: 1) The signal-to-noise ratio of vibration signals is low; 2) The signal source is single; 3) The signal analysis method is complex and relies on manual work. Feature extraction; What's more, vibration sensors are difficult to install on high-end equipment such as fully enclosed machining centers and multi-degree-of-freedom robotic arms, resulting in the inability to obtain analysis signals

[0005] However, most current fault diagnosis methods based on encoder signals focus on the analysis mode of vibration signals, that is, using traditional signal processing methods to perform complex noise reduction and feature extraction for fault diagnosis, which are limited by two aspects: First, As the encoder signal is an angle domain signal, it is difficult to directly transplant the relatively well-developed vibration signal analysis method to the encoder signal processing field; second, the signal processing method relies too much on professional knowledge and experience, and it is difficult to realize automation and intelligence. diagnosis

Of course, the current hot deep learning method is also emerging in the field of fault diagnosis based on encoder signals, but it focuses on the parameter optimization of simple networks, ignoring the exploration of encoder signals reflecting the diversity of equipment operating states and the innovation of model structures, resulting in Information leakage, resulting in insensitivity to certain fault types, reducing the reliability of planetary gearbox fault diagnosis

Images