Self-learning-based variable working condition mechanical fault intelligent diagnosis method under small sample

Abstract

The invention discloses a self-learning-based variable working condition mechanical fault intelligent diagnosis method under a small sample. The method comprises the steps of performing standardized preprocessing on a mechanical signal sample of mechanical equipment in a variable working condition operation state; constructing and training a generative adversarial network, and enhancing a trainingdata set by generating a mechanical signal sample; constructing a self-learning network model; pre-training a self-learning network model by using the mechanical signal samples generated by the generative adversarial network and not more than 30 real mechanical signal samples; constructing a mechanical equipment state classification and recognition network model; and the network weight of the mechanical equipment state classification and recognition network model is finely adjusted by using not more than 30 real mechanical signal samples and the corresponding state labels, and the mechanicalequipment state classification and recognition network model after fine adjustment of the weight can realize intelligent diagnosis of mechanical faults under variable working conditions. The method has the characteristics of high diagnosis result accuracy and strong anti-interference capability.

Description

technical field [0001] The invention relates to the field of fault diagnosis of mechanical equipment, in particular to an intelligent diagnosis method for mechanical faults under variable working conditions based on self-learning under small samples. Background technique [0002] During the operation of mechanical equipment, its main components such as bearings, gears, rotors, etc. are prone to failure due to the continuous load, so it is necessary to carry out research on fault diagnosis of mechanical equipment. On the one hand, due to the constraints of the working environment and working conditions, mechanical equipment is usually in a state of variable working conditions, which usually manifests as changes in load and speed. Moreover, all kinds of mechanical signals collected on mechanical equipment will be polluted by noise, making feature extraction and state recognition difficult. On the other hand, it is difficult to obtain fault signal samples of mechanical equipme...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a self-learning-based intelligent diagnosis method of variable working condition mechanical faults in a small sample to overcome the problems in the prior art. The present invention uses a generative confrontation network to generate a large number of mechanical signal samples to enhance the training data set. Solve the small sample problem, and at the same time use the self-learning network model to obtain fault feature information from a large number of mechanical signal samples, solve the problem that it is difficult to extract effective features under variable operating conditions, and finally use the one-dimensional convolutional neural network to realize the Classification and identification of operating status of mechanical equipment

Images