An intelligent mechanical system fault diagnosis method based on multi-level multi-mode feature extraction

Abstract

The invention discloses an intelligent mechanical system fault diagnosis method based on multi-level multi-mode feature extraction. The method comprises the steps that an original vibration sequence of a to-be-detected mechanical part is acquired; Performing first feature extraction on the original vibration sequence by adopting a principal component analysis method to obtain a first target sampleof the to-be-detected mechanical part; Carrying out second feature extraction on the first target sample by adopting a non-negative matrix factorization method to obtain a second target sample of theto-be-detected mechanical component; Two vibration sequence predictors obtained through training based on a PID neural network and an extreme learning machine are used for predicting the two target samples respectively to obtain two predicted vibration sequences, and weighted fusion is carried out to obtain a fusion prediction vibration sequence; And outputting the fault type of the to-be-detected mechanical part according to the fusion prediction vibration sequence by using an intelligent mechanical system fault classification model obtained based on training of the support vector machine. The method is high in mechanical part fault diagnosis accuracy and high in robustness and timeliness.

Description

technical field [0001] The invention relates to the field of mechanical system fault identification and prediction, in particular to a multi-level and multi-mode intelligent mechanical system fault diagnosis method. Background technique [0002] Fault diagnosis and identification of the mechanical system is an important prerequisite to ensure the normal operation of the mechanical system with high reliability and maintainability. The train bogie is the connection coupling device between the train body and the track, and its safety guarantee during operation is very important. The main fault problem of the train bogie is the vibration of the components of the bogie due to the unevenness of the track. Phenomenon. The current mechanical fault diagnosis and identification technology is mainly an intelligent fault diagnosis method based on the steps of signal acquisition, feature extraction, fault identification and prediction. This type of method uses the sensor network distri...

AI Technical Summary

Problems solved by technology

[0003] However, the existing fault diagnosis method extracts a single characteristic parameter for the fault signal, resulting in low identification of mechanical fault diagnosis. Therefore, it is urgent to study a mechanical system fault diagnosis method with high timeliness and identification.

Images