Multi-feature and phase space-based dynamic prediction method of life of antifriction bearing

Abstract

Provided is a multi-feature and phase space-based dynamic prediction method of life of an antifriction bearing. The dynamic prediction method comprises steps of firstly, obtaining vibration signals of antifriction bearings, extracting time domain and frequency domain features of the antifriction bearings and carrying out standardization and slipping processing so as to obtain relative feature indexes; then, using the PCA technology to fuse the multiple feature indexes, obtaining a comprehensive attenuation performance index and carrying out life detection; then, carrying out space reconstruction on the comprehensive attenuation performance index in a history degenerative process, and comparing the history degenerative process with a current degenerative process in a phase space so as to obtain predicted invalid time; matching and combining the predicted invalid time with history invalid time, estimating the probability density distribution thereof, and calculating average life; and continuously accumulating and expanding by analyzing samples to obtain average life at different observation moments. According to the invention, effects of multiple factors in a degenerative process of the antifriction bearing can be effectively reflected; the prediction method is free from restriction of hypothesis conditions; prediction errors are reduced; and dynamic and precise prediction of life can be achieved.

Description

technical field [0001] The invention relates to the technical field of rolling bearing life prediction, in particular to a dynamic prediction method for rolling bearing life based on multiple features and phase space. Background technique [0002] Rolling bearings are important parts and one of the key components that are vulnerable to damage in mechanical equipment. Their performance and reliability play a vital role in the performance and reliable operation of the entire mechanical equipment. Rolling bearing life prediction is actually based on its current degradation state, predicting the time it takes from the start of operation to failure, so as to facilitate planned maintenance and lay the foundation for equipment maintenance decisions. Therefore, predicting the life of rolling bearings during operation is of great significance for predictive maintenance of equipment. [0003] The selection of performance indicators is very important to accurately predict the life of ...

AI Technical Summary

Problems solved by technology

[0003] The selection of performance indicators is very important to accurately predict the life of rolling bearings. The traditional life prediction uses a certain feature of the vibration signal as a decline performance indicator to predict, which cannot effectively reflect the effects of multiple factors in the degradation process of rolling bearings.

Due to the nonlinear and non-stationary characteristics of the rolling bearing degradation process, the traditional method of establishing a life prediction model through algorithms such as BP neural network and support vector machine for life prediction is constrained by assumptions, and is limited by the small number of observation samples and randomness in the early stage of prediction. Due to the influence of the error, the error of the prediction result is relatively large, and the accurate and dynamic prediction of the life of the rolling bearing cannot be realized.

Images