Rolling bearing residual life prediction method based on cv and bhsmm model

By using a hidden semi-Markov model improved with logistic regression and beta distribution, the problems of data redundancy and low accuracy in the prediction of the remaining life of rolling bearings are solved, achieving accurate prediction of rolling bearing failures and reducing computational complexity.

CN116522524BActive Publication Date: 2025-10-24JIANGSU UNIV OF SCI & TECH
View PDF -1 Cites 0 Cited by

Patent Information

Application Number
CN202310351881.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-04-04
Publication Date
2025-10-24
Estimated Expiration
2043-04-04

AI Technical Summary

Technical Problem

Existing methods for predicting the remaining life of rolling bearings suffer from problems such as data redundancy, limited technology, and low prediction accuracy. Furthermore, traditional models cannot effectively handle noise, leading to inaccurate diagnostic results.

Method used

An improved hidden semi-Markov model (BHSMM) based on logistic regression and beta distribution is adopted. Vibration signal clutter is cleared by evaluating the CV index, and the observation probability matrix and autocorrelation are introduced to optimize the state dwell time distribution and establish a rolling bearing remaining life prediction model.

Benefits of technology

It improves the accuracy and real-world fit of rolling bearing failure prediction, reduces computational complexity, and enables accurate prediction of the remaining life of rolling bearings.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116522524B_ABST
    Figure CN116522524B_ABST
Patent Text Reader

Abstract

The application discloses a rolling bearing residual life prediction method based on a CV and BHSMM model in the field of fault prediction, collects vibration signals of different fault degrees of a rolling bearing under no-fault and different fault conditions, obtains column vectors of the vibration signals, obtains a logistic regression function CV as an observation value according to the column vectors, effectively removes clutter in the vibration signals of the rolling bearing, optimizes an observation probability value according to a Gaussian distribution, obtains a probability with autocorrelation, optimizes a state residence time according to a Beta distribution, obtains a BHSMM model according to the optimized probability with autocorrelation, designs a classifier for the BHSMM model, obtains a classifier with a maximum conditional probability, calculates a corresponding residence time, obtains a residual life according to the residence time, realizes accurate prediction of a residual service life of the rolling bearing, can better describe a degradation process, has higher real fitting, adopts a probability revaluation as an intermediate variable, and reduces complexity of calculation.
Need to check novelty before this filing date? Find Prior Art