Bearing fault diagnosis method based on total variation and compensation distance assessment

Abstract

The invention discloses a bearing fault diagnosis method based on total variation and compensation distance assessment, comprising the steps that a sensor collects a multi-channel vibration signal ofa rolling bearing; the vibration signal is segmented according to time sequence, forming a sample set; time domain characteristic of the sample is extracted, and the vibration signal total variation is introduced, obtaining time domain one-dimensional characteristic row vector; the time domain one-dimensional characteristic row vectors of different channels form a high-dimensional data characteristic set, and a sensitive characteristic set is obtained by adopting a compensation distance assessment algorithm; training is conducted by means of a support vector machine with PSO optimization and afault diagnosis model is established in a bid to determine the bearing fault type and output the result, and bearing fault types are determined and results are output. The invention is advantageous in that through collecting the vibration signals of the rolling bearing under different running states, multiple sample sets can be obtained based on time sequence, and time domain characteristics canbe extracted, and total variation characteristics are introduced, forming a multidimensional time domain characteristic set; the sensitive characteristic index can be selected through a compensation distance assessment algorithm, and a PSO-optimized support vector machine is adopted for training the fault diagnosis model, and thereby the fault diagnosis precision is high, and training speed is fast.

Description

technical field [0001] The invention relates to a bearing fault diagnosis method, in particular to a bearing fault diagnosis method based on total variation and compensation distance evaluation. Background technique [0002] Rolling bearing is an important component in rotating machinery, and it is also a relatively common and easily damaged part in rotating machinery. It plays a key role in rotating machinery. Whether its working status is normal or not directly affects the performance of the entire unit. Therefore, whether it can be quickly and Accurately detecting the existence and severity of bearing faults is of great significance for ensuring safe and reliable operation of bearings and reducing equipment downtime costs. According to statistics, 30% of the faults of rotating machinery are caused by bearings, so the multi-classification of rolling bearing faults is particularly important. [0003] The traditional rolling bearing fault diagnosis method is to extract the ...

AI Technical Summary

Problems solved by technology

Because each feature index has a different impact on fault diagnosis, some indexes are very important, and some have little or little relationship, so the redundant information of high-dimensional feature vectors will cause dimensionality disaster for subsequent pattern analysis, increasing calculation time, Affects fault classification accuracy

However, traditional dimensionality reduction methods such as principal component analysis and linear discriminant analysis are limited to dealing with linear features, and there are often complex nonlinear relationships in the feature sets of rolling bearing vibration signals, so traditional dimensionality reduction methods face challenges

[0004] At present, rolling bearing fault diagnosis algorithms generally use data-driven machine learning algorithms, such as neural networks, fuzzy recognition, Bayesian classification, etc., which have the disadvantages of long training time or low classification accuracy.

Images