A Fault Diagnosis Method for Rolling Bearings Based on Approximate Support Vectors

Abstract

The invention discloses a rolling bearing fault diagnosis method based on an approximate support vector. First use the MiniBatchKmeans algorithm and support vector machine to quickly find the "approximate support vector", then use the "approximate support vector" to filter out samples near the "approximate support vector" in the original sample, and finally use the filtered samples to train the support vector machine. The invention greatly reduces the training time of the support vector machine, and at the same time ensures the accuracy of fault diagnosis.

Description

technical field [0001] The invention relates to the field of bearing fault diagnosis, in particular to a rolling bearing fault diagnosis method based on approximate support vectors. Background technique [0002] With the continuous development of science and technology, the use of modern large-scale machinery by human beings has gradually matured, and a large part of the mechanical structure tends to be more precise and complex. Such modern machinery has greatly improved people's production efficiency. However, production accidents caused by mechanical failures are also increasing. Once a certain part fails, the entire mechanical system will stop working, resulting in abnormal production, and even causing great damage to people's lives and property. Therefore, it is particularly important to discover the type of mechanical failure in time and take corresponding measures. [0003] As the most important part of machinery, rolling bearings are also one of the most prone to fa...

AI Technical Summary

Problems solved by technology

However, according to the research of the inventor, the above method still has the problem that it cannot reasonably identify non-support vectors or the recognition accuracy is not enough, and it takes a lot of time for clustering: for example, training support vectors by clustering centers after K-means algorithm clustering Although this method can reduce a certain amount of training time, when the amount of data is large, the K-means algorithm itself also needs to consume a lot of time. According to the principle of the K-means algorithm, the clustering center after clustering does not exist originally. samples, but the mean value of all samples belonging to this class; at the same time, most of the cluster centers are not support vectors, but the results of non-cluster centers are support vectors, thus, it will not be possible to filter out suitable samples To train the support vectors, and then make the fault diagnosis data disorder, which seriously affects the troubleshooting and the use of machinery

Images