Research on Bearing Fault Detection Algorithm Based on Convolution Neural Network and SVM
HaoNing Pu, Zhan Wen, Bing Wan
Citation
HaoNing Pu, Zhan Wen, Bing Wan, “Research on Bearing Fault Detection Algorithm Based on Convolution Neural Network and SVM”, International Journal of Engineering and Applied Computer Science, vol. 04, no. 07, pp. 01-08, August, 2022.
Abstract
The traditional bearing fault detection method is using the inner diameter micrometer. This method of manual measurement not only has a large workload and low efficiency, but also has poor reliability and a high rate of leakage, which affects the product quality and makes it difficult to meet the needs of current production. This paper mainly uses the support vector machine (SVM) model and the convolution neural network (CNN) model to solve the shortage of bearing fault manual detection. This paper is based on the Anaconda platform, and uses Python language to program bearing fault detection. To detect bearing fault in SVM model, the main characteristics of bearing signal are extracted by principal component analysis (PCA), and then the main feature parameters are optimized by Particle Swarm Optimization (PSO), which can identify the fault of rolling bearing more quickly and accurately. To achieve the fault diagnosis of CNN model, CNN can use convolution kernel to automatically mine the features that are difficult to extract from the fault signal, and has superior fault diagnosis performance. Therefore, using the SVM model and CNN model to classify and detect bearing faults has the advantages of being fast and efficient, and can well overcome the shortcomings of traditional methods.
h-index
6
i10-index
4
Impact Factor
2.35
International Journal of Engineering and Applied Computer Science
EMPIRICAL RESEARCH PRESS LTD. LONDON UNITED KINGDOM
