Fault Detection of Bearing using Support Vector Machine-SVM

Modern spinning machinery is a crucial component of rolling element. The principal aim of this project is to create a support vector machine model, which is one of the AI techniques to detect and diagnose bearing fault at early stage. The development of the model should be able to forecast the bearing fault diameters based on the collected input variables. In order to achieve this objective, a set of bearing raw vibration frequency signal is acquired. The raw vibration signals were extracted. The extracted features are used as the inputs containing different motor loads, different motor speeds and different locations. The support vector machine approach is being used to run the simulation. The selection of kernel functions and other parameters are very important in the development of a reliable model. Trial and error method are used to identify the best combination of parameters for SVM model by comparing the MSE and CC values. The best kernel functions and parameters are set and the model is ready to be used to run the real data since it can provide the best and most accurate precision in early detecting bearing failures. Recommendation was made to improve the architecture of SVM model.