Classification of faults in mechanical components using machine learning theory is attracted intense scrutiny and interest from both scientists and engineers. Generally, every mechanical system is exhibit personalized vibration behaviors under different assemble and work conditions. Furthermore, for the lack of faulty samples of real-world mechanical systems, fault classification using machine learning methods, such as support vector machine (SVM), neural networks (NNs), etc., are often difficult to achieve agreeable results. In this paper, a personalized faults diagnosis method using finite element method (FEM) simulation and SVM is proposed. Firstly, the finite element method (FEM) simulation is performed to generate a large number of simulation signals with different faults. Secondly, the simulation signals are employed as training samples to train SVM. Finally, the actual signals of unknown samples (test samples) are inserting into the trained SVM to classify faults. More specifically, the personalized fault diagnosis method is applied to diagnosis bearing faults, and the final results confirm the effectiveness of the method for mechanical fault diagnosis.
展开▼
