Support Vector Machines based Rotor Fault Diagnosis with Improved Particle Swarm optimization

Artificial intelligence methods such as support vector machines (SVM) are widely used in fault diagnosis of mechanical rotors. SVM is a general machine-learning tool based on structural risk minimisation principle that performs good generalisation. Many studies have shown that the performance of SVM classifiers in fault diagnosis can be improved by optimizing parameters. For easy implemention, intelligent optimization methods are often used to optimize complex engineering problems. In this work, two improved particle swarm optimization algorithms are used to optimize parameters of SVM classifiers to improve identification rate of mechanical rotor faults. The basic particle swarm optimization (PSO) is also discussed for comparisons. Simulation shows the performance of SVM classifiers can be improved in rotor fault diagnosis by using the two improved PSO.