Artificial Immune Based Support Vector Machine Algorithm for Fault Diagnosis of Induction Motors

The use of induction motors is widespread in industry. Many researchers have studied the condition monitoring and detecting the faults of induction motors at an early stage. Early detection of motor faults results in fast unscheduled maintenance. In this study, a new artificial immune based support vector machine algorithm is proposed for fault diagnosis of induction motors. Support vector machines (SVMs) have become one of the most popular classification methods in soft computing, recently. However, classification accuracy depends on kernel and penalty parameters. Artificial immune system has abilities of learning, memory and self adaptive control. The kernel and penalizes parameters of support vector machine are tuned using artificial immune system. The training data of support vector machine are extracted from three phase motor current. The new feature vector is constructed based on park's vector approach. The phase space of this feature vector is constructed using nonlinear time series analysis. Broken rotor bar and stator short circuit faults are classified in combined phase space using support vector machines. The experimental data are taken from a three phase induction motor. One, two and three broken rotor bar faults and 10% short circuit of stator faults are detected successfully.