The application of pso-lssvm in fault diagnosis of subway auxiliary inverter

Focused on the non-stationary characteristics of the fault signal of subway auxiliary inverter and the fault diagnostic accuracy problem, this paper establishes the fault diagnosis model on the basis of least squares support vector machine (LSSVM) and particle swarm optimization (PSO). This paper firstly extracts the frequency domain energy feature vector by wavelet packet transform, secondly establishes the multi-fault classification based on LSSVM to achieve the fault pattern recognition, and thirdly optimizes the structure parameters of LSSVM by means of PSO, which enhances the global search capability and avoids the blindness of parameter choice. Experiment results demonstrate that the proposed fault diagnosis model not only achieves better classification effect, but also is superior to traditional LSSVM in diagnostic accuracy and generalization ability, which is applicable to the fault diagnosis of subway auxiliary inverter. © 2013 ISSN 2185-2766.