Study on Fault Prediction of Vehicles Synchronous Generator

Arming at the complex electromagneter relationship of synchronous generator, it can work with wrong. Using wavelet packet transform to reduce voice, extracting the unsteady feature of the output voltage. According to LSSVM can make nonlinear prediction and HMM has the explanative ability, the fault prediction model based on LSSVM+HMM is presented., training the HMM by the normal features. Then the fault likelihood of unknown signal feature and the predicted feature calculated by LSSVM can be getted by the trained HMM. Decided the state of generator by comparing the values. Effectiveness of the model was verified from the real example.