Research on prognostic for switch power based on HMM

The fault prediction technique plays a vital role in enhancing security, reducing the life cycle cost and improving maintenance support efficiency for equipment. The hidden Markov model is one of fault prediction techniques to solve the prognostic problem on buck Switch Power. The health degeneration process is analyzed in detail, and the output ripple voltage, inductance current and output power are selected as monitoring parameters. Each state of sample series is used to train HMM, and then the observation sequences of signals to be measured are tested by the model above. So we can get the likelihood possibility of signals to be measured. Experimental results show that this method can accurately predict switching power supply condition.