Gaussian Process based Remaining useful life Prediction for Electric Energy Metering Equipment

Electric energy metering equipment (EEME) will fail in advance not as designed running in extreme environments. A multi-kernel Gaussian process regression model using measurement error data to perceive remaining useful life (RUL) for EEME is proposed. Firstly, the gauss kernel and periodic kernel are used to match the health index trend of EEME under a variety of typical environmental stresses. Furthermore, the Bayesian method and Monte Carlo Markov chain method are used to solve the model, and the Weibull distribution is used to fit the posterior trajectory to get the probability density estimation of the RUL.