Abstract:

A method based on optimal relevance vector machine (RVM) is proposed to solve the problem of electronic system condition time series prediction. Based on the phase space reconstruction of electronic system condition time series, the RVM regression model is established. A quantum behaved particle swarm optimization (QPSO) algorithm is employed to realize automatic selection of the established model parameters, which adopts cross validation error as the optimization objective function and takes the kernel parameter as the particle position in quantum space. Experimental results show that the proposed method has higher point prediction accuracy and can provide probabilistic predictions, which is conducive to determine the future health status of electronic systems more reliably.