Fault Prognosis Method of Industrial Process Based on PSO-SVR

For latent faults or situations where the pre-failure characteristics are not obvious, fault prognosis techniques are needed. This work proposes a fault prognosis method based on support vector regression (SVR), in which particle swarm optimization (PSO) algorithm is utilized to optimize the parameters to improve the prediction accuracy. The SVR algorithm and grey prediction are tested on benchmark data taken from Tennessee-Eastman process and the "NASA prognosis data repository", and the experiments compare the prediction accuracy difference between the two algorithms.