Corrosion Rate Prediction of Storage Tank Based on Particle Swarm Optimization and Least Squares Support Vector Machine

A prediction method for corrosion rate of large storage tanks was proposed based on particle swarm optimization (PSO) algorithm and least squares support vector machine (LSSVM), which utilize the global optimization capability of PSO algorithm to optimize the regularization parameters and kernel parameters of LSSVM. The corrosion rates of storage tanks were predicted by the method, and the prediction accuracy of the model was verified by measured data. The results show that the predicted corrosion rates obtained using PSO-LSSVM were in good agreement with the actual corrosion rates. The mean absolute percentage errors of the predicted results of the tank top, the first tank wall and the tank bottom were 2.265%, 3.077% and 1.18%, the root mean square errors were 0.010%, 0.012% and 0.011%, and the corresponding coefficient of determination were 0.973, 0.982 and 0.976, respectively. So this method can effectively predict the corrosion rates of storage tanks.