Predictive Models of Aluminum Reduction Cell Based on LS-SVM

Bath temperature and alumina concentration are two important but hard to measure online parameters of aluminum reduction cell. To this problem, a novel method based on least squares support vector machine (LS-SVM) and chaos optimization is proposed to establish predictive models of the two parameters. This method employs chaos optimization technique to iterate and search in feasible regions so as to find optimal LS-SVM algorithm parameters and corresponding model parameters. The simulation results show that this method has smaller absolute error and relative error than those of neural network method.