Research of temperature predictive control based on LSSVM optimized by improved PSO for thick steel plate Roller hearth Normalizing Furnace

According to process requirements of Roller-hearth Normalizing Furnace, and non-linear characteristics of the temperature, the paper proposes a new nonlinear system prediction control algorithm instead of the tradition, which the accuracy of model is not high. The new control algorithm uses least squares support vector machine (LSSVM) optimized by improved particle swarm optimization (APSO) to establish the predictive model. This model is simulated and studied by using lots of data acquired from the site. The result indicates that this prediction model based on APSO and LSSVM has higher control accuracy and good application in future.