ZHU Yonghong, ZHANG Gaohui, XIA Li, CHENG Xien, LI Jie

(Jingdezhen Ceramic Institute, Jingdezhen 333403, Jiangxi, China)

Abstract: In view of the backward control method of ceramic shuttle kiln at present, a PID parameter optimization control method based on fuzzy depth confidence network is proposed in this work. At the same time, the model of fuzzy depth belief network is designed, while the structure of the controller and the corresponding algorithm are provided. The improved LM algorithm is used to estimate the network parameters to reduce the sample data storage space and computational complexity. In this method, the temperature of ceramic shuttle kiln is controlled by collecting real-time temperature data and on-line adaptive adjusting PID controller parameters with the fuzzy depth belief network. Finally, the method is applied to the ceramic shuttle kiln model to verify the control effect. As compared with the traditional PID controller, the controller designed in this study results in the maximum absolute error of tracking the shuttle kiln ideal temperature reduced by 19%. Therefore, the method proposed in this paper is effective and feasible.

Key words: ceramic shuttle kiln; fuzzy neural network; depth belief network; PID control; LM algorithm