Research On Intelligent Control Method Of Ceramic Shuttle Kiln Based On Deep Reinforcement Learning

Ceramic shuttle kiln is a kind of thermal equipment which is widely used in ceramic firing process.The temperature control method in the sintering process directly affects the quality of porcelain products.However,so far,ceramic shuttle kiln automation and intelligent degree is relatively low with single control means.Ceramic shuttle kiln temperature is mainly controlled by artificial PID.Its control is not stable and the control effect is not ideal for ceramic shuttle kiln parameters time variability,atmosphere pressure coupling and many other factors.In order to improve the control method of ceramic shuttle kiln and to improve the control effect,an indirect identification method is proposed to identify the temperature model of shuttle kiln online,and then the optimal control method based on deep reinforcement learning is used for control.The specific work and innovations are as follows:1.According to the relevant theoretical basis of heat transfer in the sintering process of shuttle kiln,the temperature system has been selected as the research focus from temperature system,pressure system and atmosphere system in this paper.According to the theory of heat transfer,the heat transfer equation of ceramic shuttle kiln was solved under three kinds of boundaries,and the optimum sintering temperature curve was finally fitted.2.An online identification algorithm for ceramic shuttle kiln is proposed.This algorithm indirectly restores the continuous delay model by identifying the discrete delay model.The innovation of the indirect method lies in that it takes time delay into the model parameters,which has been applicable to both integer and fractional delays,and realizes the accurate real-time tracking of time-varying delay systems under the framework of the least square method.3.An adaptive optimal control method based on deep reinforcement learning has been proposed for ceramic shuttle kiln.Two iteration algorithms such as value iteration and policy iteration for optimal control were given and their convergence was proved.In order to eliminate the external disturbance of the system,disturbance observer and compensation controller based on disturbance observer were designed.Finally,the proposed method was applied to the ceramic shuttle kiln to carry out simulation experiments,and the simulation results show the effectiveness and feasibility of the proposed method.