DHP algorithm based multi-variable optimal control for cement calcination process

Cement precalciner kiln(PCK) clinker calcination process is a matter of mass transfer, heat transfer, physical and chemical reactions, and more complex multi-variable nonlinear system with more disturbances. In order to reduce energy consumption and to ensure the quality of cement clinker burning, one needs to explore different control methods from the traditional way. In this paper, PCK technology is conducted a detailed analysis, and its model is established by artificial neural network. New controller has been designed to control the model by choosing the appropriate control variables. Dual Heuristic Programming (DHP) is the advanced form of Adaptive Dynamic Programming (ADP) algorithm. Typical DHP structure is consists of three modules: Critic Network, Action Network, and model network. Its Critic network output cost function J's partial derivative to the state variable and therefore have a higher accuracy, with the corresponding its calculation is much more complex. Its purpose is when minimizing the cost-to-go function, one can find the optimal or sub-optimal control signal, so that the discrete-time nonlinear systems to obtain the desired control trajectory. Simulation results show that the controller response time faster, the parameters have small overshoot which help the stability of the actual system operation. DHP approach with multi-variable control of the clinker calcination process, is an effective way and demonstrate the potential of real-time optimal control.