基于双启发动态规划的预分解窑控制器设计

针对水泥预分解窑熟料煅烧过程多变量、多扰动、非线性，难以建立精确的数学模型，实际生产中对工人的生产经验依赖性较强等问题，提出采用误差反向传播(BP)神经网络建立烧成系统数学模型，设计预分解窑双启发动态规划(DHP)控制器。DHP评价网络输出代价函数J关于状态量的偏导数，获得最优或者次优的控制信号，然后由动作网络输出控制信号，使系统得到期望的控制轨迹。仿真结果表明控制器响应时间较快，各参量超调量均不大，有助于实际系统的稳定运行。

Design of cement calcination process controller based on dual heuristic programming algorithm

For the multiple variables, disturbances, nonlinearity and other properties of the cement clinker kiln process, it is very difficult to establish an accurate model of the cement kiln system. It is strongly dependent on the experience of workers in the actual production. The error Back-Propagation (BP) neural network was used to establish the firing system model, and a controller was designed for the kiln based on Dual Heuristic Programming (DHP). DHP critic network output the partial derivative of cost function J with the state to obtain the optimal or sub-optimal control signal. The action network output control actions to control the system to achieve desired trajectory. The simulation results show that the controller has faster response time and less overshoot. These features contribute to the stable operation of the real system.