一类间歇过程的优化及非线性预测控制

基于主产物浓度和反应温度的RBF神经网络模型，使用粒子群优化算法(PSO)求解该间歇反应主产物产率最大化问题，进而得到反应温度优化曲线。利用RBF神经网络建立了反应器冷却水控制温度阶段的预测模型，采用非线性预测控制，并引入了模型误差项，增强了控制方法的鲁棒性和间歇过程的抗干扰性能。利用 Lyapunov 原理对该预测控制算法做了稳定性分析，确定了系统稳定条件下的参数的取值范围。同时编制控制程序在多功能过程及控制实验装置(MPCE)装置上实现了算法的控制，并与以升温速率为基准的特殊PID调节器的控制结果比较，结果证明了基于RBF神经网络非线性预测控制方法的有效性。

Optimization and non-linear predictive control of a class of batch processes

Based on the radial basis function neural network (RBFNN) model of the concentration of the main product and the reaction temperature, an optimization curve for reaction temperature has been obtained by means of a particle swarm optimizer (PSO). A predictive model of a batch reactor was set up at the water cooling temperature. Use of a non-linear predictive control system with a model error item enhanced the robustness and anti-disturbance performance of the model. The value range of parameters was obtained through a stability analysis of predictive control based on Lyapunov theory. The program was set up to realize a control strategy based on a multifunctional process and control experiment (MPCE) system. The proposed optimal control profile is effective and superior to the control result using a special proportional-integral-derivative (PID) regulator.