中文摘要：

针对传统的控制理论对实际的工业生产过程中的被控系统，特别是具有强非线性的系统控制效果不是很理想，而应用非线性模型预测控制算法能够较好解决非线性系统的控制问题，提出了一种基于回声状态网络（EchoState Network，ESN）模型进行非线性系统辨识和粒子群优化（Particle Swarm Optimization，PSO）进行滚动优化的非线性模型预测控制系统的算法。ESN能够很好地辨识非线性系统，其计算时间、数据训练和稳定性相对于传统递归神经网络有了较大进步，PSO具有全局优化和较快的寻优速度。针对典型化工非线性对象连续搅拌槽反应器（Continue Stirred TankReactor，CSTR）的仿真实例表明，此模型在预测控制优于BP和PSO结合的非线性预测控制，以及传统的PID控制，证明了该算法运用于非线性模型预测控制中的有效性。

英文摘要：

To the problem that the control objects in practical industry processes are nonlinear systems, and the traditional control theory can not deal with them perfectly, the nonlinear model predictive algorithm is introduced. The algorithmn of nonlinear model predictive control system based on the echo state network （ESN） model and the particle swarm optimization （PSO） is proposed. The ESN can identify nonlinear system perfectly, and has larger progress in computing time, data training and stability compared with the tradi- tional recursive neural network. The PSO algorithm has the global optimization and faster speed for optimum. The simulation results of continue stirred tank reactor shows that it is significantly superior compared to the neural networks based prediction control and traditional PID control, and the effectiveness of it in nonlinear model predictive control is proved.