一种新型的神经网络预测控制方法

预测控制算法的核心是预测过程中的滚动优化，滚动优化方法选择是确定其是否实用的关键，针对这一特点，在此提出了一种以径向基函数（RBF）神经网络为多步预测模型的非线性预测控制算法。算法采用RBF神经网络建立系统预测模型，并以微粒群优化（PSO）算法作为滚动优化算法，用来实现在有限时域内对控制序列的寻优，提高了优化过程的收敛性和求解精度。仿真结果表明了算法的有效性和高效性，获得了良好的控制效果。

Novel Method of Neural Network Predictive Control

The rolling optimization is the core of predictive control algorithm in the predictive process, and the selection of rolling optimization method is a hinge to practice. Aiming at the characteristics, a nonlinear predictive control algorithm taking the radial basis function （RBF） neural network as multi-step predictive model is presented. The RBF neural network is adopted in the algorithm to establish the system model, and the particle swarm optimization algorithm is applied to performing the online nonlinear optimization to achieve the optimization of control sequences in the limited time domain, which improves the convergence and accuracy of the optimazation process. The results of the simulation show that the algorithm is valid and has got a good control effect.