摘要
针对最优控制问题的数值求解,提出了一种混合小波神经网络-粒子群(WNN-PSO)算法。算法首先利用小波神经网络的非线性逼近能力参数化最优控制轨迹,将最优控制问题转化为非线性规划(NLP)问题,其决策变量为小波神经网络的参数,然后采用粒子群(PSO)算法优化小波神经网络参数,获得NLP问题的全局最优解。针对Bang-Bang最优控制问题和一个经典的化工过程最优控制问题进行仿真研究,验证了所提出算法的可行性和有效性。
To solve optimal control problems with numerical methods, a hybrid wavelet neural network - particle swarm optimization （WNN-PSO） algorithm was developed. The first step of WNN-PSO was to parameterize the optimal control trajectory based on the non-linear approximation capability of the wavelet neural network. Then the optimal control problem was transformed into a non-linear programming problem where the decision variables are the parameters of the wavelet neural network. Lastly, the parameters of the network were optimized by the particle swarm optimization （PSO） algorithm and the global optimal solution of the NLP was obtained. Simulation study on a Bang-Bang optimal control problem and a benchmark chemical process optimal control problem shows the feasibility and effectiveness of the proposed method.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2013年第3期425-429,共5页
Journal of System Simulation
基金
国家自然科学基金(60974039)
国家科技重大专项资助(2008ZX05011)
关键词
最优控制
小波神经网络
参数化
粒子群
optimal control
wavelet neural network
parameterization
particle swarm optimization
