Research upon Multistage Optimal Control by Wavelet Neural Network

For having stronger learning and generalizing power of functions, wavelet neural network (WNN) can solve multistage optimal control problem. In the course of solving, optimal control law was fitted by using WNN, and a Lagrangian function was constructed to translate into optimization problem from optimal control one. A weight factor was introduced to regulate tradeoff between control system and fit performance by utilizing WNN from the state space to the action space, and then the optimal control performance was reached. Simulation example shows that WNN can solve the multistage optimal control problem better, and different value of weight factor affects the simulation result