The Controller Design And Application Of Neural Network Optimized By Genetic Algorithm

Neural network is a computation model which simulated human physiological mechanisms of neural. It has the ability to non-linear mapping. In ideal circumstances, it can approximate any linear and non-linear relationship. At the same time, it has the self-learning autogenous shrinkage characteristics, and has the strong robustness and fault tolerance. In the past 15 years, neural network technology has penetrated into all fields.In Neural Network, when the learning algorithm of BP trains neural network, it is difficult to achieve high accuracy to time-varying systems, because it need to provide teacher's signals on the network training. In addition, because the BP learning algorithm is based on the gradient descent of this nature, there is the inevitable convergence of the learning process slow, vulnerable to the shortcomings of local minimum points.For BP neural network has some shortcomings and genetic algorithm can converge to global optimal solution and has the strong robustness, this article combines genetic algorithm with neural network so that it can not only plays generalized mapping capabilities of neural network, but also makes neural networks with fast convergence and strong le??ng ability. In order to verify the effectiveness of BP neural network optimized by genetic algorithm, the paper is applied this algorithm to a straight-line stability control of the inverted pendulum and produced a simulation of the inverted pendulum control and experimental real-time control software by LabVIEW language which has some advantages of interface development and data entry, network communications, simple hardware control. It is showed by the simulation that controller design of genetic algorithm optimization of BP neural network is feasible and is able to achieve good stability of the inverted pendulum.