Research On NN-HSIC Strategy Based On The MEA In The Inverted Penduium System

Intelligent Control is a rising study field which has been developing with the progress of the computer technique applying to classical control theory and modern control theory. Compared with the traditional control theory, intelligent control theory has the bigger fitness to the complexity of circumstance and assignment. It is not only the constructed model, but also has multilevel description precision. It developed self-study, self-adaptive and self-organizing conception, so it can be applied in more extensive field.The inverted pendulum is a kind of typically high ranks, nonlinear, strong-coupling, uncertain and absolutely unstable system. It can be stabilization only when the effective control method will be applied. The inverted pendulum system is a typical experimentation equipment to value control theory algorithm and its control methods and ideas have comprehensive purposes to the general industry process, so the research of its control methods has big significance.At first, through the deeply analysis of neural networks(NN) and mind evolutionary algorithm(MEA), their idea souls, mostly algorithms, features and applying develops, the paper rises one idea that the shortages of neural network can be remedied by the mind evolutionary algorithm. Use the MEA, the weights of NN evolve to avert the local least value and enhance the convergence speed. Then, thispaper gains simulating human intelligent control(HSIC) rule of the inverted pendulum from the analysis of the system.Finally, this paper combines the ameliorative NN and the simulating humancontrol theory and rises one new intelligent control strategy-Neural NetworkHuman Simulating Intelligent Control Strategy Based on Mind Evolutionary Algorithm. Then, this control strategy applies on the inverted pendulum system and gets excellent control result.This strategy firstly uses mind evolutionary algorithm to confirm control rule parameters and the values of NN weights. Then, through the training of NN, the control rule parameters can adjust on-line by the trained NN and the training time is short. Through applying to the inverted-pendulum system, the good control result proves the strategy is easy and available and the control result is good.