The Application Research Of Neural-network Decoupling Controller In Multi-variable System

For multi-variable system with the characters of non-linearity, heavy delay andstrong coupling (such as the artificial environmental chamber), researchers aredifficult to find an effective control strategy. In the past people often took the PIDcontroller into practice. With the development of the intelligent controltechnologies in recent years, many researchers prefer to use intelligent controltechnology in order to gain more satisfactory controlling effects in some respects.Ref.[1] presents a method called feedforward decoupling of fuzzy control for theenvironmental chamber , which takes both fuzzy-controller and neural-networkcontroller in the control system. The method was proved feasible by an emulatorprogram in Matlab. Yet the Ref.[1] didn't develop a practical neural-networkprogram to apply to the application of environmental controlling. Neither has itdiscussed more efficient algorithm to improve the controller ability.On the base of the environmental chamber control research, this article gavesome characters of multi-variable system controlling firstly, then the theory ofneural-network BP algorithm was introduced detailedly. Two decoupling controlschemes for multi-variable system were presented in the follow. For the firstscheme, the article put an emphasis on discussing how to achieve the C++program of BP algorithm, which can be called by a Visual Basic program in DLLmanner. For the other scheme, the article also proposed a more efficientlyneural-network controller called PID nerve-cell network. The algorithm of PIDneural network was discussed and both it's C program code and emulation resultswere given..