Developed Based On Fuzzy Neural Network Control Of The Smart Valve Positioner

Pneumatic control valve is a kind of important implements in industrial process control, valve locator is the major attachment of control valve, it can improve the development property of valve apparently, raises flexibility, speed and the precision of control. Since valve locator is a system, the key work of this paper lies in the control part of valve locator.To develop the intelligent valve locator of high performance, this paper makes the work of some following aspects mainly.Firstly, the article expounds the working course of valve locator.Secondly, on the foundation of each component elasticity of ample consideration, it establish the dynamics model of pneumatic control valve, the mathematics model of controlled object and deduce the Transfer function of open loop, and find out the factor which causes parameters' change.Thirdly, it sums up and summarizes the virtue and shortcoming of fuzzy control and nerve network control, since fuzzy control has certain complementary with nerve network control, therefore, the combination of fuzzy control and nerve network control are necessary to development. Fuzzy nerve network control hasn 't only fuzzy control's robustness, structural distinct but also the nerve network control's study and memory.Fourthly, for fuzzy nerve network controller, the inference rule of a groupof optimization has the most important influence for its performance. How to get the rule of optimization becomes one of most important works. For the system that has adequate inputs and outputs sample knowledge, have various methods to getting. For instance, DCL ( differential competitive learning) algorithm can draw fuzzy control rule from input and output data. But to the system that inputs and outputs sample knowledge lack or do not perfect, DCL algorithm be unable to help. Reinforce based fuzzy nerve network controller of learning algorithm (RBFNNC), rely on the fuzzy sum of errors of systematic state and appraisement system only to strengthen type factor, delete and optimize to get a group of effective control rule.Fifthly, According to the condition of parameter change, it uses RBFNNC controller and PID controller to carry out emulation experiment respectively. By contrast, can discover that the performance index of the systematic output curve of application RBFNNC controller is better than the output curve of the system of application PID controller obviously.