Control Study, Based On Fuzzy Neural Network Lag

Time delay,which is commonly found in industrial production, usually cause the quality of control systems deterioration or even instability, at the same time, time-delay systems are often accompanied with a lot of unknown and uncertain factors, with the environment, working conditions and unpredictable changes. Therefore, according to time-delay systems, those that based on quantitative mathematical model of the traditional control method has effective control impossibly, we must search for new control strategies.This paper focuses on the control strategy of time-delay systems based on fuzzy neural network, mainly including identification and prediction by using neural network, the study of fuzzy neural network control. The basic idea is to use neural network predictor to identify and predict the time-delay object, then calculate the deviation and the deviation change, in accordance with the output value of neural network predictor and a given value. And as two inputs of the fuzzy neural network controller, the time-delay object is controlled by using the conrol strategy of fuzzy neural network.The identification and prediction by using neural network is based on approximation and learning ability, and can be studied to extract mathematical model to be a more accurate expression of the characteristics of the actual object from the input and output data received, requiring less priori knowledge. The methods have static neural network and dynamic neural network; Elman network is used to identify and predict the actual object and achieves good results, unlike the static neural network which is complexity required to identify the structure and can be difficult to train the model response to the actual object.In this paper, the control strategy based on fuzzy neural network overcomes the shorts of fuzzy control which the fuzzy rules are uncertain, arbitrary, subjective and difficult to obtain effective knowledge of rules. The control strategy can overcome the shorts that the neural network should not make good use of existing experience of knowledge. Fuzzy neural network controller has the standard model of fuzzy neural network structure and the T-S model of fuzzy neural network structure. The initial weight value of fuzzy neural network usually only set zero or random numbers at the training stage, so, it results in bad initial state, and even cause system instability. The optimization technique of Particle swarm extracts the control law from the sample data firstly, which was a basic response to the system requirements for fuzzy neural network controller. In order to control time-delay systems effectively, back-propagation algorithm is uesd to continuously optimize the structure parameters of fuzzy neural network controller real-time.In order to facilitate the study based on the control strategy of fuzzy neural network, the paper also build a real-time control system by using OPC technology to communicate between Forcecontrol and MATLAB. And it is applied to the control system of the boiler water temperature.