A Class Of Nonlinear Systemo F Neural Network Fuzzy Control

Combined with fuzzy system and neural network can also has two kinds of advantagesand overcome their shortcomings, improving study's skills and ability of system, and also isan important kind of the growing intelligent control method at present. In the fuzzy controlreference neural network technology, solve the membership function and the fuzzy rules inthe optimal design control field have a wide range of applications. This paper mainly topredict the fuzzy control, inverse fuzzy control reference neural network technology wasstudied, in addition to fuzzy neural network algorithm proposed new the organization. Theconcrete work of this paper mainly embodied in the following aspects:1. The research based on neural network predictive fuzzy control. Using the neuralnetwork optimize of the fuzzy controller parameters of fuzzy control. As for delay nonlinearsystem and over fitting phenomenon of neural network, designed a fuzzy control systemcombined with multi step forecasting and optimal control. Used Bayesian algorithm trainedthe neural network, in order to ensure the accuracy of the modeling and improvedgeneralization ability, then combined this algorithm with fuzzy optimization control, learnedand revised the estimated system control value from forecasting model. The results indicatethe method get a better output tracking.2. Learning inverse fuzzy control based on the data algorithm. Fuzzy control use neuralnetwork to drive the fuzzy reasoning. According to some problems appeared in fuzzy modelin inverse control which is the rolling data window online have high calculation and lowaccuracy control model, put forward a new method that is the inverse fuzzy learningalgorithm based on data, and the algorithm is applied to establish the fuzzy model of inverse.First, the modeling data in time and space of adjacent specific, the accumulation of data fromthe system to find out and the current mode matching input data, ensure the accuracy of thecontrol model and greatly reduce the computation time, then the adaptive algorithm onlineadjust the system model parameters, realize real-time tracking control of nonlinear system.This method increases the system control precision and computational efficiency.3. Research the neural network self-organizing fuzzy control. Control input space isdivided by fuzzy rules which is remembered with neural network. A new self-organizingalgorithm for fuzzy neural network is proposed, which can revise the rules of the fuzzycontrol and optimize the structure of fuzzy neural network based on input and output data.Through the analysis of the nonlinear function approximation of the sewage treatment systemand water quality model, the results indicate the self-organization method of the sewagetreatment has a good prediction and effectiveness.