Research And Application On Process Modeling And Control Of Water Processing Based On RBFNN

Water pollution is an urgent problem in the world at present; and people come to have an agreement to meet water demand by water treatment. The technology of activated sludge is an economic and simple and convenient method to improve water quality.To exactly control DO in the biochemical pool can get good treatment results. But the biochemical reaction process is a complex biochemical and nonlinear system, in the real water treatment, the stream is fluctuant, it is very difficult to exactly control DO. In order to resolve this problem, the paper proposes a new control method-IMC based on radial basis function neural network (RBFNN) to accurately control DO, and applies this control method to a water processing system of Lv yuan pharmacy company in Dalian. The main research contents of the thesis are as follows:First, a lot of research is done on the waste water treatment system, and the control object and control strategy are selected. In order to model the system, a lot of research is done on the RBF algorithms, and a new method that combines the improved subtractive clustering method and dynamic nearest neighbor clustering algorithm is brought forward. It greatly improves the capability of approaching and computing speed, realizing adjusting network parameters.Then, through the analysis on characteristics of biochemical reaction process, using the data collected, the RBFNN is successfully applied to build up model. Through the model, we can know in advance the tendency of the states change of the biochemical reaction process under given conditions.Last, based on RBF neural network's advantage of linearity in parameter, an RBF network based nonlinear internal model controller (IMC) is designed, and is applied to biochemical reaction system. At last, simulation experiments are carried and the hardware design method is given.Through research on the control method based on neural network, this paper attempts introducing the technology of neural network intellectualized control into waste water treatment. The results simulation shows that IMC system is well self-learning, adaptive and robust, and can realize accurate control on nonlinear biochemical reaction processing.