The Application And Research Of Intelligent Control In Wastewater Treatment

With the rapid development of economy, the environment problem have become increasingly prominent, especially the change of water environment, it poses a very bad impact on people's health both physically and mentally, and it has become a constraint to the sustainable development. So sewage treatment has become the top problem of city construction. In this article, briefly introduce some research work on intelligent control of sewage treatment on the basis of existing theory.In this article firstly propose soft-sensing technique to the problem that key parameter of water quality can not be measured online, and then considering characteristics of sewage treatment: non-linear and difficult to set up its model, I apply the neural network to set up the soft-sensing model of sewage treatment. In the final chapter I propose that model reference adaptive control (NNMRAC) method based on the neural network could be applied to realize the closed-loop control of sewage treatment.In the process of applying neural network to set up the soft-sensing model of sewage treatment, I set up the soft-sensing model of sewage treatment based on BP network, and Elman network respectively, and make simulation through sample data, and analyze the performance of networks. Then, in order to solve the problem that the topological structure of neural network can not be ascertained and the network lacks of learning algorithm, I use genetic algorithm to optimize the weight and the threshold of Elman networks. From the simulation results, we can see that the generalization ability is well, and the accuracy is high.In the final chapter I set up sewage treatment system that based on neural network model reference adaptive control(NNMRAC). The system does not depend on the accurate mathematics model of the controlled object, however it can adjust the controller if the parameter of controlled object and environment changes. In the identifier part of the system I choose Elman network, and in controller part of the system I apply RBF network, and then I use genetic algorithm to optimize the weight and the threshold of Elman network and RBF network, from the simulation results, we can see that the pace of study is fast, the training accuracy is high.