Research On The Predictive Control Model Of Wastewater Treatment Based On Neural Networks

China’s economy has developed rapidly in recent years, the resulting is that living waterand industrial water consumption grows rapidly, so that wastewater increases accordingly,the shortage of freshwater is exacerbated by above mentioned two aspects---usage andpollution. So the wastewater treatment is becoming increasingly important, but wastewatertreatment is a high energy consumption and high cost industry, at the same time, the qualityof wastewater treatment will also have direct influence for our living environment. So apredictive control system, which can improve the technical level of wastewater treatmentplants and production efficiency and the quality of discharge water, is urgently needed. It’simportant in river conservation and sustainable development.This paper is based on ASM1platform, using the Elman neural networks to build amodel for wastewater biological treatment process of activated sludge to predict control. Themain contents as below:(1) Comparison of the existing wastewater treatment control technologies, analyzing BPneural networks and Elman neural networks’ advantages and disadvantages for the wastewatertreatment of predictive control, so that to find out which model is more suitable.(2) Improve the existing Elman neural networks, leads into the feedback of output layernodes and adds the feedback of link layer nodes, heighten the performance of Elman neuralnetworks. It adopts the additional momentum method to improve the ability of over the localminimum value of training Elman neural networks.(3) Building a predictive control model of the aeration tank based on Elman neuralnetworks, this model includes two neural networks， and one is neural networks which use topredict the future output as Neural Network Identifier (NNI). Another is a Neural NetworksController (NNC) for output control sequence. And also use this model to predictive andcontrol the DO concentration and pH value.The simulation shows that the Elman neural networks predictive control system hasachieved satisfactory results in the accuracy and feasibility, it owns important significance inimproving the predictive control level of wastewater treatment.