Application of improved wavelet neural network in MBR flux prediction

Membrane Bio-Reactor (MBR) technology plays an important role in modern sewage treatment but the performance of the MBR technology is seriously affected by the membrane fouling. In general, the result of membrane fouling is decline of MBR membrane flux, and the effect of MBR sewage treatment is directly affected by the decrease of membrane flux. In order to predict MBR membrane flux accurately and rapidly, the forecasting model of MBR membrane flux based on particle swarm improving wavelet neural network algorithm (PSO_WNN) was established. In view of the complexity of the MBR membrane fouling factor, in the beginning, the main components of the factors affecting the flux of MBR membrane were analyzed. The important factor is extracted as the input of the PSO_WNN prediction model, and the membrane flux is used as the output. Then, the PSO_WNN simulation model is established, and the prediction results are obtained by using the model. By comparing the predicted data and experimental data, the predictive accuracy of this algorithm is high on the membrane flux, and compared with the BP neural network model, the comparative results show that the PSO_WNN forecasting model has higher predicted accuracy.