Application and Research of Elman Neural Network Model Optimized Input by Improved Immune Genetic Algorithm in MBR Membrane Flux Prediction

Abstract

Membrane flux is an important indicator to show and prove the extent of membrane pollution . Our team adopted the Elman neural network to establish the MBR contamination simulation prediction model. By setting up the non-linear relationship of influencing MBR membrane pllution’s factors and membrane flux, to complete the membrane flux prediction. Research shows that the MBR membrane flux prediction model based on Elman neural network still has some problems. After reviewing a large number of literatures, we applies improved adaptive immune genetic algorithm based on antibody concentration to optimize this model. At the same time, we compares the optimization effect of this algorithm with the optimization effect of traditional genetic algorithm. By the contrast and analysis, the Elman neural network model optimized input by the improved immune genetic algorithm is superior to the Elman neural network model optimized input by the traditional genetic algorithm on the prediction accuracy and stability of MBR membrane flux.