MBR Membrane Fouling Based On Artificial Immune System Under Big Data Platform

Membrane bioreactor(MBR)as a new technology in the field of wastewater treatment,has a good development prospect,and has been duly noted by experts and scholars in related fields.However,membrane fouling has seriously affected the development of MBR technology,how to regulate the operating conditions and parameters of the system,slow down the membrane fouling phenomenon,and how to establish the simulation model of membrane fouling have become hot research topics.The purpose of this paper is to establish the non-linear relationship between the factors affecting the MBR membrane fouling and the membrane flux characterizing the degree of membrane fouling,and thus to accomplish the membrane flux prediction.Before the simulation prediction model is established,it is necessary to process the data generated during the operation of the membrane bioreactor system and extract the required data for building the model.With the development of MBR,the system will produce more and more data,and therefore this paper introduces a big data distribution system to complete the storage and analysis of the data related to membrane fouling.Using Hadoop commonly used components to build a complete data processing model,and completing the data processing automatically with the shell script.Including using the Flume components to complete the data collection,using the MapReduce program to complete the data cleaning,using the Hive components to build data warehouse and analysis the data in the table,and using the Sqoop components to import the result data into the relational database.Theoretical studies have shown that neural networks can approximate arbitrary continuous functions with arbitrary precision,and can establish a complex non-linear relation model between input and output,which have been used widely in the prediction field.In this paper,Elman dynamic neural network is used to build the simulation prediction model,which belongs to the local regression feedback network,with the associative memory function,and the prediction accuracy is superior to the traditional BP neural network.However,because of the Elman neural network still adopts the gradient descent method of feedforward network to train the network,there are some problems such as slow network training speed,easy to fall into the local minimum point,sensitive to the initial of network weights and thresholds,and difficult to determine the network structure.Aiming at these problems,an adaptive immune genetic algorithm based on the concentration of antibodies was introduced to optimize the Elman network model,and thus the dynamic recursive neural network can be constructed and designed automatically.The experimental results show that,the Elman network optimized by immune genetic algorithm is superior to the un-optimized Elman neural network,and it is found that the Elman network optimized by immune genetic algorithm is more effective than the model optimized by the traditional GA algorithm.Moreover,the prediction accuracy of the network has been improved greatly and the optimized network's prediction is relatively stable.