Application Of ASM-CFD Model And ADABOOST Algorithm In MBR Simulation

The membrane bioreactor(MBR)technology as a new technology of wastewater treatment in recent years,with stable water quality,not subject to site constraints,easy to achieve automatic control,get more and more attention and concern of scholars and experts,the development prospects are also increasingly being optimistic about the related field personnel.However in the whole system of MBR that includes complex physical changes,chemical changes and biochemical reaction.Characteristics of water quality simulation reactor how to efficiently and accurately predict and find the operation conditions and control parameters of flux suitable are concerned problem in the industry,and it is the recent research focus of MBR.The classic model of activated sludge(ASM),although able to describe the solute concentration change in solution affect microbial biochemical reaction conditions from a certain level,but it describes an average level change simulation of MBR reactor.It is not good to simulate the change of the liquid concentration in the specific location of the MBR reaction.And Simple computational fluid dynamics(CFD)method is used to simulate the hydrodynamic environment of the whole MBR.It will also ignore the influence factors of biological response on water quality change.The simulated water quality is not ideal.,so this paper put forward the ASM model and CFD method to combine research,and the introduction of computational fluid dynamics The mathematical model and computer simulation are more popular in FLUENT software.The distribution map of solute concentration is basically consistent with the actual situation of the theory,and has certain feasibility.In addition,it has found that in the model study of intelligent simulation prediction for MBR membrane fouling and membrane flux,the traditional BP neural network has a good fitting effect to a certain extent,also can establish the nonlinear relationship between any input and output,but for the whole MBR system,the operating conditions and the exact control parameters,not only can save the cost,but also get more out of water.So,this paper introduces the AdaBoost algorithm based on the BP model,the reinforcement learning method to the traditional model has been strengthened,the AdaBoost_BP prediction model is obtained,and the longitudinal experiment is compared with the single BP neural network.In addition,Based on the idea of iteration and reinforcement,this paper introduces another excellent algorithm of machine learning,that is,gradient boosting decision tree,and applies it to the intelligent prediction of MBR membrane flux,retraining and modeling experimental data,and doing a transverse experimental comparison with AdaBoost model.Show that comparing experiments,the enhanced BP neural network and the gradient boosting decision tree have good prediction ability,and the performance of both of them is quite good.And the precision is a lot better than the traditional single BP,It can well complete the prediction of MBR membrane flux,which has a certain promotion effect.