Study On The Multi-intelligence-algorithm Application In Multi-objective Optimization In Wastewater Treatment Process

Wastewater treatment process refers to a dynamic non-line biological reaction process characterized by time-variety and great time-lag.In addition,this process is often affected by various variables simultaneously.Due to these factors,related researches on on-line monitoring and process optimization of wastewater treatment process are also being limited.Nevertheless,with the development of artificial intelligence in recent years,intelligent algorithm has been applied to water quality prediction and process optimization of wastewater treatment process,which achieved some satisfying results.In consideration of the increasingly stringent effluent discharge standards,it undoubtedly becomes a key problem to be solved in the field of wastewater treatment to explore how to reduce the energy consumption at the same time of improving the effluent quality.This study analyzed the multi-objective optimization problems existing in activated sludge process,ANAMMOX and denitrification collaborative nitrogen and carbon removal process,and systematically studied the thinking and methods of constructing multi-objective optimization model and soft sensor model based on a comprehensive understanding of the application status of wastewater treatment simulation benchmark model?BSM1?,anammox technology?ANAMMOX?,ANAMMOX and denitrification collaborative carbon and nitrogen removal technology and intelligent algorithm.Therefore,this study is of great significance for on-line monitoring and process optimization of anammox water quality,as well as the tradeoff of energy saving and emission reduction in activated sludge process.The main contents and conclusions are as follows:?1?Automatic monitoring system for wastewater treatment was developed in this study.According to the control requirements of the industrial wastewater intelligent control pilot system,the selection of related software and hardware equipment and the configuration of the operation interface of the wastewater treatment automatic monitoring system were completed.?2?Based on principal component analysis?PCA?,particle swarm optimization?PSO?,genetic algorithm?GA?and Elman artificial neural network?ANNe?,the prediction model of effluent quality in ANAMMOX process was constructed,and the initial weight and threshold of ANNe was optimized by hybrid intelligent algorithm based on PSO-GA.The results showed that the mean average percentage error of the optimized model for the prediction of ANAMMOX effluent NH₃-N concentration,TN removal concentration and gas production were 0.10,0.033 and 0.114,respectively,and the convergence rate distribution of TN removal prediction and gas production prediction was improved by 15.88%and 26.61%,respectively.Taking the prediction of gas production as an example,compared with the original ANNe,the prediction accuracy was improved by 32.9%,the stability was improved by 83.7%,and the prediction accuracy increased with the increase of the number of training samples,which indicated it has the great potential of engineering application.?3?A multi-objective optimization model of ANAMMOX and denitrification collaborative carbon and nitrogen removal technology was established based on least square support vector machine?LSSVM?and fast non-dominant genetic algorithm?NSGA-??.The results revealed that the correlation coefficients between the predicted and real values of NH₃-N removal,TN removal and COD removal in the test phase of the soft sensor model based on LSSVM were0.997,0.969 and 0.990,respectively.When the influent COD/TN,p H and NH₃-N/NO₂-N were0.24,7.44 and 0.928,the corresponding removal rates of NH₃-N,COD and TN were 90.15%,75.25%and 89.97%respectively,indicating the denitrification process has better carbon and nitrogen removal performance.?4?A multi-objective optimization model of the benchmark simulation model BSM1 was constructed to optimize the effluent quality and operation energy consumption.The results showed that the effluent quality under the optimization strategy was improved by 72.2%,while the operation energy consumption was reduced by 5.68%compared to default strategy.And for all the influent quality and control conditions,the optimization strategy was always better than the original strategy in terms of effluent quality and operation energy consumption.