Multi-stage Upgrading And Renovation Of Sewage Treatment Plants:Model Research And Application

Waste water treatment plant construction and upgrading is an important way to promote economic environment.At present,most of the construction of waste water treatment plants rely on the comprehensive subjective judgment of the decision-making team,without systematic analysis and comprehensive comparison,resulting in unreasonable technology and construction scale,resulting in double losses of economy and environment.In order to carry out the improvement,modernization,and building of facilities for treating wastewater plant more reasonably,the cost model is developed using a multi-stage approach that incorporates genetic algorithm optimization and backpropagation neural network（GA-BP-ANN）techniques.The process selection model for the second stage is determined through a process comparison model that is based on fuzzy two-stage stochastic trusted constraint programming.Finally,the operation control analysis of the final stage of the construction of the waste water treatment plant is carried out by the Bayesian optimization of long and short time memory（LSTM）,and the decision scheme of the multi-stage upgrading of the waste water treatment plant is obtained.The primary topics and findings of the article are summarized as follows:1.The cost model based on GA-BP-ANN is constructed.Based on incomplete information and unclear relationship between data lines,this model comprehensively considers investment minimization and environmental benefit maximization,and takes design water quantity,treatment degree（primary,secondary and advanced treatment）,intake BOD₅ concentration and water area as input conditions,and total cost and construction cost as output conditions.By optimizing the weight of each neuron in the neural network through genetic algorithm,a neural network with good performance is trained,and the neural network provides an advantageous scheme for the investment and construction of waste water treatment plant.2.The study developed a process comparison scheme using a fuzzy two-stage stochastic confidence constraint programming model to select an appropriate wastewater treatment process for the Kunming wastewater treatment plant.The research also provided the best environmental and economic process for other regions to choose their wastewater treatment process based on the first-stage investment scheme.In this scheme,the newly added wastewater from the waste water treatment plant due to population expansion is taken as the fuzzy number,and the credibility is selected as[0.5,0.9]by the language credibility definition.The results show that the credibility constraint can prevent the waste of useful information,and the results are more in line with the strict environmental protection requirements.For the heavily polluted areas,the most suitable treatment process for economy and environmental protection can be selected.3.Two models were developed using a neural network approach based on Bayesian optimization of short and long-term memory methods.Model1,which is a multi-input-single output model,is based on the concentration of biological processes.Model2,which is a single input-output model,is based on time series data.The two models start from the treatment capacity control of waste water treatment plant and combine the ability of the model to predict and control the water quality of the wastewater treatment plant using hardware sensors was evaluated.The results show that:the error of the training set is between[-0.4,0.4]（g/m³）and the error of the test set is between[-0.2,0.15]（g/m³）,which can be utilized to forecast the quality of the treated wastewater that is discharged from the treatment plant.Model 1 can be used as other data to replace the hardware sensor design when it is convenient to obtain,and model 2 can predict the historical concentration obtained.This is highly significant in ensuring the efficient operation of the wastewater treatment plant.