Research On Data-driven Intelligent Optimal Decision For Sewage Treatment Processes

As water shortage and water pollution problems become increasingly serious in China,relevant departments have made further requirements for sewage treatment discharge standards,making it imperative to optimize the management of sewage treatment plants.Although the current sewage treatment plants can guarantee stable operation,they mostly rely on manual experience in management and decision-making.The lack of effective evaluation and management plans makes it impossible to flexibly respond to the highly uncertain sewage treatment process and the sharp increase in labor costs.Therefore,it has become one of the important goals of current and future sewage treatment plants to propose an effective intelligent management plan to reduce the uncertainty and reduce labor in the sewage treatment process.However,the management research on sewage treatment in China started relatively late,and the level of intelligent control and optimization is low.Currently modeling and optimizing the sewage treatment process is still a big challenge.For this reason,this paper puts forward an optimized decision-making scheme based on data-driven hybrid neural network model and fuzzy inference system,which is applied to the intelligent management and decision-making of a sewage treatment plant in Chongqing.And put forward corresponding management suggestions based on the optimization results.The main work of this paper is as follows:（1）Data collection and preprocessing:This study takes a real sewage treatment plant as an example,the data collected includes the dissolved oxygen（DO）concentration of the 6 oxic tanks in the sewage treatment plant,the outlet chemical oxygen demand（COD）concentration,the outlet ammonia nitrogen（NH₃-N）concentration,the outlet total phosphorus（TP）concentration,and outlet total nitrogen（TN）concentration.The data preprocessing in this part includes missing value,outlier processing and normalization processing,and data description of 6 parameters and 4 water quality factors.（2）Modeling of sewage treatment process based on hybrid neural network:Taking into account the time series characteristics of the sewage treatment process and the backflow process,this paper proposes a data-driven model based on a hybrid neural network（Intelligent Management Mechanism Based on Hybrid Neural Computing,IM-HNC）to abstract the sewage treatment process.First,adopts a convolutional neural network to construct a feature space with strong expressive ability.Then introduces bidirectional long short-term model to capture the two-way time series characteristics of the sewage treatment process.And based on a series of experiments,it is concluded that the model has good performance and stability.（3）Intelligent optimization decision-making system for sewage treatment based on fuzzy inference:According to the IM-HNC model established,this paper proposes an intelligent optimization system based on fuzzy inference,then obtain the optimal aeration range combination of the sewage treatment system by analysis of means（ANOM）.The experimental results show that when the DO concentration levels of the 6 oxic tanks in the A²O sewage treatment plant are at levels 4,1,2,2,3,and 4,the pollutants after sewage treatment are closer to the optimal value.And based on the optimization results and a series of experiments,proposes three aspects of data-driven,intelligent management and aeration optimization decision-making recommendations for the sewage treatment plant.In this paper,constructs a fuzzy inference system based on gray correlation coefficient through the analysis and integration of the sewage treatment process modeling method and optimization system.To study the optimization strategy of reducing energy consumption under the premise of ensuring the stable outlet water quality.This research is beneficial to improve response ability of changes in effluent quality in the future,and has important theoretical and practical significance for realizing the intelligent management of the sewage treatment plant.