Research On Intelligent Operation Strategy Of Wastewater Treatment Process Based On Data-driven

With the rapid development of society and economy,the amount of sewage discharge has increased correspondingly,which has brought more and more operating pressure on sewage treatment plants.In order to meet the needs of sustainable development,the pollutant discharge standards of sewage treatment plants have also been improved accordingly.In this case,many sewage treatment plants use excessive dosing and excessive aeration in order to ensure that the effluent reaches the standard,resulting in unnecessary waste of energy and materials.In order to achieve the goals of stable compliance and energy saving and consumption reduction,effective optimization management and decision-making mechanisms are needed to rationally allocate energy and material resources.However,sewage treatment is a typical non-linear,multi-variable,strong coupling system,which is highly complex and time-varying,and its optimal operation management faces many difficulties and great challenges.The management mode based on the mechanism model relies on the simulation of physical processes and biochemical reactions,with numerous parameters and complex operations,it’s hard to deal with the time-varying nature of sewage treatment scenarios,and therefore it’s difficult to achieve timely and accurate optimized operation management.In recent years,advanced computing technologies such as neural networks have provided data-driven solutions to overcome some industrial problems.They start from the perspective of statistical learning and dig out potential laws from massive amounts of data.To this end,this paper uses a data-driven method to build a hybrid neural network model to predict the quality and quantity of influent and effluent water from sewage treatment plants,and provide a basis for decision-making for the compliance of the sewage treatment process on the premise of meeting the discharge standards.Optimize the specific sewage treatment system to reduce energy and material consumption.The main work of this paper is as follows:Firstly,construct a multi-source data acquisition and a basic database.The database includes the relational database and the original database of the sewage treatment plant.Based on the analysis of urban sewage sources,the influencing factors of sewage treatment business volume are divided into three main aspects: economy,population,and weather,and various small factor data are extracted from the three main modules to form a relational database.The original database of the sewage treatment plant includes influent water flow,influent water quality,effluent water quality and energy consumption,material consumption,etc.Secondly,establish a new GRA-CNN-LSTM model for predicting sewage traffic.Base on the correlation data,the business volume of the sewage treatment plant is predicted,including the influent water volume and the influent water quality.CNN-LSTM is a model that establishes the relationship between various factors and sewage traffic.In addition,GRA is used to screen weakly correlated data to ensure the operating efficiency of the model.Experimental results show that the GRA-CNN-LSTM model is better than other common models such as BP and MLR,with the prediction accuracy of 99%.Thirdly,establish a mixed PCA-BPNN model based on deep neural network.It is trained by a large amount of real historical data from sewage treatment plants,and can be used to predict sewage effluent water quality parameters.The PCA method is used to perform principal component analysis on the original data,and the BPNN model is used to predict and evaluate the efficiency and stability of the PCA-BPNN and on a real data set.Genetic algorithm was introduced to optimize the energy consumption and material consumption under the multi-objective effluent parameters,and establish an optimization plan.Based on historical data of two different types of processes of sewage treatment plants,each of the 10 typical influent parameters was screened,and the two sewage treatment plants were optimized through feedback adjustment and iteration.The optimized results were compared with the original data.According to the calculation,the results show that the total energy and material costs have been reduced to varying degrees,among which the power consumption is reduced by up to 23%,and the material is reduced by up to 60%.Finally,a specific operation plan is proposed through the optimal value and the internal equipment of the sewage treatment plant.Through the establishment of multi-source data collection and the construction of basic database,this paper constructs a prediction system based on gray correlation coefficient to study the prediction of the quality and quantity of influent water of sewage treatment plant through correlation data.Through the establishment of wastewater treatment process simulation methods and the analysis and integration of genetic algorithms,an optimization decision-making system is constructed to study optimized operation strategies for reducing energy and material consumption under the premise of ensuring that the effluent quality reaches the standard.The proposed energy-saving and consumption-reducing management method is not only a new solution for optimized operation and management of sewage treatment,but also beneficial to economic and social development.