Early Warning Method Of Effluent BOD Based On Modular Neural Network

In recent years,water pollution caused by industrial activity is particularly concerned in environment protection.Wastewater treatment is an important approach to effectively solve water pollution and realize the recycling of water resources.Biochemical Oxygen Demand(BOD)is an important variable to assess the content of biodegradable organic matter in water.Excessive discharge of effluent BOD from wastewater treatment plants will cause eutrophication of water.The design of effluent BOD concentration early warning method can detect and deal with the problem of exceeding the standard in advance,which is significant to avoid the occurrence of water eutrophication and improve the efficiency of wastewater treatment.However,the measurement of BOD concentration is time-consuming,complicated in measurement operation,and unavailable in real time,which brings many difficulties in timely and precisely early warning of effluent BOD in WWTP.Aiming to solve the above problems,this thesis studies an effluent BOD early warning method based on Modular Neural Network(MNN)to realize the timely and precisely early warning of effluent BOD in WWTP.The main research contents of this thesis include the following points:(1)Establishment of a multi-step prediction model based on MNN for effluent BOD.The effluent BOD shows different nonlinear performance when WWTP works under different weather conditions,and a single ANN encounters the decline of the modeling accuracy under this situation.To solve this problem,this thesis investigates the multi-step prediction modeling method based on MNN for effluent BOD concentration.Firstly,a weather oriented task decomposition method is designed,and the effluent BOD prediction task is decomposed according to weather categories.Secondly,sub-networks based on Err-Cor-RBFNN are designed to model the prediction task of effluent BOD concentration under each weather category.Through the design of task decomposition and sub-network in Modular Neural Network,a multi-step prediction model of effluent BOD concentration is established.Finally,the simulation results show that the designed multi-step prediction model based on MNN for effluent BOD can divide the BOD prediction tasks according to weather categories effectively,model the divided prediction task by sub-networks accurately,and predict effluent BOD concentration precisely.(2)Design of a hierarchy early warning evaluation method.Aiming at the problem of alarm delineation and early warning evaluation of effluent BOD concentration,a hierarchy early warning evaluation method based on threshold and trend evaluation is designed to perform early warning evaluation on the multi-step prediction results of effluent BOD concentration.Firstly,according to the relevant standards of sewage discharge,the alarm threshold of effluent BOD is determined.Secondly,combining the trend information extracted from the prediction results with alarm thresholds of effluent BOD,a comprehensive hierarchy alarm strategy is built.Finally,the designed early warning method is used to evaluate the multi-step prediction results of effluent BOD concentration for early warning.The simulation results show that the designed early warning method can accurately alarm the potential problem.(3)Software development of the effluent BOD early warning method based on MNN.Aiming at the problem of software application for the proposed effluent BOD concentration early warning method,an intelligent effluent BOD early warning software based on the MATLAB GUI function was designed and developed in this thesis.The software design mainly includes user management part and intelligent warning part,including user registration function and login function,software introduction function,data import function,intelligent effluent BOD concentration warning function and help function.The software is easy to use and maintain.It can carry out adaptive training according to the actual needs of users in wastewater treatment,meet their individual needs,realize rapid early warning of effluent BOD concentration and visualize the prediction results,which has practical significance and application value to ensure the stable and efficient operation of wastewater treatment plants.