Multi-Time Scale Hierarchical Model Predictive Control For Municipal Wastewater Treatment Process

Due to the rapid growth of population,the development of industry and agriculture,and the aggravation of water pollution,all countries in the world are facing a serious shortage of water resources.Wastewater treatment provides an effective solution to this problem.Not only can it reduce the discharge of water pollutants and realize the harmless utilization of wastewater,but also produce renewable water sources and maintain the balance between ecology and natural material circulation.Activated sludge wastewater treatment process(WWTP)has the advantages of flexible operation mode,high treatment efficiency,low operation cost and not easy to produce secondary pollution.It has been widely used in municipal wastewater treatment plants to purify water quality.However,the reaction rates of various biochemical reactions in the activated sludge process are different,and the collection devices of each variable have different physical characteristics,which make the variables in WWTP have different sampling periods,and the variables reflect the difference of time scale.It may reduce the operation performance of the closed-loop system and even destroy the stability of the system.Therefore,it is a challenge to design an effective control strategy to ensure the stable and efficient operation of WWTP with the time scale difference.In order to overcome the challenge caused by the difference of time scale of WWTP,a multi-time scale hierarchical model predictive control(MSHMPC)method for WWTP is proposed.Firstly,a multi-time scale prediction model of WWTP based on fuzzy neural network(FNN)is established to obtain the prediction outputs of controlled variables in different time scales.Secondly,a multi-time scale model predictive controller(MSMPC)of WWTP is designed to realize the fast time scale control of controlled variables with slow time scale sampling characteristics,and ultimately improve the control performance of WWTP.Then,an MSHMPC strategy,including a hierarchical control structure and a hierarchical optimization algorithm,is designed.The controlled variables of WWTP with different time scales are tracked by different hierarchical controllers,which futher improves the control performance of WWTP.Finally,an MSHMPC system for WWTP is designed,which realizes stable operation of WWTP under different time scale sampling information,and obtains satisfied control performance.The main research work in this thesis includes the following points:1.The design of the multi-time scale prediction model for municipal wastewater treatment process.A multi-time scale prediction model of WWTP based on FNN is designed to obtain the prediction outputs of controlled variables in different time scales.First,two FNN models are constructed to predict dissolved oxygen(DO)concentration and nitrate nitrogen concentration,respectively.Second,FNNs are pre-trained using historical data with different time scales to initialize the prediction time scales of FNNs.Third,an online updating algorithm based on sliding window mechanism is designed to correct the parameters of FNNs.Finally,the experimental results on the benchmark simulation model No.1(BSM1)show that the proposed multi-time scale prediction model can predict the system outputs with different time scales,and has high prediction accuracy.2.The design of the multi-time scale model predictive controller for municipal wastewater treatment process.A multi-time scale model predictive controller(MSMPC)is designed to control the controlled variables with slow time scale sampling characteristics in the fast time scale to improve the operation performance of WWTP.First,a multi-time scale model predictive control structure is constructed to achieve the accurate tracking of set-points in fast time scale.Second,a multi-time scale model prediction controller,including a fast time scale prediction model and the gradient descent algorithm,is designed to calculate the control law.Third,the stability of MSMPC is proved theoretically.Finally,the proposed MSMPC is simulated and analyzed in detail on BSM1.The experimental results show that MSMPC has suitable operation performance in terms of control accuracy.3.The design of the multi-time scale hierarchical model predictive controller for municipal wastewater treatment process.A multi-time scale hierarchical model predictive controller(MSHMPC)is designed to achieve the accurate tracking of set-points with different time scales.First,a hierarchical model predictive control structure is constructed.The controlled variables in different time scales are tracked by different hierarchical controllers.Second,a hierarchical optimization method is designed to solve the optimal control problem and obtain a reasonable control law.Third,the stability of MSHMPC is proved theoretically.Finally,the proposed MSHMPC is simulated and analyzed in detail on BSM1.The experimental results show that MSHMPC has satisfactory operation performance in terms of control accuracy.4.The design of multi-time scale hierarchical model predictive control system for municipal wastewater treatment process.In order to verify the effect of MSHMPC in practical WWTP,an MSHMPC system for WWTP is developed.First,the functional modules of the system are determined according to the demand analysis.Second,the functional modules of the system are encapsulated to complete the integration of MSHMPC system of WWTP.Finally,the MSHMPC system of WWTP is applied to a wastewater treatment plant in Beijing.The practical application results show that the system can improve the operation efficiency of WWTP.