数据和知识驱动的城市污水处理过程多目标优化控制

Data-knowledge Driven Multiobjective Optimal Control for Municipal Wastewater Treatment Process

城市污水处理过程优化控制是降低能耗的有效手段,然而,如何提高出水水质的同时降低能耗依然是当前城市污水处理过程面临的挑战.围绕上述挑战,文中提出了一种数据和知识驱动的多目标优化控制(Data-knowledge driven multiobjective optimal control,DK-MOC)方法.首先,建立了出水水质、能耗以及系统运行状态的表达关系,获得了运行过程优化目标模型.其次,提出了一种基于知识迁徙学习的动态多目标粒子群优化算法,实现了控制变量优化设定值的自适应求解.最后,将提出的DK-MOC应用于城市污水处理过程基准仿真模型1(Benchmark simulation model No.1,BSM1).结果表明该方法能够实时获取控制变量的优化设定值,提高了出水水质,并且有效降低了运行能耗.

The optimal control is an effective method to reduce energy consumption for municipal wastewater treatment process.However,it is still a challenge to improve the effluent qualities and reduce energy consumption simultaneously for the municipal wastewater treatment process.To solve this problem,a data-knowledge driven multiobjective optimal control(DK-MOC)method is proposed in this paper.First,the expression relationship among effluent qualities,energy consumption and system operation state is established to obtain the operational optimal objective model.Second,a dynamic multiobjective particle swarm optimization algorithm,based on knowledge transfer learning method,is proposed to obtain the optimal set-points of control variables adaptively.Finally,the proposed DK-MOC method is applied to the benchmark simulation model No.1(BSM1)of the municipal wastewater treatment process.The results demonstrate that this proposed method can obtain the optimal set-points of control variables online,which not only improve the effluent qualities,but also reduce the operation energy consumption effectively.