摘要
针对国际污水处理基准仿真1号(BSM1)模型,对第五池中氨氮浓度采用串级控制,对第二池中硝态氮浓度采用单回路控制。利用神经网络建立氨氮浓度和硝态氮浓度的设定值与能耗的神经网络模型。同样,利用神经网络建立氨氮浓度和硝态氮浓度的设定值与出水水质合格率的神经网络约束模型。优化问题的目标函数和约束条件均通过神经网络建立,并利用变搜索系数(NLJ)算法求解该优化问题。将最优解分别作为氨氮和硝态氮控制器的设定值。仿真结果表明在关键水质达标的基础上,降低了能耗。
Regard to the international wastewater treatment benchmark simulation model No. 1(BSM1), cascade control strategy is used to control the ammonium concentration in the 5th compartment and a single loop PID controller is used to control the nitrate concentration in the 2nd compartment. A neural network is developed to model energy consumption using the set points of ammonium concentration and nitrate nitrogen concentration. Similarly, another neural network is created to model constraint of effluent quality using these set points. The optimal problem containing constraint model and energy consumption model is then solved by NLJ optimization algorithm. Solutions of the optimal problem are used as the set points of ammonium concentration controller and nitrate nitrogen concentration controller respectively. The results show that the energy consumption of wastewater treatment process is reduced while satisfying key effluent quality constraints.
出处
《计算机与应用化学》
CAS
2015年第11期1338-1342,共5页
Computers and Applied Chemistry
基金
国家自然科学基金资助项目(61304071)