基于改进粒子群算法的污水处理过程神经网络优化控制

摘要

Due to the high energy consumption of activated sludge wastewater treatment process, a new intelligent optimal control system is designed in this paper by considering the effluent quality and the relationship between the biochemical reaction parameters. This control system is used for the benchmark simulation model (BSM1) proposed by the International Water Association (IWA). The APSO is utilized to optimize the dissolved oxygen and MLSS levels in the fifth compartment and the nitrate level in the second anoxic tank. Meanwhile, the outputs of BSM1 are predicted by the neural network, and the energy consumption is cut down whthin the effluent water quality standar-ts. The simulation results show that, comparing to the cloose-loop control strategy, the totle energy consumption of this proposed optimal control system is lowered by 4. 614% , the neural network optimal control strategy can significantly reduce the energy consumption of activated sludge wastewater treatment process.%针对活性污泥法污水处理过程高能耗的问题,综合考虑污水处理出水水质和生化反应参数之间的关系,文中设计了一种智能优化控制系统.该系统以国际水协(IWA)开发的基准仿真模型BSM1为研究对象,利用改进粒子群算法优化BSM1第2分区的硝态氮浓度和第5分区的溶解氧浓度、混合液悬浮物固体浓度的设定值；同时利用感知器神经网络预测污水处理过程的输出,在出水水质达标的前提下降低污水处理能耗.仿真实验结果表明,系统总能耗相比闭环控制策略降低4.614％,该神经网络智能优化控制系统能够有效降低污水处理的能耗.