基于Hopfield 神经网络的污水处理过程优化控制

针对前置反硝化污水处理过程的优化控制问题,提出一种基于拉格朗日乘子法的Hofield神经网络优化方法.构造了污水处理过程约束优化问题的数学表达式,通过Hopfield神经网络优化计算生化池第5分区溶解氧浓度和第2分区硝态氮浓度的设定值,并采用PID控制器实现底层的跟踪控制.基于国际标准的Benchmark基准仿真平台进行仿真实验,结果表明污水处理系统在出水关键水质达标的基础上,能够显著降低能耗.

Optimal control for wastewater treatment process based on Hopfield neural network

For the optimal control problem of predinitrification wastewater treatment process, a Hopfield neural network optimization method based on the Lagrange multiplier is proposed. Firstly, under the constrain of some key effluent pollutant qualities, a wastewater treatment optimization objective function is constructed to minimise the energy consumption. Then, the set points in bioreactor of both dissovled oxygen concentration in the 5th compartment and nitrate concentration in the 2nd compartment are optimized by Hopfiled neural network, respectively. Both concentrations are controlled by PID controller. Finally, based on the international standard benchmark, the simulation results show that through the optimization of Lagrange multiplier Hopfield neural network, the energy consumption of wastewater treatment process is reduced obviously under constraints of effluent pollutant qualities.