基于自组织递归小波神经网络的污水处理过程多变量控制

Multivariate Control of Wastewater Treatment Process Based on Self-organized Recurrent Wavelet Neural Network

污水处理过程(Wastewater treatment process,WWTP)是一个包含多个生化反应的复杂过程,具有非线性和动态特性.因此,实现污水处理过程的精准控制是一项挑战.为解决这个问题,提出一种基于自组织递归小波神经网络(Selforganized recurrent wavelet neural network,SRWNN)的污水处理过程多变量控制.首先,针对污水处理过程的动态特性,根据小波基的激活强度设计一种自组织机制来动态调整递归小波神经网络控制器的结构,提高控制的性能.然后,采用结合自适应学习率的在线学习算法,实现控制器的参数学习.此外,通过李雅普诺夫稳定性定理证明此控制器的稳定性.最后,采用基准仿真平台进行仿真验证,实验结果表明,此控制方法可以有效提高污水处理过程的控制绝对误差积分(Integral of absolute error,IAE)和积分平方误差(Integral of squared error,ISE)的精度.

The wastewater treatment process(WWTP)is a complex process containing multiple biochemical reactions with nonlinear and dynamic characteristics.Therefore,it is a challenge to achieve accurate control of the wastewater treatment process.To solve this problem,a multi-variable control of wastewater treatment process based on the self-organized recurrent wavelet neural network(SRWNN)is proposed.Firstly,to deal with the dynamicity of wastewater treatment process,according to the firing strength of the wavelet base,the self-organizing mechanism is designed to dynamically adjust the structure of the recurrent wavelet neural network controller to improve the control performance.Then,an online learning algorithm combined with adaptive learning rate is used to learn the parameters of controller.In addition,the stability of the controller is proved by the Lyapunov stability theorem.Finally,the benchmark simulation platform is used to conduct simulation.The experimental results show that this control method can effectively improve the integral of absolute error(IAE)and integral of squared error(ISE)of the wastewater treatment process.