Research On Neural Network Control Algorithm Based On Biological Regulation Mechanism

With the growth of population and the development of social economy,the problem of water resources seriously restricts the further development of society.Open-channel water diversion is a common way of water transfer projects all over the world,and it is also an important measure to rationally utilize water resources and optimize the allocation of water resources.However,in recent years,with the expansion of the scale of channels,the impact of coupling between channels has become more and more serious.If decoupling cannot be effectively achieved,it will lead the water level to fluctuate greatly,and even cause overtopping accident.Therefore,aiming at solving the decoupling control of water diversion channel system,this paper research the neural network control algorithm improved by biological regulation and apply it to the system of the middle route of the South-to-North Water Diversion for simulation.(1)In this paper,based on the downstream control operation mode of the main canal water delivery system in the middle route of the South-to-North Water Diversion Project,the effect requirements of control effect and decoupling effect are analyzed,which provides specific requirements for the design of decoupling control algorithm.Considering the coupling effect in the process of water conveyance,taking two adjacent canals as the research object,the mechanism of the relationship between the liquid level height of the canal and the gate flow is analyzed.Thus,with the actual canal parameters,the coupling model of the system is established,which provides the object under the coupling effect to verifying the effect of the control algorithm in the middle route of the South-to-North Water Diversion.(2)Inspired by the interaction mechanism between the biological endocrine system and the nervous system,this paper proposes a self-feedback ANN algorithm based on the endocrine regulation mechanism by improving the traditional BPNN.It is mainly composed of a selffeedback ANN unit and the endocrine regulation mechanism unit controlling the concentration of hormones are composed,thereby strengthening the ability of rapid adaptive adjustment and having the ability of memory.This algorithm is applied in the forward control unit to design an intelligent controller to adjust the parameters by online training,which can also improve the control accuracy and speed of the controller.Finally,the simulation results prove that the forward control effect is timely and effective.(3)Aiming at the decoupling control problem of the middle route of the South-to-North Water Diversion with the characteristics of large hysteresis,strong coupling,and time-varying,a predictive decoupling control Algorithm is proposed.It is inspired by the decoupling and coordinated regulation mechanism of the neuroendocrine system.By designing the forward estimation unit based on incremental variables,the improved ANN is used to obtain the estimated pseudo output value of the forward channel,so as to solve the problem to accurately obtain the uncoupled output value in the time-varying system,thereby improve the practicability of the algorithm.The objective function of the decoupling compensation neural network is constructed through the pre-obtained pseudo output value for online training,and the decoupling compensation control variable is obtained in advance to reduce the coupling effect and improve the hysteresis of the decoupling.On this basis,a composite objective function is used to replace the single objective function of the decoupling compensation ANN,and the proportion factor is designed to adjust the composite objective function to further optimize the decoupling control scheme.In order to ensure the reliability of the algorithm,the stability of the system is proved by the Lyapunov stability theorem.In paper,the designed predictive decoupling control algorithm is applied to the central line of South-to-North Water Diversion for simulation research,compared with other control algorithms.The results show that the amplitude and adjustment time of predictive decoupling control algorithm are reduced by 60.15% and 76.62% respectively,which significantly improves the decoupling performance compared with the traditional algrithom.In particular,the algorithm improved by the compound objective function are reduced by 54.41% of amplitude and 44.25% of adjustment time,which realizes the real-time dynamic decoupling in the system of central line of South-to-North Water Diversion.