Study On Model Predictive Control Of Large-scale Membrane Reverse Osmosis Seawater Desalination System

With the rapid increase of the population of the world and the rapid development of industrialization,the shortage of freshwater resources has become increasingly serious and a major obstacle to social development.In order to obtain clean fresh water resources,the world has set its sights on the ocean.Developing desalination technology is an important way to solve the global water shortage problem.It has formed a consensus in the global scientific and technological community.With the development of low-cost,high-salt membrane technology and the advancement of high-efficiency energy recovery devices,seawater reverse osmosis(SWRO)desalination technology has become the most popular and concerned by its economy and convenience.Adopting appropriate control strategies is the key factor to ensure SWRO system reliability,improve water quality,and reduce operating costs.In the domestic reverse osmosis seawater desalination system,most of the traditional control methods are used.However,the SWRO system has the characteristics of coupling,time lag and nonlinearity.It is difficult to achieve the ideal control effect by using traditional control schemes such as PID.Therefore,according to the characteristics of reverse osmosis desalination,this paper proposes three control strategies:PID control,decoupled control and Model Predictive Control(MPC).Through horizontal comparison,it can be found that the advanced control strategy can significantly improve the system.Control effect.In order to apply advanced control strategies in the actual process,this paper designed a special control software,which can be connected with DCS or PLC through OPC(OLE for Process Control).The main work of this paper has the following points:(1)Based on the mass equation,momentum equation and energy equation,the nonlinear mechanism model of SWRO system is established.Because the SWRO mechanism model is more complicated,it is difficult to directly use it in control system design.Therefore,this paper uses the system identification technology to linearize the nonlinear model under typical conditions,and obtain a linear model to facilitate the subsequent control system design.(2)Based on the linear model,the design method of the conventional control scheme is given,including the design of the PID control system and the decoupled control system.Firstly,the traditional PID control scheme is given.The shortcomings of traditional PID control are analyzed by tracking and anti-interference test.Then,in order to effectively cope with the coupling characteristics of the system,a feedforward compensation decoupled controller is designed.Through the tracking and anti-interference simulation test,it is shown that the decoupled control can improve the control performance of the system.(3)Introduce the model predictive control algorithm to design the SWRO control system.It can deal with practical problems such as coupling and constraints,and can further improve the performance of the control system.In order to achieve unbiased tracking,unmeasurable disturbances are introduced into the model,and the linear model is expanded as necessary.Real-time state estimation can effectively deal with the effects of factors such as model mismatch,unmeasurable disturbance,and changing working conditions.(4)In order to apply the advanced control algorithm to the actual system,a special predictive control software was designed.Under the VS2010 platform,the MPC algorithm is packaged to form an advanced control software with online control,offline simulation,OPC communication and other functions.