Software Package Design And Simulation Studies On Predictive Control

Modern control theory in aerospace, aviation and other fields have achievedbrilliant results. However, it requires the control object must have a precisemathematical model. The objects involved in the industrial process are often multi-input multi-output high-dimensional complex systems, so their mathematical models aredifficult to accurately obtain. Predictive control methods have been proposed in1970sto overcome the imbalance between control theory and applications. It includes severalforms of predictive control techniques such as Model Algorithmic Control (MAC),Dynamic Matrix Control (DMC), Generalized Predictive Control (GPC), etc. Thesetechniques are similar to some extent, where MAC relies on impulse response model,DMC step response model and GPC control auto-regressive integrated moving average(CARIMA) model. Requirement for model in predictive control is low. Using the past,present and future information to get the control input, and effectively handlingconstraints and multivariable problems, predictive control soon attracts widespreadinterest and attention from the industrial control sector and academic circles.At present, there are many simulation studies on the MAC and DMC algorithm.For example, improved MAC algorithm in the discrete linear system; fractional PIDcontrol combined with the DMC control in the second-order two-capacity water tanksystem. However, the actual industrial processes are usually nonlinear, multivariabletightly coupled, large time delay, etc.This article is focus on the MAC, I-DMC and II-DMC algorithm faced all cases.The constraint problems are converted into the quadratic programming problem solvedby using quadprog function in the MATLAB optimization toolbox. CSTR is a highlynon-linear chemical reaction system in typical process industry and has been widelyused in dyes, pharmaceuticals, reagents, food and synthetic materials industry, and playsan important role in the core of the chemical production equipment. This articleestablishes the continuous single-input single-output nonlinear mathematical model andthe two inputs and two outputs state space model for the CSTR system, and solves theunconstrained and constraint cases by real-time simulation on MAC, I-DMC andII-DMC, and finally analyzes the control performances of the algorithms, providing aneffective theoretical support for the actual industrial process control algorithm selection.This article makes the software package of predictive control based on simulation programs of the three types of algorithm, and achieves the goal thar the user can choicethe predictive controller arbitrarily.