Research On Analytical Tuning Method Of Model Predictive Control For Multivariable Fractional Dead Time System

Model predictive control has been widely used in industrial processes because it can deal with system constraints in low model requirements,and can keep strong robustness.Due to many subsystems of industrial processes that can be described by a first-order plus dead time model,it is urgent to analyze the closed-loop performance of the predictive controllers and then propose effective analytical parameter tuning methods.Most of the existing parameter tuning methods in industrial applications are mainly based on engineering experience or numerical approximation,which leads to a certain blindness in predictive control parameter tuning.In addition,for some systems with fractional dead time in the industrial process,the existing methods usually approximate the fractional dead time to the integer for the next optimization,which results in the uncertain system performance and even worsening the closed-loop performance of the system.Aiming at the above problems,this paper divides the multivariate systems which can be described by first-order plus fractional dead time models into three cases for analysis and research,and proposes several corresponding analytical model predictive control tuning methods on multivariate fractional dead time systems.1.For multivariable fractional dead time systems that for each output the dynamics and time delay from the outputs are the same,the mathematical model is firstly derived,and then the closed-loop transfer function is obtained to perform decoupling analysis which reveals the quantitative relationship between the predict control parameters and the closed-loop performance.Next,the analytical tuning formulas for predictive control parameters are obtained.Finally,the simulation results in p H automatic control system and mixing system are used to verify the effectiveness of the proposed algorithm.2.For multivariable fractional dead time systems that each input affects the outputs with the same dynamic and time delay,firstly according to the same dead time,the mathematical model is firstly derived,and then the closed-loop transfer function and decoupling condition are obtained to achieve its analytical predictive control analysis.On the basis of different dead time,because closed-loop systems are difficult to completely decouple,a new predictive control analytical tuning method based on optimization problem is proposed.Finally,the effectiveness of the methods are verified using a turbine generator system as an example.3.For general multivariable fractional dead time systems,based on the above research,the mathematical model is firstly derived,the closed-loop system transfer function is obtained and then the parameter tuning problem is transformed into a pole placement problem.Therefore,an effective analytical parameter tuning method for model predictive controllers is proposed.Finally,the simulation result in Methanol-water distillation processing system is used to verify the effectiveness of the proposed algorithm.