Prediction And Compensation Algorithm About Time-delay Systems

Through the development of control theory for several decades, it has experienced two stages: Classical Control Theory and Modern Control Theory. Using frequency methods, Classical Control Theory is mainly used for single input and single output systems. But it is not suitable for the space science which is developing rapidly nowadays. The methods for Modern Control Theory are time analysis. It is widely used in multiple inputs and multiple outputs systems. One problem is that exact mathematics models are required in Modern Control Theory based on State Space theory, but they are hardly to be got because of the uncertainty of the structures, parameters and environments in industrial process control, so the Model Predictive Control algorithm is generated and widely used because its models are non-parameters ones. The models in this algorithm are gained through experimental methods. The main ideas of the algorithm are "model predicting—feedback revising—rolling optimizing". By calculating the quadratic performance index of the system, the optimal control variable will be got. There are three typical algorithms of Model Predictive Control (MPC): Model Algorithm Control (MAC), Dynamic Matrix Control (DMC), and Generalized Predictive Control (GPC). In this paper, our discussions mainly base on State Space description of DMC.In industrial process control, the problem of time-delay is always hardly to be avoided, but which is one of the most important factors resulting in control failure. So in this paper, our main researching purpose is using the idea of prediction to diminish the affection of time-delay. First we analyzed the main factors resulting in time-delay, and then constituted the system time-delay model based on DMC State Space description, also the Time-delay Compensation on DMC was introduced to improve the control effect. Because of the affection of time-delay, the control inputs can not trace the system changing, the Compensation Gene and the Transition Gene were introduced to adjust the predictive control inputs and the delayed control inputs to realize control stability and speediness. The algorithm is also suitable for the systems which saltate frequently.Because of the complication of industrial process control, the problem of time-delay becomes more and more serious. Using the idea of prediction, this problem was analyzed in the method of Dynamic Matrix Control (DMC) on the model of subsystems, also the relationships among the subsystems were well concerned and the price-driven coordination method was used to optimize the linking constrains. To diminish the affection of delay the algorithm of Time-Delay Compensation on DMC (TDCDMC) for large-scale systems was introduced. The simulation results show that this improved algorithm has a good delay compensation performance and the control stability and speediness are guaranteed.