Researching On TDF-IMC-PID And Mulit-Model Switchlng Control

In recent years, with the continuous improving of the production process, the performance required for the process control is also increasing. To some control loops it is difficult to get good performance through using normal PID controller. In the industrial production the PID controller is the highest utilization rate controller, so the research on improving PID controller is very meaningful. In this context the PID controller based on two degree of freedom internal model control structure is proposed. It often encounters a larger range of operating conditions or frequently changes and other issues in the actual production process. The multi-model control provides a good method to solve the above problems. So this paper also launched a number of studies on multi-model switching control. Specific studies include the following aspects:First, we study how to derive the PID controller parameters from the two degree of freedom internal model control(TDF-IMC) structure. The IAE performance indicators and the structure singular value are used to get the two freedom degree of internal model controller’s feedback filter parameter. The IAE performance indicators and overshoot constraint are used to get the feedforward filter parameter. According to the relationship between the two freedom degree of internal model controller and the PID controller we can get the parameters of PID controller. The simulation shows this method has excellent control performance, it is worth to promote in practical applications.Secondly, in this context a multi-model TDF-IMC structure is proposed. It uses a wide range of multi-model approach to working condition and the TDF-IMC to ensure good tracking performance and interference robustness. By the improved K-means clustering algorithm to classify the collected data and establish the model set. Through simulation and practical application show that the method for complex conditions, the long time lag single circuit has good control performance. Applying on the de-hexane tower unit successfully; improve the product quality and economic efficiency.Finally, to solve the problem that too many models will cause a long transition time the method that use multiple models set is proposed in this context. The method that uses the principle of universal MVC3and optimization problem of linear quadratic Gaussian’s inversion to tune the weighting matrix parameters of model predictive controller is proposed. And combine the DMC with the work set to achieve optimize operations.