| Dynamic matrix control (DMC) is one of the most extensively apllied algorithms of predictive control. Parameters of the classic DMC controller have usually been got through some simulations and real time tests before running. But these parameters usually can't guarantee good control results. This is a widespread problem to DMC. Currently, the auto-tuning method and technique for DMC controllers still lack research, and as a result that DMC's extensive applications in industry are seriously limited. Circulating Fluidized Bed Boiler (CFBB) is a typical thermal plant, and it is very important to implement predictive control for such a plant in order to guarantee power stations safety and efficiency. With complicated industry process modelling control as background, this thesis study the auto-tuning method of predictive control system in the complicated industrial processes based on the overview of currently domestic and international forerunner methods in thermal process control.Chapter 1 is an introduction of the thesis. The background and significance are firstly described, then currently domestic and international research situation are summed up, and predictive control applications in the thermal processes are also summarized. Based on DMC's problems at present and the complication of CFBB thermal control, the thesis put forward its main targets.Chapter 2 study the classic DMC and its parameter auto-tuning. In the context of simulations, there are two topics to be probed into: there are some parameters to choose to be main tuning parameters; Acording the prompted tuning strategy, some typicalplants have been numerically simulated. In the industrial application, cascaded DMOPID predictive control is employed to control over-heat steam temperature, and the results were better.Chapter 3 study DMC algorithm for integrating processes (non-self-regulating processes), and put forward the parameter auto-tuning strategies for SISO non-self-regulation processes. To the vapor bag level, the prompted method are tested and verified in real simulations. The results show that the method is better than other common design methods. In the aspect of the industrial application, the predictive control software package that the author partly took part in design was applied to the control CFBB level of a certain power plant. Compared with PID control, the DMC can attain satisfying results.Chapter 4 discusses multivariable DMC algorithm and its parameter auto-tuning strategy, and pilot to study the method's applications in the CFBB load control. With CFBB load control system, the thesis made a simulation and analysis for a typical Two-Input-Two-Output process. The control results were better than the classic method.Chapter 5 presents the software development work. The full DMC parameter auto-tunning software package has been developed in the Matlab environment. The package can be applied to the typical process control loop, and systematically test, verify and simulate, then carry on on-line control. Using it, better control results can be acquired, and adjusting time will decrease.Additionally to take in as a member, the author partly participated application study work of sub-topic for the project of 863. It includes to successfully embed the developed software package into DCS of Shanghai Xinhua control engineering company.It carryed out the satisfying optimal control in the 130t/h CFBB machine unit of a certain heat power station, and obtained satisfying applied result. |
PDF Full Text Request