Modular Multivariable Generalized Predictive Control And Its Applications

The demand for the process control continuously advanced along with the rapid development of modem industry. The constrained multivariable multi-object dynamic optimal control problem must be solved.The static Modular Multivariable Control (MMC) can explicitly deal with constrained multi-object static optimal control problem. With the hierarchical modular structure, it assigns each object into one corresponding control module according to the priorities of the objects, and satisfies these objects from the high priority to the low priority based on the lexicographic minimization. The hierarchical modular structure explicitly expresses engineering trade-offs and decision. It follows the acquirements of the industrial process control.The stair-like control technique decreases calculated complexity of future control variables, with the combination of modular multivariable control and stair-like Generalized Predictive Control (GPC), this paper presents Modular Multivariable Generalized Predictive Control (MMGPC), which can handle dynamic modular multivariable optimal control problem.Dynamic MMC can explicitly handle the constrained multivariable multi-object dynamic optimal control problem which needs to be solved in modern industry. Stair-like GPC can handle the control problem with large time constant, large pure time delay, strong disturbances, and it has strong robustness. The combination of them makes MMGPC obtain good effect in practical applications.The Intelligent Supervisory Shell (ISS) is very important in promoting continuous runtime and running rate of the advanced controllers. Based on the conclusion of the research actuality, task and content of the intelligent supervisory shell from the aspects of identification supervision and controller supervision, this paper presents general framework of the supervisory shell for identification and the supervisory shell for controller. The object-oriented modular framework of ISS makes the design and reconfiguration of the practical samples of ISS very convenient.The application research of dynamic MMC proves its good quality and practicality. The application research includes three parts:Part I: MMGPC for Main-steam Temperature of Boiler in Power PlantBy the analysis of operation technics, the control for main-steam temperature of boiler in power plant is actually a constrained multi-object optimal control problem. MMGPC is closer to the essence of the problem compared with other algorithms which only traces the setpoint.In order to obtain the model of the controlled objects, this paper presents a new modeling method based on the combination of the Least Squares Identification (LSI) and the Correlative Coefficient Identification (CCI). It can obtain the order, pure time delay and coefficients of the model simultaneously.Part II: MMGPC for Temperature of Multi-segment Cool-excitated Ammonia Synthetic Tower...