Efficient Incremental Model Predictive Control:Algorithms, Realization And Applications

Model predictive control (MPC) is a computer control algorithm based on online optimization, with the advantages of constraint handling, performance optimization and multi-variables control. It has been broadly applied to industrial control. MPC algorithm calculates the control law online using the optimization algorithm, but the existing algorithms and the realization are always complex and hard to apply to the fast system. Meanwhile, the embedded system which is used in the design of controller broadly has characteristics of small size, low cost, relatively limited process and storage abilities, so it can not satisfy the real-time requirement of the complex optimization algorithm. Therefore, it has great theoretical and applied value to study an efficient model predictive control algorithm and its application.Considering the multi-variable linear system with constraints, the dissertation discusses the efficient MPC algorithm, as well as its realization and application. The efficient incremental MPC with constraints is researched on aspects of controller simplification and optimization algorithm by using incremental state-space model as the predictive model. An efficient incremental model predictive controller based on PIC MCU is designed, and is applied to the Gas-Liquid Cylindrical Cyclone (GLCC) liquid-level system. The main researches and innovations of the dissertation are as follows:(1) An efficient incremental MPC algorithm is proposed for a multi-variables linear time-invariant system with limit inputs. An incremental state-space model is used as the predictive model to build the MPC rolling optimization problem. The stair-like control strategy is used to compress the decision variables of the optimization problem to reduce the MPC online computation. Furthermore, the coordinate alternation and golden section algorithms are introduced to online optimize the value of predictive control increment, aiming to finish the online rolling calculation of the predictive controller with constraints. Finally, the simulation illustrates the superiority of the algorithm by comparing with truncation and interior point methods. (2) The controller software and hardware implementation are designed for the efficient incremental predictive control algorithm. Using PIC MCU as master chip, the dissertation designs the basic interface circuits such as data collection and process, human-computer interaction, communication and storage circuit. The design of software is modularized, and completes the implementation of user interface and MPC algorithm.(3) The efficient incremental predictive controller is applied to the GLCC liquid-level control system, which has double inputs and single output. Then, the proposed algorithm is compared with the PID algorithm through experimental research on the designed controller, and the result verifies the practicability and superiority of the proposed algorithm.Finally, the conclusion and future work are presented.