Arm-based Model Predictive Control System Research And Development

PID is the most widely used industrial control method. However, its performance is not satisfactory when the process is of non-linear, time-variables, coupling or uncertain parameters. It is proved that such complex processes would be controlled well by using Model Predictive Control (MPC) both in theory and practice. Because of the technical barriers and monopoly in application of industry, MPC has not been a universal industrial control algorithm. The research of embedded MPC is the breach of the MPC popularization. It is also a new field of research.In the paper, a solution of embedded MPC, which is built within the framework of RS485 bus, is proposed to serve the small system. The research of embedded MPC system includes algorithm, controller hardware, controller software, PC software and applications of the system. The details are as follows:1. The shifted matrix is introduced to make the control increment convergence to zero in the control horizon on study of Dynamic Matrix Control (DMC). Computing a number's reciprocal can substitute for the matrix inversion. The simplified algorithm reduces the computing time of the control vector, so it is easier to apply to small-physical-size embedded system. In the light of the convergent trend of the control increment, the algorithm is called Convergent DMC. The simulation confirms the validity of the algorithm.2. MPC circuit is designed on ARM to meet the requests of the embedded MPC. The circuit is divided into two parts, CPU system board and expansion board. The hardware is convenient to upgrade and expand.3. Linux, of which the kernel is trimmed, is used to be the operating system of embedded MPC. U-Boot is selected as the bootloader of the system. Cramfs, a read-only file system, is established. MPC software is programmed on simplified DMC using C++.4. The real-time PC monitoring software ARM-MPC Monitor is finished with MFC of Visual C++. The implementation of the software includes system configuration, algorithm tuning, data communication and results output.5. Speed control of the DC motor is taken for example to describe the details in using embedded MPC system, and then analyze the different results of the control algorithms. Embedded MPC replaces the conventional local loop in the system. The experimental result confirms that the control quality of MPC is better than that of traditional PID control. The thesis is concluded with a summary and prospect of future researches.