Research On Model Predictive Controller Based On FPGA/SOPC

With its natural ability in handling constraints,Model Predictive Control(MPC) has become an established control technology in the petrochemical industry, and its use is currently being pioneered in an increasingly wide range of process industries. It is also being proposed for a range of higher bandwidth applications, such as ships,aerospace and road vehicles. New applications have made new demands to MPC controller, such as high real-time performance, miniaturization, high reliability and low-cost. These are also urgent problems which MPC algorithm has to face in new applications. The solving speed of the whole algorithm is restricted by the complexity of online optimization, particularly in the rapid system.Therefore, the online solving speed of MPC has become the research focus both at home and abroad.In the last decade, many researchers have presented several efficient optimization algorithms for MPC, which have improved the MPC algorithm's performance. Other researchers proposed some new methods to implement the MPC controller. This thesis presents a new method to implement the MPC controller based on the FPGA chip. Compared to the traditional method, the controller based on FPGA can meet the needs of new demands in new applications.The new MPC controller can extend applications of MPC to miniaturized devices and embedded systems.Refer to the hardware implementation based on FPGA, there are two ideas. one is to use to implement the MPC controller totally on a FPGA chip by writing HDL code. The MPC controller by this idea has a simple system structure and high real-time performance.But its weakness is its difficult development process and long development cycle.The other is SOPC scheme based on Nios II softcore processor. The MPC algorithm runs in the Nios II processor.Besides, custom instructions and hardware accelerators in the Nios II processor can largely improve the solving speed of MPC. This scheme can also reduce costs and has a shorter development cycle.The main tasks of the thesis are as follows:1. To analyze linear constrained MPC and establish state-space model of the electronic throttle. To design linear constrained MPC controller for electronic throttle and simulate the controller in Matlab / Simulink;2. To build up a SOPC system and complete the design of hardware and software systems on FPGA chip based on SOPC scheme, mainly including the configuration of the Nios II CPU core and standard IP components, design of custom instruction and custom components, implementation of MPC algorithm in C programming language , etc.;3. To set up real-time simulation platform based on FPGA and dSPACE real-time simulation system and run real-time simulation of the MPC controller based on FPGA. The results show that the MPC controller based on FPGA has high real-time performance and characteristics of miniaturization, high reliability and low-cost.It can meet the demands in new applications.Above all, this thesis has implemented a MPC controller based on the FPGA/SOPC, and its real-time simulation shows that MPC controller based on FPGA has high real-time performance. The result has proved that the implementation of controller based on FPGA/SOPC is feasible and significant.