Model Predictive Control


The primary focus this project is the development of the concept of Model Predictive Control (MPC) on a chip, and finding solutions to issues related to its feasibility. In particular, topics of interest include :

  • Prove the feasibility MPC for control of systems with fast dynamics.

  • Determine the feasibility of MPC on a chip for various different systems.
    • For given contraints, sample time, system states determine the viability of a hardware implementation.

  • Apply MPC to control various examples of real systems.
    • Use the theoretical work to implement a real MPC controller in hardware for the control of an Inverted Pendulum, which is a classic example of simple non-linear example.
    • Extend the work to use MPC to control the 3D Crane problem.
    • Implement MPC on an FPGA for the cantilevered beam problem, to compare the resulting hardware against the existing DSP solution.

  • Develop the concept of MPC on a chip.
    • Explore the scope for parallelisation in hardware based implementations to decrease computation time for MPC algorithm, and thus increase the complexity of systems that may feasibly be implemented.
    • Develop a suite of generalised architectures and hardware modules that can be scaled to meet the requirements of a wide range of problem specifications.
    • Develop proof-of-concept custom circuits to demonstrate the feasibility and benfits of deploying MPC onto FPGAs and ASICs

Maintained by Dr. Adrian Wills
University of Newcastle
22 Aug 2009, © Copyright