Research Of Real-time Model Predictive Control For Three-phase Asynchronous Motor

Model predictive control is an advanced control method based on receding horizon multi-parametric optimization. Traditionally, it has been mainly applied to process control industry. Due to its inherent high computational demand, it is difficult to apply model predictive control directly to high sampling rate real-time scenarios.Explicit model predictive control addresses the computational load issue of the classical model predictive control by computing the feasible state partition and its corresponding control law offline. And the real-time implementation of the control law is achieved by matching the current state with the feasible state partition. This greatly reduce the on-line computational time. With explicit model predictive control, the feasible application area of model predictive control can be expanded. In this thesis, the explicit model predictive control has been applied to the real-time control of three-phase asynchronous motor. Through the simulation study, and comparing the results obtained with widely used PID controller, it is shown that explicit model predictive can be a feasible solution for high-sampling rate real-time control applications.Fast online model predictive control based on convex optimization has also been studied. This control method is based on the CVXGEN toolbox. The multi-parametric optimization problem is converted into a convex optimization problem which can be solved efficiently with CVXGEN. In this thesis, the procedure required to use the CVXGEN toolbox has been investigated and has been applied to the control of asynchronous motor. By comparing the results obtained with widely used PID controller, it can be seen that the fast online model predictive control method can greatly reduce the computational load and solve the optimization problem within the short time-frame.