Model Predictive Control Analysis

2004 Words9 Pages
It is well known that the common drawback of non-adaptive control techniques such as PID and classical fuzzy control is that it is designed and tuned for specific operating point which in our case is the wind direction. The parameters of these controllers are static and don’t adjust to the change of this operating point. The proposed technique Model predictive control (MPC) has its roots in optimal control. It is about using a dynamic model to predict system behavior, and solve optimization problem to produce the best control move at the current time. As an optimal control technique its move depends on the initial state of the dynamic system, so it use the past record of measurements to determine the most likely initial state of the system.…show more content…
Gu. Model Predictive Control Basics and Operation 4.3.1 Illustrative Example Our daily work plan is a perfect example for showing how model predictive control technique works. At the beginning of the day an eight hours plan is put for all tasks needs to be done by this part of the day. Also, factors which could be adjusted to reach day plan objective to be put as well. For good tracking of plan progress, plan will be reviewed each one hour and see if it is need to be adapted using predefined factors according to achievements after this interval. If we map this to control area we can say that: Work Background is plant model, Adjustable factors are manipulated variables, one hour is sampling interval time and finally eight hours is our prediction horizon which refer to the period we wish the future to be predicted for. Depending on Controller model structure, there are three main general approaches to MPC design. Step response models and Model finite impulse response (FIR) models are used in the past. Recent years, the transfer function model-based predictive control and state-space design methods is…show more content…
Where Np is number of samples, and the current time is ki. x (ki) is the state variable vector which is gotten from measurement and it provides the current plant information. ∆u(ki),∆u(ki +1), . . . ,∆u(ki + Nc − 1) is the future control trajectory , where Nc is called the control horizon dictating the number of parameters used to capture the future control trajectory. With given information x(ki), the future state variables are predicted for Np number of samples. Finally, x (ki +1 | ki), x(ki +2 | ki), . . . , x(ki + m | ki), . . . , x(ki + Np | ki) is future state
Open Document