Scribd
Upload
29 views

Model Predictive Control

Uploaded by

yazen.shakir
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
29 views

Model Predictive Control

Uploaded by

yazen.shakir
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 12

Model Predictive Control

Yaser Nabeel Ibrahem


September 2024
References

• Rawlings, James Blake, David Q. Mayne, and Moritz


Diehl. Model predictive control: theory, computation, and
design. Vol. 2. Madison, WI: Nob Hill Publishing, 2017.

• Kouvaritakis, Basil, and Mark Cannon. "Model predictive


control." Switzerland: Springer International Publishing 38
(2016): 13-56.
Introduction

• It is well known that handling the required state and control constraints and
achieving the system stability are the main priority needs for tracking a
trajectory by the high non-linear dynamic models.

• The MPC controller has emerged as the most popular and valuable control
technique since the 1980s, known as Receding Horizon controller as well.

• It is a worth describing the method based on the principle of prediction to a


finite horizon at each sampling time starting from the current state.

• Subjected to the state and control constraints of the system dynamics, the
optimization problem is solved to determine the optimal control input.
The concepts and The origins of predictive control

• Easy to understand and powerful controller.


• Extensible to multivariable systems.
• Idea of a receding horizon.
• Computation of control signal by optimizing predicted plant behavior.

• Model Predictive Control Also known as:


• * Receding Horizon Control (RHC).
• * Dynamical Matrix Control (DMC).
• * Generalized Predictive Control (GPC).
Main components of model predictive control

• Actions depend on predictions.


• Predictions are based on a model.
• Current input is based on achieving a best output.
• Limited time window (receding horizon).
• Precise control requires an accurate model.
• Handles constraints (QP or LP) in a systematic way.
• solving an optimization problem at each step.
feature of MPC is that we can include different
kinds of models such as:

• Linear.
• nonlinear.
• SISO (Single input Single output).
• MIMO (Multiple Input Multiple Output) with strong coupling.
• Models with constraints.
• Models with time delays.
• Models with inverse response.
Constrained Linear Model Predictive Control

• The linear model predictive control LMPC approach is designed by relying on


the linearization of the dynamic model. The discrete time dynamical model
description is shown in equation (1).
Xk+1 = Ak Xk + Bk uk (1)

• The finite horizon optimal controller solves the following constrained


optimization problem at each time instant k to implement the MPC algorithm.
The cost function is presented by:

(2)
• Where the terminal state and its desired state are denoted by 𝒳N and 𝒳*N,
respectively, and the reference state is denoted by 𝒳*k. The prediction
horizon is denoted by N. Qf and Q are positive semidefinite matrices and R is
a positive definite matrix.

• The constrained optimization problem for linear MPC at each time instant k is

(3)

• 𝓧𝑲 ∈ 𝓡𝓷 , 𝒰𝐊 ∈ 𝓡𝓶 , A𝑲 ∈ 𝓡𝓷𝒙𝓷, B𝑲 ∈ 𝓡𝓷𝒙𝓶
• Where the state and input constraints are presented as

(4)

• The linear optimization algorithm is implemented to obtain a minimum Finite


horizon cost of the desired states and control inputs while the constraints are
satisfied.
• A quadratic programming (QP) or a linear programming algorithms are used
to solve the low level optimization problem at each time instant k, which can
produce an input sequence.
Constrained Non-linear Model Predictive Control

• Non-linear Model Predictive Control NMPC approach is proposed in this


chapter to handle the high non-linearity based on prediction method at each
sampling time subjected to the state and control constraints.

• For each sampling k, the optimal control sequence can be found by solving
the optimization problem including constraints in order to perform the NMPC
method. A nonlinear programming (NLP) or a quadratic programming (QP)
algorithm can be used to solve a constrained optimization problem at each
time instant k.
• The constrained optimization problem for non-linear MPC at each time
instant k is:

(5)
Figure (1) Basic concept of MPC
Where 𝒱 ≤ 𝑓
The goal of the model predictive controller is to determine a set of control
inputs at the time instant 𝐤, that will make the output of the system to follow a
prescribed output trajectory (desired control trajectory) over the prediction
horizon.

You might also like