An alternative use of the Riccati recursion for efficient optimization
Introduction
Model Predictive Control (MPC) is one of the most commonly used control strategies in industry. Some important reasons for its success include that it can handle multi-variable systems and constraints on control signals and state variables in a structured way. In each sample, some kind of optimization problem is solved. In the methods considered in this paper, the optimization problem is assumed to be solved on-line. The optimization problem can be of different types depending on which type of system and problem formulation that is used. The most common variants are linear MPC, non-linear MPC and hybrid MPC. In most cases, the effort spent in the optimization problems boils down to solving Newton-system-like equations. Hence, a great deal of research has been done in the area of solving this type of system of equations efficiently when it has the special form from MPC. It is well-known that these equations (or at least a large part of them) can be cast in the form of a finite horizon LQ control problem and as such it can be solved using a Riccati recursion. Some examples of how Riccati recursions have been used to speed up optimization routines can be found in, for example, [1], [2], [3], [4], [5], [6], [7], [8], [9], [10], [11].
The main purpose of this paper is twofold. First, this paper introduces a new factorization of the KKT coefficient matrix of an MPC problem given in dense form (states eliminated). It is shown that this factorization is closely related to the Cholesky factorization of the coefficient matrix, when the order of the optimization variables is reversed. Second, using this new factorization, a standard Cholesky factorization can be efficiently computed. This makes it possible to develop a drop-in replacement for a generic Cholesky factorization in, for example, standard range-space or null-space Active Set (AS) methods. Compared to the generic factorization, computational performance can be significantly improved for long prediction horizons. The benefit with the proposed factorization compared to standard Riccati based methods is that it requires only small changes to the solver which it is plugged into since the result from the algorithm is a standard Cholesky factorization. This factorization can, as the generic Cholesky factorization, be reused to solve several systems of equations with different right hand sides, and it can be updated using standard methods.
In this article, denotes symmetric matrices with columns. Furthermore, () denotes symmetric positive (semi) definite matrices with columns. The matrix denotes the exchange matrix, which is a matrix with ones in the counter-diagonal and zeros in other elements. Note that for a matrix is nothing but with the rows and columns in reversed order, which can be performed using zero flops. A Sans Serif font is used to indicate that a matrix or a vector is, in some way, composed of stacked matrices or vectors from different time instants. The stacked matrices or vectors have a similar symbol as the composed matrix, but in an ordinary font. For example, .
Section snippets
Problem formulation
The problem considered in this work is where the states , the initial condition , the control inputs , the penalty matrices for the states , penalty matrices for the control inputs , the cross penalty matrices , the linear penalties on the states , the linear penalties on the control signals , the off-set terms , and the prediction horizon
A closer look at the optimality conditions
This section is opened with a standard result that summarizes the properties of the Cholesky factorization.
Theorem 1 Assume . Then, there exists a matrix such that is lower triangular, has nonnegative diagonal entries, and which is known as the Cholesky factorization of . Furthermore, if , then is unique and has positive diagonal entries.Cholesky Factorization, [14]
To be able to exploit problem structure, it is common to use a permutation matrix such that a matrix can be factored as ,
Conclusions
This work presents a method by which Riccati recursions commonly used for high speed MPC applications can be used to compute a generic Cholesky factorization often used in non-structure exploiting optimization routines. The result provides both a new factorization of the coefficient matrix in the KKT system as well as a new way to obtain fast numerical linear algebra for these problems. The classical Riccati method has a computational complexity that grows as and the generic Cholesky
References (15)
- Jonson Henrik, A Newton method for solving non-linear optimal control problems with general constraints. Ph.D. thesis,...
- et al.
Application of interior-point methods to model predictive control
J. Optimiz. Theory App.
(1998) A primal–dual interior-point method for robust optimal control of linear discrete-time systems
IEEE. Trans. Autom. Control
(2000)- et al.
Quadratic programming algorithms for large-scale model predictive control
J. Process Contr.
(2002) - Lieven Vandenberghe, Stephen Boyd, Mehrdad Nouralishahi, Robust linear programming and optimal control. Technical...
- et al.
Efficient solution of second order cone program for model predictive control
Internat. J. Control
(2004) - Daniel Axehill, Anders Hansson, A mixed integer dual quadratic programming algorithm tailored for MPC, in: Proceedings...
Cited by (11)
Distributed Eco-Driving Control of a Platoon of Electric Vehicles Through Riccati Recursion
2023, IEEE Transactions on Intelligent Transportation SystemsReview, Evaluation and Application of Condensing Algorithms for Model Predictive Control based on a First-Order Method
2023, 2023 European Control Conference, ECC 2023Autonomous Tracking and State Estimation With Generalized Group Lasso
2022, IEEE Transactions on CyberneticsRecent Advances in Quadratic Programming Algorithms for Nonlinear Model Predictive Control
2018, Vietnam Journal of Mathematics