High-performance small-scale solvers for linear Model Predictive Control

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

1 Downloads (Pure)

Abstract

In Model Predictive Control (MPC), an optimization problem needs to be solved at each sampling time, and this has traditionally limited use of MPC to systems with slow dynamic. In recent years, there has been an increasing interest in the area of fast small-scale solvers for linear MPC, with the two main research areas of explicit MPC and tailored on-line MPC. State-of-the-art solvers in this second class can outperform optimized linear-algebra libraries (BLAS) only for very small problems, and do not explicitly exploit the hardware capabilities, relying on compilers for that. This approach can attain only a small fraction of the peak performance on modern processors. In our paper, we combine high-performance computing techniques with tailored solvers for MPC, and use the specific instruction sets of the target architectures. The resulting software (called HPMPC) can solve linear MPC problems 2 to 8 times faster than the current state-of-the-art solver for this class of problems, and the high-performance is maintained for MPC problems with up to a few hundred states.
Original languageEnglish
Title of host publicationProceedings of European Control Conference (ECC) 2014
PublisherIEEE
Publication date2014
Pages128-133
DOIs
Publication statusPublished - 2014
Event13th European Control Conference (ECC) 2014 - Strasbourg Convention and Exhibition Center, Strasbourg, France
Duration: 24 Jun 201427 Jun 2014
Conference number: 13
http://www.ecc14.eu/

Conference

Conference13th European Control Conference (ECC) 2014
Number13
LocationStrasbourg Convention and Exhibition Center
CountryFrance
CityStrasbourg
Period24/06/201427/06/2014
Internet address

Keywords

  • control engineering computing
  • linear algebra
  • optimisation
  • parallel processing
  • predictive control
  • Power, Energy and Industry Applications
  • Robotics and Control Systems
  • Signal Processing and Analysis
  • Transportation
  • high-performance computing technique
  • high-performance small-scale solvers
  • IP networks
  • Kernel
  • Libraries
  • linear model predictive control
  • linear MPC
  • Matrices
  • optimization problem
  • optimized linear-algebra libraries
  • Program processors
  • Registers
  • state-of-the-art solvers
  • Vectors

Cite this

@inproceedings{dc04d1911fe240a49f6628c2844b53ce,
title = "High-performance small-scale solvers for linear Model Predictive Control",
abstract = "In Model Predictive Control (MPC), an optimization problem needs to be solved at each sampling time, and this has traditionally limited use of MPC to systems with slow dynamic. In recent years, there has been an increasing interest in the area of fast small-scale solvers for linear MPC, with the two main research areas of explicit MPC and tailored on-line MPC. State-of-the-art solvers in this second class can outperform optimized linear-algebra libraries (BLAS) only for very small problems, and do not explicitly exploit the hardware capabilities, relying on compilers for that. This approach can attain only a small fraction of the peak performance on modern processors. In our paper, we combine high-performance computing techniques with tailored solvers for MPC, and use the specific instruction sets of the target architectures. The resulting software (called HPMPC) can solve linear MPC problems 2 to 8 times faster than the current state-of-the-art solver for this class of problems, and the high-performance is maintained for MPC problems with up to a few hundred states.",
keywords = "control engineering computing, linear algebra, optimisation, parallel processing, predictive control, Power, Energy and Industry Applications, Robotics and Control Systems, Signal Processing and Analysis, Transportation, high-performance computing technique, high-performance small-scale solvers, IP networks, Kernel, Libraries, linear model predictive control, linear MPC, Matrices, optimization problem, optimized linear-algebra libraries, Program processors, Registers, state-of-the-art solvers, Vectors",
author = "Gianluca Frison and S{\o}rensen, {Hans Henrik Brandenborg} and Bernd Dammann and J{\o}rgensen, {John Bagterp}",
year = "2014",
doi = "10.1109/ECC.2014.6862490",
language = "English",
pages = "128--133",
booktitle = "Proceedings of European Control Conference (ECC) 2014",
publisher = "IEEE",
address = "United States",

}

Frison, G, Sørensen, HHB, Dammann, B & Jørgensen, JB 2014, High-performance small-scale solvers for linear Model Predictive Control. in Proceedings of European Control Conference (ECC) 2014. IEEE, pp. 128-133, 13th European Control Conference (ECC) 2014, Strasbourg, France, 24/06/2014. https://doi.org/10.1109/ECC.2014.6862490

High-performance small-scale solvers for linear Model Predictive Control. / Frison, Gianluca; Sørensen, Hans Henrik Brandenborg; Dammann, Bernd; Jørgensen, John Bagterp.

Proceedings of European Control Conference (ECC) 2014. IEEE, 2014. p. 128-133.

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

TY - GEN

T1 - High-performance small-scale solvers for linear Model Predictive Control

AU - Frison, Gianluca

AU - Sørensen, Hans Henrik Brandenborg

AU - Dammann, Bernd

AU - Jørgensen, John Bagterp

PY - 2014

Y1 - 2014

N2 - In Model Predictive Control (MPC), an optimization problem needs to be solved at each sampling time, and this has traditionally limited use of MPC to systems with slow dynamic. In recent years, there has been an increasing interest in the area of fast small-scale solvers for linear MPC, with the two main research areas of explicit MPC and tailored on-line MPC. State-of-the-art solvers in this second class can outperform optimized linear-algebra libraries (BLAS) only for very small problems, and do not explicitly exploit the hardware capabilities, relying on compilers for that. This approach can attain only a small fraction of the peak performance on modern processors. In our paper, we combine high-performance computing techniques with tailored solvers for MPC, and use the specific instruction sets of the target architectures. The resulting software (called HPMPC) can solve linear MPC problems 2 to 8 times faster than the current state-of-the-art solver for this class of problems, and the high-performance is maintained for MPC problems with up to a few hundred states.

AB - In Model Predictive Control (MPC), an optimization problem needs to be solved at each sampling time, and this has traditionally limited use of MPC to systems with slow dynamic. In recent years, there has been an increasing interest in the area of fast small-scale solvers for linear MPC, with the two main research areas of explicit MPC and tailored on-line MPC. State-of-the-art solvers in this second class can outperform optimized linear-algebra libraries (BLAS) only for very small problems, and do not explicitly exploit the hardware capabilities, relying on compilers for that. This approach can attain only a small fraction of the peak performance on modern processors. In our paper, we combine high-performance computing techniques with tailored solvers for MPC, and use the specific instruction sets of the target architectures. The resulting software (called HPMPC) can solve linear MPC problems 2 to 8 times faster than the current state-of-the-art solver for this class of problems, and the high-performance is maintained for MPC problems with up to a few hundred states.

KW - control engineering computing

KW - linear algebra

KW - optimisation

KW - parallel processing

KW - predictive control

KW - Power, Energy and Industry Applications

KW - Robotics and Control Systems

KW - Signal Processing and Analysis

KW - Transportation

KW - high-performance computing technique

KW - high-performance small-scale solvers

KW - IP networks

KW - Kernel

KW - Libraries

KW - linear model predictive control

KW - linear MPC

KW - Matrices

KW - optimization problem

KW - optimized linear-algebra libraries

KW - Program processors

KW - Registers

KW - state-of-the-art solvers

KW - Vectors

U2 - 10.1109/ECC.2014.6862490

DO - 10.1109/ECC.2014.6862490

M3 - Article in proceedings

SP - 128

EP - 133

BT - Proceedings of European Control Conference (ECC) 2014

PB - IEEE

ER -