A fast MPC algorithm for reducing computation burden of MIMO
☆
Rongbin Qi
⁎
, Hua Mei
⁎
, Chao Chen, Feng Qian
Key Laboratory of Advanced Control and Optimization for Chemical Process, Ministry of Education, East China University of Science and Technology, Shanghai 200237, China
abstractarticle info
Article history:
Received 3 June 2015
Received in revised form 17 July 2015
Accepted 29 July 2015
Available online 19 October 2015
Keywords:
Fast MPC algorithm
Computation burden
One-bit operation
Dimension reduction
The computation burden in the model-based predictive control algorithm is heavy when solving QR optimization
with a limited sampling step, especially for a complicated system with large dimension. A fast algorithm is pro-
posed in this paper to solve this problem, in which real-time values are modulated to bit streams to simplify the
multiplication. In addition, manipulated variables in the prediction horizon are deduced to the current control
horizon approximately by a recursive relation to decrease the dimension of QR optimization. The simulation re-
sults demonstrate the feasibility of this fast algorithm for MIMO systems.
© 2015 The Chemical Industry and Engineering Society of China, and Chemical Industry Press. All rights reserved.
1. Introduction
As one of the most widely applied advanced control algorithms,
model-based predictive control (MPC) is introduced in 1 970s from
industrial applications as one of the most effective control strate-
gies. It solves many problems in i ndustrial control field while tradi-
tional proportion- integral-der ivative (PID) alg orithm and modern
control theory fail. However, complex on-lin e QR optimizat ions
and matrix multiplication in a limited sampling step leads to
heavy computation burden, whi ch l imits the application of MPC
algorithm.
Reduction of computational burden of MPC algorithm has been
paid much atte ntion in academia and industry. Many efforts have
been made to improve the algorithm from mathematic vision or
adjust the optimization structure to reduce computation burden.
Diehl et al. [1] proposed a real-time iteration approach to simplify
QR optimization process. Kouvaritakis and Can non [2] described
an expan ded Newto n–Raphson algorithm to replace a universal
QP optimization formulation. Wang and B oyd [3] described a cus-
tom method with a pa rticular structure that can work much faster
than the generic opt imization me thod. Bem porad et al. [4] and
Johansen et al. [5] proposed an explicit MPC algorithm, which
sets state variables to optimal inputs as a piecewise affine func-
tion. Ding and Huang [6] and Wan and Kothare [7] calculated ma-
nipulated variables in p rediction horizon off-line approximately
with a recursiv e relation to simplify the on-line opt imization.
Medagoda and Gibbens [8] used inconsistent predictive horizon
interval to reduce computation burden. Another method simpli-
fied matrix multiplication by using bit-steams based schema [9].
Although some methods are efficient, most of them are too com-
plicated to re alize.
This work presents a fast MPC al gorit hm for solving these prob-
lems i n QR optimization and on-line matrix multiplicat ion. Real
variables are modulated to bit streams represented by ±1, to replace
conventional coefficient multi plication. Control variables are
expressed by current control actuation approximately with a recur-
sive relation rather t han obtaining all of t he control actuations as
control variables. The algorithm is applied to MIMO system to dem-
onstrate its feasibility.
2. Fast MPC Algorithm
The fast MPC algorithm is proposed to decrease the computation
burden.
2.1. Dimension reduction
An expanded method based on [10] is de duced t o reduce the
matrix dimension of QR optimization problem. For a linear time
Chinese Journal of Chemical Engineering 23 (2015) 2087–2091
☆
Supp orted by the National Natural Science Foundation of China (61333010,
61203157), the Fundamental Research Funds for the Central Universities, the National
High-Tech Research a nd Development Program of C hina (2013AA040701), Shanghai
Natural Science Foundation Project (15ZR1408900), and Shanghai Key Technologies
R&D Program Project (13111103800).
⁎ Corresponding authors.
E-mail addresses: qirongbin@ecust.edu.cn (R. Qi), hmei2006@ecust.edu.cn (H. Mei).
http://dx.doi.org/10.1016/j.cjche.2015.10.008
1004-9541/© 2015 The Chemical Industry and Engineering Society of China, and Chemical Industry Press. All rights reserved.
Contents lists available at ScienceDirect
Chinese Journal of Chemical Engineering
journal homepage: www.elsevier.com/locate/CJChE