彭海军

个人信息Personal Information

教授

博士生导师

硕士生导师

性别:男

毕业院校:大连理工大学

学位:博士

所在单位:力学与航空航天学院

学科:动力学与控制. 计算力学. 工程力学

电子邮箱:hjpeng@dlut.edu.cn

扫描关注

论文成果

当前位置: 中文主页 >> 科学研究 >> 论文成果

Explicit expression-based practical model predictive control implementation for large-scale structures with multi-input delays

点击次数:

论文类型:期刊论文

发表时间:2018-06-01

发表刊物:JOURNAL OF VIBRATION AND CONTROL

收录刊物:SCIE

卷号:24

期号:12

页面范围:2605-2620

ISSN号:1077-5463

关键字:Large-scale structures; vibration suppression; model predictive control; multi-input delays; explicit expression; Newmark- method

摘要:In this paper, two practical model predictive control (MPC) implementation algorithms with multi-input delay (NFMPCMID1 and NFMPCMID2) are developed in discrete-time formulation for vibration control of large-scale structures. By introducing a particular augmented state vector, the controlled dynamic equation with multi-input delay is transformed into the standard form without any explicit time delay. Because of no approximation for multi-input delay involved, the system performance and stability are easily guaranteed. In order to solve the computation efficiency and memory requirement for large-scale structure, a novel explicit expression form of Newmark- method is derived, from which the future states can be easily predicted without computing matrix exponential and its integration. By applying this explicit expression form into MPC, the control input of NFMPCMID1 method can be computed by some matrix-matrix multiplications, and also the control input of NFMPCMID2 method can be computed just by two off-line transient analyses and one on-line transient analysis at every sampling instant on the structure. For no computation of matrix exponential and its integration in NFMPCMID1 and NFMPCMID2 methods, the off-line computation efficiency is greatly improved, and the memory requirement is greatly reduced, especially for the NFMPCMID2 method. In additional, due to the small amount of on-line computation, the on-line computation efficiency is also guaranteed. At last, the stability, feasibility and efficiency of the proposed methods are verified by several typical numerical examples.