摘要
针对用单批次激励试验数据进行多变量系统的智能辨识的不准问题,提出了一套基于M次不相关激励和汇总智能优化的多变量系统辨识方法。该方法的要点是:对于有M维输入的多变量系统的准确辨识至少需要M次不相关激励试验,然后采集试验数据,再用汇总智能优化指标进行多变量系统模型的辨识计算。通过理论推导证明了多变量系统辨识需要M次的不相关激励试验的必要性。通过一个应用实例验算,演示了新的多变量系统辨识方法的易用性、正确性和工程实用性。
Aiming at the inaccuracy problem of identification of multivariable system with single batch excitation experiment data,a new multivariable identification method based on M-times uncorrelated excitation and assembly intelligent optimization is proposed.The main point of this method is that for the accurate identification of multivariable system with M-dimensional input,at least M-times identification excitation experiments are required,and the experiments data is obtained.And then,the identification calculation of multivariable system model is carried out with the aggregated intelligent optimization index.Throuth theoretical derivation,the necessity of uncorrelated excitation tests requiring M times for multivariable system identification is proved.An application example is given to demonstrate the usability,correctness and engineering practicability of the new multivariable system identification method.
作者
徐春梅
杨平
李芹
康英伟
于会群
彭道刚
XU Chun-mei;YANG Ping;LI Qin;KANG Ying-wei;YU Hui-qun;PENG Dao-gang(College of Automation Engineering,Shanghai University of Electric Power,Shanghai 200090 China)
出处
《自动化技术与应用》
2021年第9期94-98,共5页
Techniques of Automation and Applications
基金
上海市“科技创新行动计划”高新技术领域项目(编号17511109400)
上海市科学技术委员会工程技术研究中心项目资助(编号14DZ2251100)。
关键词
多变量系统
系统辩识
不相关激励
汇总优化
智能优化
multivariable system
system identification
irrelevant excitation
assemble optimization
intelligent optimization
