2019 IEEE 15th International Conference on Control and Automation (ICCA)

3 / 2018-12-31 01:02:43

MIMO Closed-loop Identification based on Channels Isolation Using Leaky Least Mean Squares Based- Partial Correlation Method

Closed loop identification ; Partial Correlation ; Subsystem interactions ; Least Squares

Abstract-Closed-loop identification of multivariable systems faces some difficulties due to subsystems interactions. The correlation between the process input signals is the key factor that can be assessed to determine the strength of these subsystems interactions and subsequently the necessity to be included in the identification process. The conventional partial correlation analysis is the adopted technique to assess these inputs correlations. However, this technique is based on least squares estimation and hence, suffers from incorrect estimation of the model error due to the unmeasured disturbances and strong correlations between the process input signals. Therefore, this paper proposes the use of the Leaky Least Mean Squares-based partial correlation method (LLMS-Pr) for the isolation of insignificant interaction dynamics of multivariable closed-loop systems. Unlike the conventional partial correlation method, the proposed method clearly discriminates the interaction channels that have a significant contribution to the interconnected subsystems from the ones that do not by reducing the model error that arises due to the process inputs correlation. The efficacy of the proposed method is illustrated through a case study.