Open-Loop System Identification of MIMO Integrator Model

Abstract

This paper considers the open-loop system identification of a multi-input multi-output (MIMO) integrator model which is a multiplication of a constant matrix, called the action matrix, and the integrator. Our problem is to estimate the action matrix from the measured input and output data. The estimation algorithm is proposed by removing trends from the numerically integrated input and output so that the effects of input and output disturbances are eliminated. The proposed method is applicable to the basic scenario (constant input and output offsets), which is similar to previous studies, and the non-constant output offset; however, it is not applicable to the non-constant input offset. The discussion in this paper indicates the essential difficulties in the open-loop system identification with the integrator.

• system identification
• parameter estimation
• multi-input multi-output
• integrator