A Varying Interval-Based Multi-Innovation Filtered Gradient Method for Bilinear-in-Parameter Systems With Missing Measurements

Colored noise is quite common in the industrial field, posing difficulties to the parameter estimation of system identification. Moreover, missing measurements exacerbate these challenges. In this paper, to overcome the colored noise interference, a filtering-based identification model is established by designing a linear filter for input and output data. Subsequently, a varying-interval-based sampling method is designed to address missing measurements by adjusting sampling intervals. Eventually, this paper proposes a varying interval-based multi-innovation filtered gradient (VI-MIFG) method for bilinear-in-parameter systems, where the residual is able to be represented by linear combinations of the innovation. The proposed algorithm can fully leverage the observed measurements and multi-innovation information. A computation analysis and a numerical case validate the effectiveness of VI-MIFG.