Kalman Filter-Based Identification of Systems with Randomly Missing Measurements and Linear Constraints

Abstract

Abstract The available information of linear constraint in linear dynamic systems, which is often unexplored in previous works, is taken advantage of to improve the accuracy of the parameter estimation, particularly in the presence of randomly missing measurements. Specifically, a Kalman filter-based identification for systems without constraint but with the randomly missing measurements is first introduced. Then the result is extended to systems with linear constraint under normal conditions. By doing so we show that the accuracy of the estimation is improved by taking the constraint into account, both theoretically and numerically.