A model uncertainty compensation filter

The Kalman filter theory was developed assuming ideal conditions (i.e., all plant dynamics and noise statistics are known exactly). These assumptions are not always valid when applying this theory to real world problems. This paper presents a new estimator, the Model Uncertainty Compensation (MUC) algorithm, that resolves the problem of applying Kalman filter theory to real-world problems where significant modeling errors exist. The MUC algorithm uses a feedback formulation, based on the LQG control scheme to compensate for non-Gaussian modeling errors. For this paper, a fourth order simulation is used to evaluate the performance of MUC algorithm. The system is chosen to highlight the capabilities of the MUC algorithm.