H (formula) based filtering for systems with time delays and application to vehicle tracking

Abstract

In this thesis, the filtering problem for linear systems with time delay is studied. The standard mixed sensitivity problem is investigated and the duality between H∞ control problem and H∞ filtering problem is established. By using this duality an alternative H∞ filtering method is proposed. An optimum H∞ filter is designed for single output system. However the proposed technique does not only work for single delay in the measurement but also works for multiple delays in both state and measurement if the linear system has more than one outputs with delay. For this case, different suboptimal filters are designed. This work also deals with an important aspect of tracking problems appearing in many different applications, namely state estimation under delayed and noisy measurements. A typical vehicle model is chosen to estimate the delayed state and the performances of the designed filters are examined under several scenarios based on different parameters such as amount of delays and disturbances.