Targer State Estimator Design Using FIR Filter and Smoother

The measured rate of the tracking sensor becomes biased under some operational situation. For a highly maneuverable aircraft in 3D space, the target dynamics changes from time to time, and the Kalman filter using position measurement only can not be used effectively to reject the rate measurement bias error. To cope with this problem, we present a new algorithm which incorporate FIR-type filter and FIR-type fixed-lag smoother, and demonstrate that it has the optimal performance in terms of both estimation accuracy and response time through an application example to the anti-aircraft gun fire control system(AAGFCS). The Target State Estimator(TSE) takes a role of estimat- ing the target state through combining the measured data from tracking sight and a prior knowledge of the target motion in a statistically optimal fashion. Commonly, the filter is designed and evaluated in time domain and such is the case of the well- known Kalman filter. This filter requires us to know the exact knowledges of the target motion which can be described by the dynamic state equation and measurement model so as to com- pute the state estimate accurately. However, real motion of the target is frequently deviated from the assumed model, and es- timation error of the filter grows and diverges in some worst case. In this situation, adaptive tracking filters may be used, and they can be conveniently categorized into three different groups. First, switching of filter model is introduced by monitoring the estimated error. When the model needs to be changed, the statis- tics of the model uncertainty may be adjusted depending on the magnitude of the residual(1) or the order of the state model may be augmented(2) or the forcing input of the model may be computed directly through manipulation of the residual se- quence(3)(4). Second, the multiple model filters are presumed