An alternative to nonlinear estimation in gun fire control systems

Abstract

A Kalman filter with assumed linear system dynamics was derived from a spherical coordinate state vector and applied to a fire control system tracking environment. Two models of target dynamics were developed: a constant-velocity model and a correlated random acceleration model. The predicted position performance of the filters derived from the two basic models was optimized using Monte Carlo methods against a set of test trajectories that include constant-velocity maneuvering target profiles. The results of the Monte Carlo simulation of these filters are compared with the results obtained from a filter derived from a Cartesian Coordinate state vector. Switch-on-range adaptive filters were developed from the two basic models and evaluated by Monte Carlo methods. The results of the simulations of the adaptive filters with the two different state vectors are compared.