Muliple Model Adaptive Estimation (MMAE) Based Filter Banks for Interception of Maneuvering Targets
George Marks · AIAA Guidance, Navigation, and Control Conference and Exhibit · 2007
The Multiple Model Adaptive Estimation (MMAE) approach to determining an unknown parameter within a known range has been around since it was introduced by D. T. Magill in 1965. With the ever increasing computing power of digital computers it has become practical to explore multiple Kalman filter solutions. Signal, Image, and Navigation processing have all exploited this increased computing capability. In particular Yaakov BarShalom has extended the multiple-model Kalman Filter theory and approach to include interaction between the Kalman filters for both target tracking and navigation. This paper presents the MMAE approach as a viable alternative to engage maneuvering targets. The basic approach is to tune several Linear Kalman filters to different frequencies or maneuver types and then apply the Likelihood Function and Bayes' rule to determine the probability that each Kalman filter is the correct model. The composite weave frequency and filter states are determined by summing the weighted frequencies and states using. The weighting terms are Bayes’ rule probabilities computed for each filter. The MMAE approach combined with the proper zero effort miss (ZEM) guidance law provides a practical systematic approach to solving the maneuvering target interception problem while preserving the ability to intercept ballistic targets. Analysis based on Six-Degree-ofFreedom (6-DOF) modeling of the interceptor will illustrate good miss distance performance achieved by the MMAE/ZEM approach for ballistic and maneuvering re-entry vehicles.