Accession Number : ADA190520
Title : Kalman Filter Residual Expert System.
Descriptive Note : Master's thesis,
Corporate Author : AIR FORCE INST OF TECH WRIGHT-PATTERSON AFB OH SCHOOL OF ENGINEERING
Personal Author(s) : Grimshaw, Jeffrey D
PDF Url : ADA190520
Report Date : Dec 1987
Pagination or Media Count : 189
Abstract : The Pilot's Associate (PA) Program has been initiated to help mitigate the extensive workload of the fighter pilot. To operate effectively, the PA system must have situation awareness: the status of important on-board and off-board systems. This knowledge is gained through sensor systems. The data from these systems must be fused together to present the PA with a coherent picture of the internal (on-board) and external (off-board) states. Although many types of information can be extracted from sensor data, this paper emphasizes those parameters that help determine target track. One common technique for fusing sensor data uses Kalman filters. In a multiple model adaptive filter (MMAF) system, the most appropriate Kalman filter is chosen. This filter provides the best estimates of the desired states. An operating MMAF system continually selects which filter to use as the basis for the state estimates. The overall accuracy of the system is closely related to how well the filters are selected. Previous filter selection techniques have proved useful, but limited. To overcome some of these limitations, an expert system, KREST, was developed so that expert rules could be used to select filters. Although no quantitative estimate of improvement is available, the MMAF expert stated that KREST exhibited a potentially significant improvement over the previously used filter selection techniques.
Descriptors : *ADAPTIVE FILTERS, *KALMAN FILTERING, *PILOTS, *ARTIFICIAL INTELLIGENCE, *FLIGHT CONTROL SYSTEMS, ACCURACY, AWARENESS, DETECTORS, ESTIMATES, FIGHTER AIRCRAFT, FILTERS, MODELS, SELECTION, TARGETS, THESES, WORKLOAD
Subject Categories : Computer Programming and Software
Military Aircraft Operations
Distribution Statement : APPROVED FOR PUBLIC RELEASE