NASA 1998 SBIR Phase I

PROPOSAL NUMBER: 98-1 01.05-5355

PROJECT TITLE: Automated On-line Health Monitoring, Failure Detection and Identification using Interacting Multiple Models (IMM) Kalman Filters


On-line autonomous sensor and actuator Health Monitoring-Failure Detection and Identification (HM-FDI) is becoming a critical issue in aircraft health monitoring, as the number of on-board and off-board sensors and actuators, as well as the complexity of each one of these sensors, has increased substantially during the recent past. SSCI propose an innovative approach to on-line autonomous Health Monitoring, Failure Detection and Identification for aircraft and spacecraft systems based on a novel combination of the Interacting Multiple Model (IMM) approach and on-line System Identification (SI), that will be called ``adaptive IMM''. The IMM algorithm was originally developed, and has been successfully used, for tracking maneuvering targets, but its application to FDI is completely new. In this new approach, each possible (anticipated) failure is mapped into a different state space model. For each one of these models, a Kalman filter (KF) computes the individual state estimates and the likelihood function corresponding to that model. The individual estimates are weighted with the probability of each model being correct (derived from the likelihood functions) to give a single (accurate) state estimate. SI is needed for detecting and identifying unanticipated failures (for example, partial degradation due to battle damage of an actuator surface or aircraft surface). On-line SI is performed using an approach based on Deterministic/Stochastic Realization Algorithms (DSRA) which identifies a state space model of the system from the multi-input multi-output (MIMO) measured data.


The Health Monitoring/FDI system to be developed during this Phase I will be of direct application to Bell Helicopter's Eagle Eye UAV (already being fly-tested). Bell has expresed its interest in the HM/FDI system proposed here, and a succesfull completion of Phase I will lead to a prototype implemetantion during Phase II to be tested at Bell's flight lab and hot-bench facilities, and commercialization during Phase III. There is also an increasing number of industrial application where advanced process control is heavily dependent on sensor information, and a sensor FDI system would be of direct application. In essence, any closed loop system that use information provided by a sensor will benefit from an FDI.


Constantino Rago
Scientific Systems Company, Inc
500 West Cummings Park, Suite 3000
Woburn , Ma 01801


Scientific Systems Company, Inc.
500 W. Cummings Park Suite 3000
Woburn , MA 01801-6503