Through the Small Business Innovative Research (SBIR) program, Intra-Tech has submitted a proposal to the Missile Defense Agency (MDA) to obtain funding for the development and commercialization of algorithms and software that provides advance warning of failures of mission critical computerized equipment and machines. The following describes Intra-Tech's novel approach:
A Model based Predictive (FDI) Failure Detection and Identification System
We propose a data driven prognostic system to predict and identify failures by developing a model reference approach using state space techniques including the use of minimal order and function observers, sub-optimal Kalman filtering and generation and evaluation of residuals. Using sub-optimal and minimal order observers will reduce computational complexity which is important since a large number of failure prediction algorithms will run simultaneously in separate threads, each looking for a specific failure signature. Intra-Tech's innovative approach reduces the FDI (Failure Detection and Identification) problem to that of mapping failures to a line generated by a 2-dimensional scatter plot. Intra-Tech will provide data driven monitoring software that uses object-oriented class libraries consisting of algorithms that will process samples of operating data from sensors and effectors. A configuration data file will instruct the monitoring software which predictive engines, models, variances, parameters and values to use. The software developed will be useful in a variety of industrial and defense systems requiring FDI capability and will have immediate benefit for the MDA by providing MDA systems with a way of receiving advance warning of failures.
The monitoring software and library developed will be useful in a variety of systems requiring failure detection including aircraft, automotive, navigation, and power plants and will benefit these systems with a way of receiving advance warning of failures. Additional benefits of our FDI Monitoring and object-oriented Class Library software are safer systems, increased mission effectiveness, optimized maintenance, prevention of catastrophic failures and reduced costs due to the new longevity and efficiency of equipment. Also, more errors can be detected and isolated due to reduced computation needed to implement each predictive engine for each failure signature.
function observers, fault detection and isolation, dynamical systems, sub-optimal Kalman filters, residual generation and evaluation, monitoring software, malfunction, algorithms