Pattern Recognition for Real-Time Performance Tracking and Failure Prediction in Complex Systems

The CLDS has developed new approaches to near-instantaneous agent-enabled learning for the purposes of real-time performance tracking, failure prediction and decision support. Information overload and data complexity challenges in distributed information networks are demanding more powerful, scalable solutions to pattern and fault recognition, especially for complex systems like aircraft and defense platforms. Our research uses new associative memory technology that is capable of recognizing patterns in performance data in order to anticipate component or system failure in vehicles such as trucks and aircraft. The learning agents are capable of observing and learning complex correlations across multiple parameters, and collaborating with other agents across a fleet. These agents lend themselves to distributed multi-agent configurations for real-time networked visibility and decision support across complex functions including supply chain management, maintenance and fleet control.