Autonomous Control System Components

We will develop and demonstrate smart system component designs and algorithms enabling real-time autonomous control systems. Autonomous control system components will improve facility safety and reduce maintenance costs by enabling automated operation, system fault tolerance, on-condition maintenance and fewer false alarms. Our innovation consists of novel algorithms for fault detection and classification that enable real-time self-diagnostic surveillance and autonomous fault accommodation by a smart system component. Our approach combines smart sensor and actuator designs, previously developed for NASA Stennis Space Center, with patented multivariate state estimation algorithms from Argonne National Laboratory (ANL) that have been shown to provide high fidelity sensor and component failure detection for nuclear power plants and rocket engines. We will develop microcontroller-compatible versions of the ANL algorithms and combine these with a new Bayesian belief network expert system algorithm to automatically classify and respond to a detected fault of a smart sensor or control component. Phase I will culminate in demonstration of prototype smart components that perform automated self-diagnostic monitoring and fault management for a hazardous gas detection and control system applicable to a wide range of NASA ground test facilities, test articles and flight systems. ~000025 ~