SBIR-STTR Award

Engineering Research And Consulting, Inc. (ERCI) proposes to develop a prototype neural network based expert system (connectionist expert system) for ASTF fault diagnosis. Neural networks are excellent for parameter estimation and recognizing patterns in signal data. The objective of this research is to design a neural network architecture to detect fault thus signifying and responding to an abnoral event. Once a successful architecture is designed, an expert system will be built using the resulting neural network output curve as a premise to a rule. The neural network based expert system can advise the engineer and/or be integrated to the Emergency Detection And Response (EDAR) function which is a part of the ASTF fault-tolerant control system. Phase I of this project will implement a prototype neural architecture using fault data generated by a simulator. The resultant network output curves will be used as rule premise for an expert system, where rule clauses will be recommendations to engineers and/or the EDAR for control of ASTF. Phase II of the project will extend the system from Phase I to build a connectionist expert system for ASTF fault diagnosis.