NASA 1996 SBIR Phase I


PROPOSAL NUMBER : 96-1 01.17-3155

PROJECT TITLE : Real-Time Health Monitoring of Flight-Critical Systems Using Fuzzy CMAC

TECHNICAL ABSTRACT (LIMIT 200 WORDS)

Many existing approaches to fault detection and diagnostic systems are after-the-fact methods, i.e. failure has occurred and the machine is down. As a result, the mean-time-between-failure (MTBF) is short and many man-hours are needed to repair the machine. The key to preventing failure to occur is the ability to identify degraded states of the machine so that machine can be repaired during regular maintenance hours.

In this proposal, we propose a new method to machine degradation monitoring, fault detection and diagnostics, which is robust to sensor noise and is also efficient in training and learning. We are also able to detect new fault conditions that have not occurred before, which may include sensor failures and hence the capability of validity self-checks. Our idea is to use a new type of biologically-realistic neural net, called Fuzzy CMAC. The Fuzzy CMAC (Cerebellar Model Arithmetic Computer) inherits preferred features of arbitrary function approximation and parallel processing from the original CMAC neural network, and the capability of acquiring and incorporating human knowledge into a system and the capability of processing information based on fuzzy inference rules from the fuzzy logic. Our learning rates are at least an order of magnitude faster than conventional neural nets. The Fuzzy CMAC can be designed in such a way that each output corresponds to one type of failure condition. Hence the neural network is able to monitor the degradation and to identify the type of failure simultaneously.

POTENTIAL COMMERCIAL APPLICATIONS
The range of applications of the Fuzzy CMAC is very large. Other applications include medical monitoring, image classification, self-repairing control systems, system health monitoring, in-process control of manufacturing processes, and real-time error compensation for machine tools using calibration data to train the network.
NAME AND ADDRESS OF PRINCIPAL INVESTIGATOR
Dr. C. M. Kwan
Intelligent Automation, Inc.
2 Research Place, Suite 202
Rockville, MD 20850
NAME AND ADDRESS OF OFFEROR
Intelligent Automation, Inc.
2 Research Place, Suite 202
Rockville, MD 20850