Using Neural Networks to Detect Failure Onset in Complex Systems

Abstract

Successful prognostic and health Muscrave, and Lin [2] describe the use of an automonitoring systems depend on being able to recognize the associative neural network (AANN) in a sensor validation signs of a failure in progress. Although such systems are application for reusable rocket engines. These findings are commonplace, little has been reported to date on fault amplified in a master's thesis by Najafi [3]. Furthermore, detection for systems where the interactions of the various these papers postulate the use of the output of the network operating parameters are subtle, complex, and correlated to replace data lost by failed sensors, due to the AANN's in unknown or difficult to elicit ways. This paper describes property of "correcting" anomalous data. Wang [4] the results of recent research into the use of neural elucidates a neural network gated expert in conjunction networks to provide detection of the onset of operational with the use of an AANN; she also corroborates the use of failure in such devices. After a preliminary exploration the AANN in monitoring the sensors in a health revealed the shortcomings of more common pattern monitoring system. Expert knowledge is used in the recognition methods, such as limit checking, a posteriori PROMISE diagnostic and prognostic system, as reported Baysean methods, and even principal component analysis, by Biagetta and Sciubba [5]. A cogeneration plant in Italy it is shown that certain types of neural networks are up to uses an expert system and system model to augment the the task. The results from simulations will show the health monitoring system. Here, the selection of the effectiveness neural network techniques in detecting the appropriate diagnosis when a detected fault could be onset of the failure. These techniques will then be ascribed to more than one cause, and to provide the demonstrated on data from a real-world system and the prognostic function is aided by expert knowledge and the results presented. system model; once a consensus is achieved, the system then produces a prognostic forecast and suggests measures

DOI: 10.1109/SYSOSE.2007.4304274

7 Figures and Tables

Cite this paper

@inproceedings{Stone2007UsingNN, title={Using Neural Networks to Detect Failure Onset in Complex Systems}, author={Victor M. Stone}, booktitle={SoSE}, year={2007} }