Model-Based Sensor Validation for a Turbofan Engine Using Auto-Associative Neural Networks

Autor:	Ten-Huei Guo, Duane L. Mattern, Link C. Jaw, C. L. Philip Chen
Rok vydání:	2003
Předmět:	Task (computing) Artificial neural network Computer science Dimensionality reduction Reading (computer) Real-time computing Isolation (database systems) Software Simulation Fault detection and isolation Associative property Turbofan
Zdroj:	International Journal of Smart Engineering System Design. 5:21-32
ISSN:	1607-8500 1025-5818
DOI:	10.1080/10255810305035
Popis:	Sensor validation is an essential task in the operation of an aircraft engine. Faulty sensor readings can lead to undesirable situations, such as dispatch delays, degraded engine performances, and possible unsafe operations. Auto-Associative Neural Networks (AANN) have been known to provide dimension reduction through the nonlinear principal component analysis. This characteristic can be applied to a system with redundant sensors and to identify and recover faulty sensor readings. In this work, we have designed an AANN-based sensor validation approach for a turbofan aircraft engine. We use a model-based approach for sensor validation in which an estimated sensor reading obtained from AANN can be used to replace failed sensor values in a feedback control system. Simulations of the sensor fault detection, isolation, and accommodation of a closed-loop operation of an aircraft engine have been done.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::2853ae8ee00b2f1384b0e25cb365d991 https://doi.org/10.1080/10255810305035 Zobrazit plný text záznamu Plný text ve formátu PDF