LSSVM with fuzzy pre-processing model based aero engine data mining technology

Abstract

The operations of aircraft fleets typically result in large volumes of data collected during the execution of various operational and support processes.This paper reports on an Airlines-sponsored study conducted to research the applicability of data mining for processing engine data for fault diagnostics. The study focused on three aspects: (1) understanding the engine fault maintenance environment, and data collection system; (2) investigating engine fault diagnosis approaches with the purpose of identifying promising methods pertinent to aircraft engine management; and (3) defining a Support Vector Machines model with Fuzzy clustering to support the data mining work in aero engine fault detection. Results of analyses of maintenance data and flight data sets are presented. Architecture for mining engine data is also presented. © Springer-Verlag Berlin Heidelberg 2007.