Contextual normalization applied to aircraft gas turbine engine diagnosis

摘要: Diagnosing faults in aircraft gas turbine engines is a complex problem. It involves several tasks, including rapid and accurate interpretation of patterns engine sensor data. We have investigatedcontextual normalization for the development software tool to help repair technicians with Contextual new strategy employing machine learning. handles variation data that due contextual factors, rather than health engine. does this by normalizing context-sensitive manner. This learning was developed tested using 242 observations an test cell, where each observation consists roughly 12,000 numbers, gathered over 12-second interval. There were eight classes observations: seven deliberately implanted healthy class. compared two approaches implementing our strategy: linear regression instance-based three main results. (1) For given problem, works better regression. (2) other common forms normalization. (3) The algorithms described here can be basis useful assisting