A Neural Network-Based Simulation Study for the Performance Monitoring of a Military Turbojet Engine

• Main Authors: CHOU, CHE-MING, 周哲名
• Other Authors: SHIAH, SHEAU-MING
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/sp43uw

碩士 === 國防大學理工學院 === 機械工程碩士班 === 102 === In viewing to the importance of performance monitoring and trouble shooting of military engines to the readiness of air force aircrafts, the study of performance evaluation thus is of the present interests. Therefore, how to apply theorem with air-thermodynamics to an engine performance monitoring tool is valuable in researches for national defense science. To focalize, the case of an turbine engine was chosen as a typical example for the performance monitoring study, due to its popularity in modern turbine engines, while in the turbine engine performance monitoring, the artificial neural networks (ANN) algorithm was adopted in simulating the turbine engine operation mechanism, for its capability to handle highly non-linear data set with less known physics. As to the whole research works, it includes: (1) a theoretical study of the turbine engine operating characteristics, based on thermodynamics and fluid kinematics, (2) an extensive calculation study of the turbine engine performance to find the parametric analysis, (3) a comprehension study of basic ANN modeling, training and validation, (4) an ANN-based case study for the performance monitoring of a turbine engine. Through above studies, a set of two MATLAB programs was developed and used to conduct a simulation case study for the US military J85 turbojet. Major results thus obtained include the J85 turbojet engine performance was simulated by using the software of GasTurb, and assessing it’s the health state based Artificial Neural Network analysis framework. Finally, provide one kind of maintenance suggestion in academic and practice.