On recognition method of aeroengine working conditions based on flight data

According to the complexity of the recognition of aeroengine working conditions and the time-consuming problem of artificial recognition, some intelligent recognition methods based on flight data were proposed in this paper. Four parameters which respectively were throttle position, high-pressure rotor speed, jet nozzle diameter and exhaust gas temperature were selected as sample data according to the working principle of aeroengine, and three multi-class classifications based on the Least Squares Support Vector Classification machine which respectively were One Versus One, One Versus the Rest and Error-Correcting Output Codes were trained and tested. Then the Error-Correcting Output Codes method which had less operation time and high test precision was used to recognize the five kinds of aeroengine working conditions which included stopping, crawling, cruising, biggest and boosting, and the results showed that it had a high recognition accuracy and good adaptability, and was convenient and quick, which laid the foundation for aeroengine condition monitoring and life expectency statistics.