Research On State Monitoring And Life Prognosis Approach For Helicopter Electro-mechanical Actuator

Electro-mechanical actuator (EMA) is the essential attitude control componentof helicopter. The EMA with DC motor has the shortcomings of high fault rate andshort life and poor reliability and restricts the enhancement of helicopter s reliability.Current research main focuses on posterior-diagnosis and prognosis offline whichcan t ensure safe flight. In order to solve the reliability problem of EMA, this paperstudies on life prognosis techniques online that can provide a basis for intelligentmaintenance and improve the reliability of the helicopters effectively.The state signal of EMA has abundant performance information. Thedegradation process can be analyzed through the changes of characters. Currentresearch which only used single character of current or vibration can t reflect thereal changes sufficiently. This paper brings forward7characters from current, rotatespeed and vibration via EMAs fault tree and mathematical model. Because theresearch object s is special, so fault criterion is hard to gain. Fault injectionsimulation is used to determine the failure thresholds which can provide sufficientcriterion for prognosis and diagnosis.Reasonable feature extraction is the key of prognosis. Based on the charactersof signals, wavelet transform is used for current s and speed s feature extraction andHilbert-Huang transform is used for vibration s feature extraction. Degradationtracks are gained through Feature extraction that provides data support for prognosis.There are many methods of prediction and modeling for nonlinear time series.However, for such a complex system like EMA, a single prognosis method isdifficult to ensure its prediction, so a combining prognosis approach named GM-SVM is proposed. The test shows the GM-SVM prognosis approach that cancombine the advantage of GM and SVM has stronger adaptive capacity and higherprediction accuracy. Then the GM-SVM approach is used for the life prediction andthe prediction accuracy Life is up to84%.Life prognosis of EMA bases on a large of degradation data. The system forstate monitoring and life prognosis is developed which has the capability ofsynchronous monitoring multiple status signals and can achieve online lifeprognosis. The system provides an good experiment platform for the research onintelligent maintenance and reliability of EMAs.