Remain Useful Life Prediction Of Engine Based On Similarity Method

With the development of mechanical technology and engineering technology,the continuous improvement of reliability requirements for key equipment,the increasing complexity of systems and components,as well as the requirements to improve the reliability of equipment,reduce operating risks and operating costs,it is necessary to take measures to evaluate the remaining useful life and state monitoring of equipment.The aero-engine is the most important component system on the aircraft.The system has high integration and complex structure,so it is necessary to take measures to monitor and maintain the aero-engine,so as to provide strong flight support for the aviation system.For this reason,PHM technology has been widely concerned by academia and industry.As one of the core technologies of PHM,the prediction and evaluation of the remaining useful life can predict the health status and future trend of the equipment in advance,and provide valuable reference for the health management of the equipment.At present,the prediction methods of residual service life are mainly divided into physical model-based methods,data-driven methods and experience-based methods.With the advent of the information age and the age of big data,a large amount of process data has been acquired,so data-driven methods have been more and more widely used.Since there is no general performance evaluation framework,some commonly used evaluation indexes are summarized to quantitatively analyze the predicted results of the remaining useful life,so as to realize the evaluation and improvement of the predicted results accuracy.Residual useful life prediction based on similarity,as a new data-driven method,can efficiently integrate a large number of historical data into the current degradation process,and has strong robustness and high accuracy.The method mainly includes the data processing part,the construction of the health indicators and similarity model,standardization of data processing,including processing and principal component analysis,the classification of health indicators to build health indicators and building health indicators,the method of determine similarity model including the time range,the choice of similarity measure,weight function and the remaining useful life prediction.The residual service life of turbofan engine was predicted by the similarity method,and the validity of the proposed method was verified.Using similarity model was carried out on the training data set,data processing,health indicators to build and similarity modeling,after remaining useful life prediction with test data set,and the different health indicators fitting model,similarity measure,life prediction of weighting function compares the performance,at the same time in the study joined the smooth measures to improve the prediction effect,Finally,the best prediction result of the similarity model is obtained.Finally,the paper reviews the work done in the whole paper,and summarizes and looks forward to the shortcomings in the research and the next research work.