Research On Life Prediction Of Equipment Under Multiple Operating Conditions Based On Similarity Trajectory Method

Mechanical equipment will constantly adjust operating conditions according to different production missions.In order to ensure the safety of equipment operation and the effectiveness of maintenance,it is of great significance to realize the life prediction of equipment under multiple operating conditions.Aiming at the problem of equipment life prediction under multiple operating conditions,although the casebased reasoning based similar trajectory method provides a feasible solution,there are still some problems,such as the similarity measurement accuracy between degradation indexes under single operating condition,the degradation model cannot be acquired accurately under multiple operating conditions,the inconsistency of mission profile between the prediction stage and the service stage of service equipments,and the unreliabile prediction results of life point by similar trajectory method.This paper aims at the above problems in the process of using similar trajectory method.On the basis of similar trajectory method,it realizes the accurate prediction of life under single operating condition by improving similarity measurement method.It realizes the life prediction under multiple operating condition by only using the monitoring data without extracting the degradation model by improving similarity measurement method.Through the decomposition of operating conditions and the use of similar trajectory method for the single operating condition,the life prediction of service equipments under the condition that the mission profile of the prediction stage and the service stage are inconsistent can be achieved.Finally,by constructing the life interval of historical data in advance,the life interval of service equipments can be predicted while using similar trajectory method to predict the life point.The main research contents are as follows:(1)The necessity of life prediction using similar trajectory method is discussed.Through the analysis of the influence of equipment working state and initial state on the life degradation curve,it can be seen that the equipment life degradation curve will be affected by many factors.Only one life degradation curve or prediction model is not suitable for most life prediction problems.Therefore,the similarity trajectory method which can make use of many degradation curves and informations in the historical data has obvious advantages in the prediction problems of multiple operating conditions and the degradation curves that cannot be accurately obtained.(2)Aiming at the problem of similarity measurement accuracy between degradation indexes under single operating condition,a new similarity measurement method based on image is proposed by using the mature algorithm in computational vision,and an improved similarity trajectory method based on image similarity is further proposed.The image-based similarity measurement method can measure the similarity of time series in the form of high dimension.Through the conversion of one-dimensional time series to two-dimensional image,not only the time sequence information between each time point in the time series can be effectively increased,but also the converted two-dimensional image can effectively reflect the shape information of the time series which is not affected by the translation or scaling of the time series.Then,the convolution neural network in computer vision is used to extract the features of the two-dimensional image and calculate the similarity.Finally,it is used in the similarity measurement of similar trajectory method,and the common test data of aero turbofan engine is used to verify that the improved similar trajectory method based on image similarity can effectively improve the life prediction accuracy under the single operating condition.(3)Aiming at the problem that the degradation model cannot be acquired accurately under multiple operating conditions,based on the characteristics of the trend component of some monitoring signals of mechanical equipment in the degradation process can be used to reflect the degradation trend while the detail component can be used to reflect the the environment and operating conditions,a similarity measurement method based on morphological pattern and symbolic aggregate approximation is proposed,and the morphological pattern and symbolic aggregate approximation based improved distance similarity trajectory method is further proposed.Empirical mode decomposition and multi-scale permutation entropy are used to decompose the monitoring signal into trend component and detail component,and morphological mode and symbolic aggregate approximation are used to encode and measure the similarity of different components.The similarity measurement method based on morphological pattern and symbolic aggregate approximation can effectively measure the similarity of time series with both trend component and detail component.Finally,it is used in the similarity measurement of similar trajectory method,and the common test data of aero turbofan engine is used to verify that the method can directly use the monitoring signal to predict the life of the equipment under multiple operating conditions without extracting the degradation model of the monitoring signal.(4)Aiming at the problem of inconsistency of mission profile between the prediction stage and the service stage of service equipments,a life prediction method based on similar trajectory is proposed for the given mission profile.The historical database is constructed by decomposing the multiple operating conditions and predicting the life of each operating conditions.The similar trajectory method is used to predict the life of the each operating condition of service equipments.Finally,according to the proportion of each operation condition in the given mission profile,the corresponding predicted life of the equipment under the given mission profile is obtained.The common test data of aero turbofan engine and the assumed mission profile are used to verify that this method can effectively solve the life prediction problem that the mission profile of the prediction stage is inconsistent with that of the service stage.(5)Aiming at the problem of unreliabile prediction results of life point by similar trajectory method,a life interval prediction method based on interval estimation and similar trajectory method is proposed.Firstly,the upper and lower limits of the life of historical data are obtained by estimating the interval of historical data.Then the life interval of service equipments can be predicted while using similar trajectory method to predict the life point.The common test data of aero turbofan engine are used to verify that the method can give the life point prediction and the corresponding life prediction interval in different prediction problems.