Research On Functional Data Analysis And Its Fault Diagnosis In Gearbox And Remaining Life Prediction In Engine

The most important thing to analyze the equipment is the analysis of equipment failure and the modeling of performance degradation.In this paper,functional data analysis is used as a research method to conduct the above research on equipment.Functional data analysis is to treat the original data as a whole,and then analyze the data from the perspective of the whole.It has the characteristics of external condition dependence,the ability to analyze data of different dimensions and the in-depth analysis of data from different angles,which makes up for the deficiencies in traditional data analysis,which reflects its good research value.The condition monitoring of equipment firstly analyzes the original signal,but the current signal analysis only analyzes the original signal on the basis of the original signal.The discrete and discontinuous characteristics of the original signal cause certain difficulties in data analysis.Therefore,in combination with the idea of functional data analysis,this paper proposes some new methods for the condition monitoring of equipment,specifically as follows:(1)A fault diagnosis method based on functional principal component analysis and kernel limit learning machine is proposed.In view of the low accuracy and robustness in the current data diagnosis,the functional data analysis has the characteristics of high robustness.In this paper,firstly,the functional principal component analysis is used to extract the principal component of the data,and then the kernel limit learning machine is used as a classifier to classify the data and identify the fault category.Experiments show that this method is better than the traditional method.(2)A degradation modeling method for basis function data analysis is proposed.After analyzing the performance degradation of the equipment,the function data is used to analyze the integrated characteristics,and all data are fitted with functions,and the functional principal component analysis is used to extract their degradation features respectively.The type degradation model is finally obtained based on experiments,and the established model helps to improve the final accuracy.(3)A new similarity life prediction method is proposed.Life for traditional similarity method error problem of the heterogeneous information fusion,this paper first to life prediction of all the degradation index,get the degradation index of residual life,and then through the Pearson correlation coefficient of each index correlation calculation,finally all degradation index according to the correlation of residual life weighted fusion,finally it is concluded that remaining service life of the equipment.Experiments show that the proposed method has high accuracy.