Rotating machinery remaining life prediction method based on FDA and SVDD

Abstract

The invention discloses a rotating machinery remaining life prediction method based on FDA and SVDD. The method is characterized by converting collected discrete data into a smooth function through a functional data analysis (FDA) method to solve the problems of information loss or model estimation distortion and the like, identifying machinery operation state through SVDD to reduce sample collection requirement and provide predication accuracy, and calculating radius of an SVDD model of a test sample and comparing the radius with a full life cycle time radius track to obtain a predication result. The method effectively reduces sample collection time required in predication of the remaining life of rotating machinery, and improves prediction accuracy.

Description

technical field [0001] The invention relates to a method for predicting the remaining life of a rotating machine, in particular to a method for predicting the remaining life of a rotating machine based on FDA and SVDD. Background technique [0002] Functional Data Analysis (FDA) method is a new branch of statistics. The concept of functional data was first proposed by Canadian scholar J.O. Ramsay, pointing out that modern sophisticated data collection systems can obtain a series of functional data, and observation data should be regarded as a dynamic concept, which can be solved by traditional data analysis methods. Research will bring problems such as loss of information or distortion of model estimation, which are mainly used in the fields of biomedicine and economics, and rarely involved in the field of mechanics. [0003] Support vector data description (Support vector data description, SVDD) was proposed by Tax and Duin, based on statistical learning theory, to establi...

AI Technical Summary

Problems solved by technology

The data-driven prediction method directly uses the relevant historical service information and online status monitoring information of the research object and the group as the data source, and extracts relevant features through a certain signal processing technology to predict the remaining life, and has been more and more researched. , but a large number of samples need to be collected and the accuracy of life prediction is limited

Images