Research on the numerical control system reliability model using censored data

Abstract

In the analysis of data for the reliability in CNC system, people commonly do not consider censored time. With the continuous improvement of reliability in CNC system, the failure rate is usually low in the timing censored test. All these produce a certain error in data analysis and parameter estimation. To solve this problem, this paper proposed to use censored data in the data analysis. First pre-process data of failure interval time, followed by the application of the cumulative failure rate to determine the distribution types primarily, then the maximum likelihood method is used to estimate the distributed parameters followed by fitting test with Hollander-Proschan method to get the distribution model. In this paper, failure data from 13 of a certain type numerical control system in four-month test were performed for processing and proved this method could improve precision in data analysis of reliability in CNC system.