Hard disk fault early warning and reliability evaluation method based on particle filtering

Abstract

The invention discloses a hard disk fault early warning and reliability evaluation method based on particle filtering. The method comprises the following steps that: through hard disk S.M.A.R.T (Self-Monitoring Analysis and Reporting Technology) index screening and secondary index construction, combining with a multisource information fusion method of multiple regression analysis to realize a purpose that high-dimensional original data is converted into a one-dimensional hard disk state degeneration variable; applying a particle filtering algorithm based on Bayesian estimation to carrying outrecursive estimation to obtain a hard disk state sequence, and making a difference with a practical hard disk state observation sequence, and obtaining a hard disk system residual error; and accordingto the statistical characteristic of the system residual error, designing an adaptive fault judgment threshold value, and combining with a multi-point continuous alarming mechanism to implement a hard disk fault early warning so as to carry out statistics on a hard disk reliability evaluation relevant index. By use of the method, the characteristic information of multiple dimensions of the hard disk can be combined to fully utilize all useful information, an online fault early warning based on hard disk state monitoring is realized, a high-level fault detection rate is obtained under a situation that a zero false alarm is guaranteed, and meanwhile, reliability evaluation index fault early warning lead time with better indication performance and guiding significance is output.

Description

[0001] The invention belongs to the field of hard disk fault diagnosis and reliability engineering, in particular to a particle filter-based hard disk fault early warning and reliability evaluation method. Background technique [0002] In today's era that is highly dependent on computers and the Internet, governments, enterprises and individuals have achieved paperless offices, and a large amount of important data and materials are stored in digital form. According to statistics, about 90% of the newly generated data in the world are stored in magnetic media, and most of them are stored in hard disks. With the increasing amount of data and people's increasing demand for high-speed and stable cloud services, data security issues have become increasingly prominent. It is not only related to personal interests and business operations, but may even affect national security. Therefore, it has always been a research hotspot in the storage field to carry out fault warning and reliabi...

AI Technical Summary

Problems solved by technology

For more complex systems such as hard disks, degradation data often appear as random and non-linear time series. Trying to use particle filter method for fault diagnosis is an effective solution to make up for the shortcomings of statistics and machine learning methods.

Images