Method of using machine learning to predict hard disk fault

Abstract

The present invention provides a method of using the machine learning to predict a hard disk fault, and belongs to the cloud storage safety technology field. The method of the present invention uses the mass smart data sets provided by the blackblaze, according to the condition that the smart data of the hard disks of different brands is distributed unevenly, uses a random forest algorithm to train and model the historical data, generates a fault prediction model, and enables the fault prediction rate to be improved.

Description

technical field [0001] The invention relates to cloud storage security technology, in particular to a method for predicting hard disk failures using machine learning. Background technique [0002] In recent years, cloud computing technology has developed rapidly, and the security and reliability of cloud computing have become the focus of attention of enterprises and individuals. To ensure that data is not lost, we must first pay attention to the security of cloud storage. Due to the huge number of disks in cloud storage, hard disks have the highest failure rate of server hardware in cloud environments. If hard disk failures can be predicted in advance, it will greatly benefit business experience and improve spare parts management. Self-Monitoring, Analysis and Reporting Technology (SMART) is one of the standard conditions that all disk manufacturers must follow in the ATA standard. It judges the health status of the disk by monitoring the status information of the motor, ...

AI Technical Summary

Problems solved by technology

[0007] Different hard disks are affected by factors such as manufacturers, environments, loads, and cumulative online time. The value of the same attribute may appear to be very different during the entire life span of the hard disk. If all data are normalized using the same parameters will seriously affect the predictive performance of the model

If different hard disks are trained and modeled separately, the process will be complicated, and the training data will be greatly reduced, resulting in underfitting

Images