Hard disk failure prediction based on improved WOA-BP

In the era of rapid development of the Internet, how to ensure the Storage security of data has become an important issue. As the physical carrier of data, once the hard disk fails, it may lead to irreparable data loss. Therefore, hard disk failure prediction is very important for data storage security of large-scale storage cluster. This article focuses on the hard disk SMART data of the large cloud storage company Backblaze, using the WOA-BP model for model training, and optimizing the random initial value problem in the WOA algorithm using Cubic mapping method. The experimental results show that the improved WOA-BP model achieved a 96% accuracy rate in hard disk fault prediction.