A Data Compression Method with an Encryption Feature for Safe and Lightweight Vibration Condition Monitoring
Yuhua Yin,Zhiliang Liu,Qiang Zhang,Yong Qin,Mingjian Zuo
DOI: https://doi.org/10.1109/jiot.2024.3412675
IF: 10.6
2024-01-01
IEEE Internet of Things Journal
Abstract:Vibration data compression is crucial for addressing the considerable data volume challenge in prognostics and health management (PHM). This challenge can be mitigated by compressing both the number of sample points and the size of individual sample points. However, achieving high-compression ratios (CRs) encounters two primary challenges. First, current optimal solutions for compressing sample point sizes, data binarization, suffer from low-compression efficiency. Second, in hybrid compression, the compression effects of individual sample point sizes are prone to being lost during the reduction of sample points, thus limiting the improvement of CRs. To address these challenges, a novel hybrid compression framework is introduced for vibration condition monitoring. Building upon this framework, an efficient compression method with encryption features is proposed. The main contributions of the proposed method are twofold. First, by introducing the concept of clustering-based binarization, compression of sample point sizes with high CRs is achieved while improving compression efficiency. Second, by designing compression sampling methods that preserve the original data properties, the failure of individual sample point size compression is prevented, and compression space is further expanded while enhancing data security. Experimental results demonstrate the overall superiority of the proposed method. Compared to existing approaches, it achieves significant improvements in CR while retaining key spectral information, enhancing compression efficiency, and ensuring better data security. Thus, it alleviates the challenges of the significant data volume posed to data storage, transmission, and processing in PHM.