A Sparsity-Free Compressed Sensing Method for PHM Data Quality Assurance Using Generative Adversarial Network

Recovering missing data to improve data quality is of great importance for prognostics and health management (PHM) of structures and machinery. Existing missing data reconstruction methods are generally limited by strong assumptions such as signal sparsity, coupled with heavy computational burdens, making them poorly adaptable to PHM data. To obtain more reliable PHM results, a novel algorithm is developed to handle missing data based on compressive sensing and enhanced context encoders (CS-ECE). The CS-ECE can reconstruct the detailed characteristics of signals in both the time and frequency domains. The effectiveness of the proposed method is validated using real-world data collected from a high-speed train. The results demonstrate that CS-ECE can obtain realistic and accurate recovered results, thereby providing a high-quality dataset for data-driven PHM studies.