Current Challenges with BIGDATA Analytics in Structural Health Monitoring

In SHM, fixed sensor networks with long-term monitoring capabilities, dense sensor arrays, or high sampling rates are perceived to produce BIGDATA. As the temporal and spatial resolution of monitoring data is improved by advances in sensing technology and with the adaptation of new data collection techniques, it is expected that efficient BIGDATA analysis strategies will become highly desirable. In addition to the massive quantity of data collected from these applications, the data's prospective heterogeneity poses a processing challenge. As capable sensing devices become more abundant and economical, it may be beneficial to integrate data collected by traditional means with emerging data types obtained by smartphones or image-based sensing systems. Previous studies have investigated the relationship between sensor network size and the corresponding information extracted by typical SHM methods. The scalability and computational sensitivity of these SHM processes in consideration of large SHM datasets have also been quantified. This paper intends to detail the current challenges posed by analyzing BIGDATA for SHM. This includes both the characteristics of BIGDATA sets produced by SHM and the expected processing challenges associated with these datasets. Novel approaches developed to overcome these challenges are reviewed and the continually evolving nature of BIGDATA is discussed.