A data-driven algorithm for detecting anomalies in underwater sensor-based wave height measurements

This paper proposes a data-driven approach for detecting anomalies in wave height measurements caused by faults in an acoustic wave and current profiler type sensor operating in an underwater environment. According to the theory of sea state and the theory of quasi-determinism of sea waves group, four statistical indicators were defined to measure the reliability of the acquired measurements. Based on these statistical indicators, a principal component analysis was performed. The resulting statistical model was used to detect sensor faults using a threshold system based on Hotelling’s T
²
and Q-statistic scores. The ability of the proposed method to detect measurement anomalies caused by sensor faults was tested using a collection of numerically corrupted experimental data. In particular, three types of measurement alterations related to three real sensor faults were considered.