Sensing Offshore Aquaculture Infrastructures for Data-Driven Dynamic Stress Analysis

Abstract

The presence of escaped fish in aquaculture facilities as a result of harsh meteorological conditions (more pressing in the face of climate change) requires a better understanding of this dynamic behaviour through vigilant monitoring and validated numerical models. In this context, data from strain and stress sensors as well as meteorological and current sensors installed at an aquaculture farm in the Region of Murcia (Spain) were collected, processed and analysed. Among them, the first results on the relationship between load and current sensors are presented. Due to the complexity of the time series, various analyses were conducted to examine their interrelation, encompassing the regression analysis of raw data and data segmented into different time intervals. Through this analysis, it was observed that employing distinct time windows better elucidated the data variability. Furthermore, an optimal data window of 240 data points was identified, demonstrating a significantly improved explanatory power, with the coefficient of determination (R2) increasing by approximately 0.8 depending on the section. This paves the way for optimising the monitoring features that must be carried out to relate cause-and-effect variables in the behaviour of these offshore infrastructures.