Page 33 - תקצירי תזות הנדסה ירוקה 2025

Studying the features of seismic reflections from the seabed to characterize its state in a non-destructive manner
Moshe Greenberg; moshegr1@ac.sce.ac.il
Dr. Uri Kushnir1, Prof. Vladimir Frid 1
1SCE - Shamoon College of Engineering, Be’er-Sheva, Israel.
Many engineering projects are carried out in a marine environment, including ports, desalination plants, gas and oil infrastructures, pipelines, and communication cables. A necessary condition for the planning of each marine infrastructure is marine soil investigation.
Most of the time, investigation of the sea floor is conducted by drilling and collecting physical samples, which requires the use of heavy engineering equipment, including drilling barges and large ships )with significant carbon emissions(.
This has a devastating effect on the marine environment, because loosed, floating sediment and carbon emissions may damage coral reefs and sensitive habitats.
The goal of this research is to develop a ‘green’, eco-friendly method for characterizing types of marine soils based on remote sensing.
The proposed method is based on acoustic signal processing and machine learning )ML(, which significantly reduce potential damage to marine ecosystems, enabling significant reduction in carbon emissions and environmental pollution, while providing a reliable source for the preliminary seabed surveys required before initiating marine engineering projects.
The desired research achievements are:
To produce a ‘green’ methodology that applies ML in combination with a sonar system for the purpose of classifying sea floor soils;
To examine the influence of soil relief, composition, voids, grain size distribution, and layers on reflected acoustic signals and to assess the influence of these factors on the implementation methodology.
To apply a non-destructive method to sub-bottom layers for the purpose of their characterization.
The results of this research demonstrate the feasibility of the new, tested classification method. Experiments were conducted on two soil types: SP )poorly-graded sand( and GP )poorly-graded gravel( and the outcomes showed clear differences between the soil types in terms of acoustic signal reflections and their spectral characteristics. These findings indicate that the method we developed, which combines signal processing and ML, is capable of effectively and reliably distinguishing between different soil types, suggesting that it holds real potential as a practical tool for future marine geotechnical surveys.
Keywords: ML, marine acoustics, marine soil classification, remote sensing, sensors.
33