Energy Reports (Nov 2022)
Energy saving and environment protection solution for the submarine pipelines based on deep learning technology
Abstract
Underwater flowlines (pipelines and cables) have become a popular option worldwide as communication lines. Correspondingly the demand for underwater trenching increased with the most desirable method internationally, jet trenching for its benefits in protecting the flowlines from external hazards. However, the trenching method is suffering from a main problem that can affect the construction and operation phases. This problem is the use of maximum jetting pressure needed for underwater trenching which leads to extra costs in the construction phase. Furthermore, the unstraightening of the trench can cause extra stress on the flowline requiring extra maintenance in the operation phase. The key cause of this problem is the classification of the underwater soil, as if the underwater soils cannot be classified as accurately as possible, the suitable jetting pressure required to trench these soils could not be applied correctly. Therefore, this research aims to develop a machine learning classifier for soil classification as a solution for enhancing the performance of the underwater trenching process. This classifier was constructed, trained, and evaluated by MATLAB R2020a to classify the underwater soil type by analyzing the soil image samples by an artificial neural network (ANN). Then a prototype that mimics exactly the real trenching mechanism was built to test the developed classifier based on three different types of soil (sand, clay & gravel). The results showed that the classifier successfully identify and classify the three different types of soil used. Finally, the results confirmed that the classifier will enhance the trenching processes. As it eliminates the need for soil classification by conventional methods while executing it in an easy fast process that will save energy, and cost, and decrease the need for future maintenance processes.
