In a significant stride for offshore engineering, a recent study published in the journal ‘Frontiers in Marine Science’ (translated from Chinese) has developed a novel approach to predict the strength of marine coral sand-clay mixtures (MCCM), a critical material used in various maritime construction projects. Led by Bowen Yang from Shanxi Ningguli New Materials Joint Stock Company Limited, the research leverages machine learning algorithms and triaxial shear tests to provide a more accurate and efficient way to assess the mechanical properties of MCCM.
Marine coral sand-clay mixtures are widely used as fill materials in offshore structures, such as breakwaters, coastal defenses, and subsea foundations. The strength of these mixtures is crucial for ensuring the stability and safety of these structures. However, predicting their strength has been challenging due to the complex interplay of various factors, including clay content, reinforcement layers, confining pressure, water content, and strain.
To tackle this issue, Yang and his team conducted a series of triaxial shear tests under different conditions to establish a comprehensive strength database for MCCM. Based on this dataset, they developed multiple predictive models, including Backpropagation Neural Network (BPNN), Genetic Algorithm optimized BPNN (GA-BPNN), Particle Swarm Optimization enhanced BPNN (PSO-BPNN), and a Logical Development Algorithm preprocessed BPNN model (LDA-BPNN).
Among these models, the LDA-BPNN model stood out, demonstrating superior accuracy and generalization capabilities compared to traditional optimization algorithms. “The LDA-BPNN model showed remarkable performance in predicting the strength of MCCM,” said Yang. “It not only improved the prediction accuracy but also enhanced the model’s ability to generalize to new, unseen data.”
The study also identified the primary factors influencing MCCM strength through sensitivity analysis. Water content, clay content, and confining pressure were found to be the most significant factors. This finding provides valuable insights for engineers and researchers working with MCCM.
One of the most practical outcomes of this research is the development of an explicit empirical formula derived from the LDA-BPNN model. This formula offers a straightforward and efficient tool for engineers without specialized machine learning expertise to predict the strength of MCCM. “We aimed to make our findings accessible and useful for practical applications,” explained Yang. “The empirical formula is a simple yet powerful tool that can be easily integrated into existing design and assessment processes.”
The commercial impacts and opportunities for the maritime sectors are substantial. Accurate and efficient strength prediction of MCCM can lead to optimized design and safety assessment of marine geotechnical engineering applications. This can result in cost savings, improved safety, and enhanced performance of offshore structures.
Moreover, the use of machine learning algorithms in this research opens up new avenues for the application of artificial intelligence in maritime engineering. As Yang noted, “Machine learning algorithms have immense potential in maritime engineering. They can help us tackle complex problems, improve our understanding of materials and structures, and ultimately lead to more innovative and sustainable solutions.”
In conclusion, this research by Bowen Yang and his team represents a significant advancement in the field of offshore engineering. By combining machine learning algorithms with traditional testing methods, they have developed a powerful tool for predicting the strength of MCCM. This tool not only enhances our understanding of these materials but also provides practical benefits for the maritime industry. As the field continues to evolve, the integration of machine learning and other advanced technologies is likely to play an increasingly important role in shaping the future of maritime engineering.