Design of jetty piles using artificial neural networks

Area of Science:

• Civil Engineering
• Computational Mechanics
• Machine Learning Applications

Background:

• Jetty pile design is complex and computationally intensive.
• Traditional finite element (FE) analysis requires significant time and resources for parametric studies.

Purpose of the Study:

• To develop a more efficient method for jetty pile design analysis.
• To utilize artificial neural networks (ANN) to predict pile stress ratios accurately.

Main Methods:

• Employed multilayer perceptron neural networks (MBPNN) with two hidden layers.
• Generated training data using 50 finite element (FE) analyses for diverse design cases.
• Input features included lateral forces and pile types; output was the stress ratio.

Main Results:

• The trained ANN demonstrated high accuracy, closely matching FE analysis results.
• The MBPNN effectively predicted stress ratios for jetty piles under various conditions.

Conclusions:

• ANNs offer a viable and efficient alternative to extensive FE analysis for jetty pile design.
• This approach can significantly reduce design time and costs, especially for complex scenarios.