Development of natural rubber under-sleeper pads for enhancing railway transition zones using an integrated artificial neural network and genetic algorithm approach

Abstract

Natural rubber (NR) from Thailand demonstrates substantial potential for improving railway infrastructure through the development of under sleeper pads (USPs). This experimental study presented an optimized NR compound designed to enhance railway track performance, particularly in transition zones where tracks shift from bridges to ballasted sections. Various NR compound formulas were invented by adjusting filler content, with their properties evaluated through tensile, thermal aging, and tear tests. The optimal compound formula was proposed and employed in the production of NR-USPs, which were tested in compliance with the BS EN 16,730:2016 standard. Key factors influencing the performance of the NR-USPs, including hardness, M100, and thickness, were analyzed. To optimize these properties, an integrated Artificial Neural Network (ANN) and Genetic Algorithm (GA) approach was developed, achieving high predictive accuracy (R² = 0.99126). Field measurements of vertical displacement were conducted at Krung Thep Aphiwat Central Terminal Station, Bangkok, Thailand, focusing on the transition zone between the bridge and ballasted tracks. These measurements informed the specification of NR-USP thickness using the ANN-GA approach. The results demonstrated that the optimized NR-USP design effectively reduces vertical displacement in transition zones, thereby enhancing track stability and performance. This study provides a robust framework for future railway infrastructure advancements in Thailand, leveraging advanced optimization techniques to improve the durability and efficiency of railway tracks.