[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-122506-en":3,"doc-seo-122506-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},122506,687197207919,"Theodora","https://ap-avatar.wpscdn.com/avatar/a000253d6f5f7c60be?x-image-process=image/resize,m_fixed,w_180,h_180&k=1779446848396160552",8,"Research & Report","Machine learning methods to predict ballast breakage in railway tracks","Ballast breakage can be quantified through the Ballast Breakage Index (BBI), yet accurately predicting BBI under repetitive loading remains difficult. This study develops BBI predictive models using machine learning, including artificial neural networks (ANN), support vector machines (SVM), and random forest (RF). A comprehensive 20-year dataset is used with inputs covering loading characteristics, particle size distribution, angularity, initial physical state, and test-apparatus stress state. Model performance is evaluated using RMSE, MAPE, and R², with ANN showing marginally superior accuracy. Results indicate ANN can capture nonlinear relationships and enable reliable breakage predictions without extensive laboratory testing, supporting field-oriented ballast performance assessment.","GeoVadis: The Future of Geotechnical Engineering – Juneja, Joseph & Dasaka (Eds)  \n© 2025 The Author(s), ISBN: 978-1-041-08547-8  \nMachine learning methods to predict ballast breakage in railway tracks  \nS. Alagesan, B. Indraratna, R. S. Malisetty and Y. Qi  \nSchool of Civil and Environmental Engineering, University of Technology Sydney, Ultimo, Australia  \nABSTRACT: Ballast breakage can be estimated using ballast breakage index (BBI), is oneof the critical parameters for assessing the performance of ballast, yet accurately predicting BBI remains challenging. In this study predictive models for BBI under repetitive loading were developed using machine learning techniques such as artificial neural networks (ANN), support vector machines (SVM) and random forest (RF) . This study utilizes comprehensive 20-year dataset, incorporating input parameters such as loading characteristics, particle size distribution, angularity, initial physical state, and the stress state applied by the test apparatus. While examining the performance based on RMSE, MAPE and R2, ANN model marginally outperformed SVM and RF models. The results demonstrate the capability of ANN in elucidating nonlinear relationships, providing accurate breakage predictions without requiring extensive laboratory testing. This study highlights the potential of advanced soft-computing tools for assessing ballast performance in railway tracks, bridging the gap between laboratory insights and field applications.  \n1 INTRODUCTION  \nThe rapid advancement of railway infrastructure necessitates efficient maintenance planning to ensure safe and reliable train operations. Ballast, the uppermost granular layer in railway tracks, distributes vertical stresses from train loads to the subgrade. However, repeated loading leads to particle breakage, reducing internal friction and increasing fines, which impedes drainage and causes permanent deformation. Over time, ballast fouling exacerbates track instability, increasing maintenance demands. In Australia, frequent ballast renewal is required to maintain track stability (Nishiura et al. 2018; Indraratna et al. 2020) and Ballast Breakage Index (BBI) is widely used to quantify ballast degradation. Extensive research has investigated the evolution of BBI through large-scale laboratory tests, considering cyclic loads, confining pressure, and ballast gradation (Lackenby et al., 2007; Anderson & Fair, 2008) . However, direct field measurement is laborious and impractical, while empirical models, though useful in controlled conditions, lack generalizability due to variations in stress conditions and loading histories. Numerical models like DEM provide valuable insights into particle-scale mechanisms but are computationally expensive (O’Sullivan et al., 2008; McDowell & Li, 2016) . Similarly, Constitutive models, though effective (Salim & Indraratna, 2004), require extensive experimental data and significant domain expertise. Given these challenges, machine learning (ML) has emerged as a promising alternative for predicting complex, nonlinear geotechnical behaviour. ML models have been successfully applied to soil classification, settlement prediction, and slope stability analysis (Penumadu & Zhao 1999, Shahin & Indraratna, 2006; Armaghani et al., 2014) . However, limited studies have focused on railway ballast, mainly addressing resilient modulus and breakage due to impact loading (Indraratna et al., 2023; Koohmishi & Guo, 2023) . This study employs Artificial Neural Networks (ANN), Support Vector Mechanisms (SVM), and Random Forests (RF) to predict BBI under train loading  \nDOI: 10. 1201/9781003645917-21  \nThis chapter has been made available under a CC-BY-NC-ND license  \nwhile emphasizing the integration of domain knowledge to ensure meaningful ML predictions beyond traditional regression-based approaches.  \n2 GEOTECHNICAL INPUT SELECTION  \nThe selection of input parameters is crucial for the development of an accurate and generalizable machine ","cbCaioneqBZ83WIh","https://ap.wps.com/l/cbCaioneqBZ83WIh","pdf",3659640,1,6,"English","en",105,"# Abstract\n# Introduction\n# Geotechnical input selection","[{\"question\":\"Why is predicting the Ballast Breakage Index (BBI) challenging?\",\"answer\":\"BBI evolves under repeated loading in complex, nonlinear ways, and field measurement is laborious. Traditional empirical or numerical approaches can be limited by generalizability or computational and data requirements.\"},{\"question\":\"Which machine learning models are used to predict ballast breakage?\",\"answer\":\"The study builds predictive models using artificial neural networks (ANN), support vector machines (SVM), and random forest (RF).\"},{\"question\":\"What input parameters are selected for the machine learning models?\",\"answer\":\"Inputs include confining pressure, number of load cycles, cyclic deviatoric stress, loading frequency, median particle size (D50), uniformity coefficient (Cu), dry unit weight, and peak friction angle, chosen to represent both material properties and stress conditions.\"}]","Machine learning methods to predict ballast breakage in railway tracks | PDF",1785810995,15,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"machine-learning-methods-to-predict-ballast-breakage-in-railway-tracks","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/machine-learning-methods-to-predict-ballast-breakage-in-railway-tracks/122506/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-04",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why is predicting the Ballast Breakage Index (BBI) challenging?","Question",{"text":75,"@type":76},"BBI evolves under repeated loading in complex, nonlinear ways, and field measurement is laborious. Traditional empirical or numerical approaches can be limited by generalizability or computational and data requirements.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"Which machine learning models are used to predict ballast breakage?",{"text":80,"@type":76},"The study builds predictive models using artificial neural networks (ANN), support vector machines (SVM), and random forest (RF).",{"name":82,"@type":73,"acceptedAnswer":83},"What input parameters are selected for the machine learning models?",{"text":84,"@type":76},"Inputs include confining pressure, number of load cycles, cyclic deviatoric stress, loading frequency, median particle size (D50), uniformity coefficient (Cu), dry unit weight, and peak friction angle, chosen to represent both material properties and stress conditions.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,114,119,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":21,"doc_module":4,"doc_module_name":46,"category_name":111,"show_sort_weight":112,"slug":113},"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":46,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]