۱- Introduction

۲- Train delay prediction problem: the Italian case

۳- Train delay prediction systems

۴- Description of data and custom KPIs

۵- Results

۶- Conclusions

References

Abstract

Current train delay prediction systems do not take advantage of state-of-the-art tools and techniques for handling and extracting useful and actionable information from the large amount of historical train movements data collected by the railway information systems. Instead, they rely on static rules built by experts of the railway infrastructure based on classical univariate statistic. The purpose of this paper is to build a data-driven Train Delay Prediction System (TDPS) for large-scale railway networks which exploits the most recent big data technologies, learning algorithms, and statistical tools. In particular, we propose a fast learning algorithm for Shallow and Deep Extreme Learning Machines that fully exploits the recent in-memory large-scale data processing technologies for predicting train delays. Proposal has been compared with the current state-of-the-art TDPSs. Results on real world data coming from the Italian railway network show that our proposal is able to improve over the current state-of-the-art TDPSs.

Introduction

Big Data Analytics is one of the current trending research interests in the context of railway transportation systems. Indeed, many aspects of the railway world can greatly benefit from new technologies and methodologies able to collect, store, process, analyze and visualize large amounts of data [1,2] as well as new methodologies coming from machine learning, artificial intelligence, and computational intelligence to analyze that data in order to extract actionable information [3]. Examples are: condition based maintenance of railway assets [4–۶], automatic visual inspection systems [7,8], risk analysis [9], network capacity estimation [10], optimization for energy-efficient railway operations [11], marketing analysis for rail freight transportation [12], usage of ontologies and linked data in railways [13], big data for rail inspection systems [14], complex event processing over train data streams [15], fault diagnosis of vehicle on-board equipment for high speed railways [۱۶–۱۸] and for conventional ones [19], research on storage and retrieval of large amounts of data for high-speed trains [20], development of an online geospatial safety risk model for railway networks [21], train marshaling optimization through genetic algorithms [22], research on new technologies for the railway ticketing systems [23]. In particular, this paper focuses on building a Train Delay Prediction System (TDPS) in order to provide useful information to traffic management and dispatching processes through the usage of state-of-the-art tools and techniques, able to extract useful and actionable information from the large amount of historical train movements data collected by the railway information systems. Delays can be due to various causes: disruptions in the operations flow, accidents, malfunctioning or damaged equipment, construction work, repair work, and severe weather conditions like snow and ice, floods, and landslides, to name just a few. Although trains should respect a fixed schedule called Nominal Timetable (NT), Train Delays (TDs) occur daily and can negatively affect railway operations, causing service disruptions and losses in the worst cases. Rail Traffic Management Systems (TMSs) [24] have been developed to support the management of the inherent complexity of rail services and networks by providing an integrated and holistic view of operational performance, enabling high levels of rail operations efficiency.