Content area

Abstract

This project is dedicated to the development of a real-time anomaly detection and monitoring system tailored to optimize railway operations. At its core, the system harnesses the power of the Random Forest model, a flexible machine learning technique, to predict estimated arrival times on train routes. Based on historical arrival and departure data from stations, the system uses GPS coordinates to meticulously calculate train speeds, thus allowing accurate estimates of journey times.

The dataset under scrutiny comprises datetime records of train movements to and from stations, a rich source of information pivotal for computing medium speeds. Preparing this dataset for model training necessitates an application of preprocessing and feature engineering techniques, ensuring data readiness and enhancing predictive accuracy.

Through a meticulous process of experimentation and evaluation, the system's performance is rigorously scrutinized. Its efficacy in furnishing real-time insights into train operations and adeptly predicting delays emerges palpable. Despite grappling with challenges inherent to real-world data the system's robustness endures. Through continuous refinement and adaptation, the system proves itself solvent in optimizing train operations and mitigating delays, ensuring seamless efficiency in real-world scenarios

Alternate abstract:

Αυτό το έργο είναι αφιερωμένο στην ανάπτυξη ενός συστήματος παρακολούθησης και ανίχνευσης ανωμαλιών σε πραγματικό χρόνο, προσαρμοσμένο για βελτιστοποίηση σιδηροδρομικών λειτουργιών. Στον πυρήνα του, το σύστημα αξιοποιεί τη δύναμη του μοντέλου Random Forest, μιας ευέλικτης τεχνικής μηχανικής εκμάθησης, για την πρόβλεψη των εκτιμώμενων χρόνων άφιξης στα δρομολόγια των τρένων. Βασισμένο σε ιστορικά δεδομένα άφιξης και αναχώρησης που προέρχονται από σταθμούς, το σύστημα χρησιμοποιεί συντεταγμένες GPS για τον σχολαστικό υπολογισμό των ταχυτήτων του τρένου, επιτρέποντας έτσι ακριβείς εκτιμήσεις των χρόνων ταξιδιού.

Details

1010268
Title
Real-Time Train Tracking and Anomaly Detection System
Number of pages
67
Publication year
2024
Degree date
2024
School code
4463
Source
MAI 86/4(E), Masters Abstracts International
ISBN
9798342109772
Committee member
Τσιχριντζής, Γεώργιος (Tsichrintzis, Georgios); Σακκόπουλος, Ευάγγελος (Sakkopoulos, Evangelos)
University/institution
University of Piraeus (Greece)
University location
Greece
Degree
M.I.S.
Source type
Dissertation or Thesis
Language
English
Document type
Dissertation/Thesis
Dissertation/thesis number
31525098
ProQuest document ID
3122641462
Document URL
https://www.proquest.com/dissertations-theses/real-time-train-tracking-anomaly-detection-system/docview/3122641462/se-2?accountid=208611
Copyright
Database copyright ProQuest LLC; ProQuest does not claim copyright in the individual underlying works.
Database
ProQuest One Academic