Bayesian Network and Croston Analysis for Axle Counter Maintenance at DAOP 7 Madiun

Malfunctions in railway signaling systems are critical issues affecting the safety and operational efficiency of train services. A key component is the axle counter, which detects train presence by counting its axles. This study aims to identify the dominant factors and causes of axle counter malfunctions and predict future malfunction frequencies to support predictive maintenance. Data were obtained from signaling equipment malfunction logs maintained by the Signaling and Telecommunications (Sintelis) Unit of PT Kereta Api Indonesia (Persero) DAOP 7 Madiun January 2024–June 2026 and expert judgments from Sintelis technicians and the Assistant Manager. Malfunction factors and causes were analyzed using a Bayesian Network based on a Directed Acyclic Graph (DAG), Conditional Probability Tables (CPT), Joint Probability Distributions (JPD), and posterior probabilities derived from Bayes' Theorem. Components with the highest malfunction frequencies were then forecast using Croston's Method to predict intermittent failures. The results show that the Head component is mainly affected by external factors, particularly metal objects (59.26%), whereas the Wheel Detection Equipment (WDE) and Evaluator Module are predominantly influenced by internal deterioration, with probabilities of 52.70% and 40.84%, respectively. Forecasting produced values of 0.39 for the Head, 0.55 for the WDE, and 0.51 for the Evaluator Module, indicating that the WDE has the highest likelihood of malfunction in the upcoming period. The integration of the Bayesian Network and Croston's Method supports predictive maintenance by prioritizing maintenance activities and improving component planning to enhance signaling system reliability.

Keywords: Axle Counter, Bayesian Networks, Croston's Method, Predictive Maintenance, Railway Signaling

Copyright (c) 2026 Journal of Dinda : Data Science, Information Technology, and Data Analytics

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

Hani'a Tsabita Fajriah Kansa Universitas Negeri Surabaya

Undergraduate Student, Department of Data Science, Universitas Negeri Surabaya

Yuliani Puji Astuti Universitas Negeri Surabaya

Department of Data Science, Universitas Negeri Surabaya. Rank: Lecturer and Coordinator of the Data Science Study Program.