An Uncertainty-Aware Bayesian Deep Learning Method for Automatic Identification and Capacitance Estimation of Compensation Capacitors

This paper addresses the challenges of misclassification and reliability assessment in compensation capacitor detection under strong noise in high-speed railway track circuits. A hierarchical Bayesian deep learning framework is proposed, integrating multi-domain signal enhancement in the time, frequency, and time–frequency (TF) domains with bidirectional long short-term memory (BiLSTM) sequence modeling for robust feature extraction. Bayesian classification and regression based on Monte Carlo (MC) Dropout and stochastic weight averaging Gaussian (SWAG) enable posterior inference, confidence interval estimation, and uncertainty-aware prediction, while a rejection mechanism filters low-confidence outputs. Experiments on 8782 real-world segments from five railway lines show that the proposed method achieves 97.8% state-recognition accuracy, a mean absolute error of 0.084 μF, and an R2 of 0.96. It further outperforms threshold-based, convolutional neural network (CNN), and standard BiLSTM models in negative log-likelihood (NLL), expected calibration error (ECE), and overall calibration quality, approaching the theoretical 95% interval coverage. The framework substantially improves robustness, accuracy, and reliability, providing a viable solution for intelligent monitoring and safety assurance of compensation capacitors in track circuits.

• Publication year

  2026

• Venue

  Italian National Conference on Sensors

• Publication date

  2026-01-01

• Fields of study

  Medicine, Computer Science, Engineering

• Identifiers
  DOI 10.3390/s26010279PMID 41516714PMCID 12788300
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar, PubMed

• No claims are published for this paper.

• No concepts are published for this paper.

• A phase space and network-based approach for diagnosing compensation capacitor faults in Jointless Track Circuits

  2026cited by this paper
• A heterogeneous transfer learning method for fault prediction of railway track circuit

  2025cited by this paper
• A comprehensive review of deep learning-based fault diagnosis approaches for rolling bearings: Advancements and challenges

  2025cited by this paper
• Trustworthy Bayesian deep learning framework for uncertainty quantification and confidence calibration: Application in machinery fault diagnosis

  2024cited by this paper
• Uncertainty-Aware Fault Diagnosis Under Calibration

  2024cited by this paper
• A novel health state prediction approach based on artificial intelligence combination strategy for compensation capacitors in track circuit

  2024cited by this paper
• Probabilistic Uncertainty-Aware Decision Fusion of Neural Network for Bearing Fault Diagnosis

  2024cited by this paper
• Functional Safety and Performance Analysis of Autonomous Route Management for Autonomous Train Control System

  2024cited by this paper
• Uncertainty-Aware Deep Learning: A Promising Tool for Trustworthy Fault Diagnosis

  2024cited by this paper
• Multi-scale attention network (MSAN) for track circuits fault diagnosis

  2024cited by this paper
• Method for Fault Diagnosis of Track Circuits Based on a Time–Frequency Intelligent Network

  2024cited by this paper
• Track Circuits Fault Diagnosis Method Based on the UNet-LSTM Network (ULN)

  2024cited by this paper
• Uncertainty-aware deep learning for monitoring and fault diagnosis from synthetic data

  2024cited by this paper
• Uncertainty-aware deep learning for reliable health monitoring in safety-critical energy systems

  2024cited by this paper
• Evaluating deep learning models for fault diagnosis of a rotating machinery with epistemic and aleatoric uncertainty

  2024cited by this paper
• A Review on Deep Learning Techniques for Railway Infrastructure Monitoring

  2023cited by this paper
• Train-Centric Communication Based Autonomous Train Control System

  2023cited by this paper
• Fault prediction of track circuit compensation capacitor based on MFO-LSTM

  2023cited by this paper
• Towards trustworthy machine fault diagnosis: A probabilistic Bayesian deep learning framework

  2022influential reference
• A Probabilistic Bayesian Parallel Deep Learning Framework for Wind Turbine Bearing Fault Diagnosis

  2022cited by this paper
• Bayesian and machine learning-based fault detection and diagnostics for marine applications

  2022cited by this paper
• A Review of Deep Learning Applications for Railway Safety

  2022cited by this paper
• Research on Fault Location Method of Track Circuit Compensation Capacitor Based on Probabilistic Neural Network

  2022cited by this paper
• A literature review of Artificial Intelligence applications in railway systems

  2022cited by this paper
• Rail track condition monitoring: a review on deep learning approaches

  2021influential reference
• Assessments of epistemic uncertainty using Gaussian stochastic weight averaging for fluid-flow regression

  2021cited by this paper
• Fault Diagnosis of Compensation Capacitor in Joint-less Track Circuit via Long Short-Term Memory Network

  2021cited by this paper
• An Uncertainty-Informed Framework for Trustworthy Fault Diagnosis in Safety-Critical Applications

  2021influential reference
• Advanced Dropout: A Model-Free Methodology for Bayesian Dropout Optimization

  2020cited by this paper
• Intelligent fault diagnosis of rotating machinery using a new ensemble deep auto-encoder method

  2020cited by this paper
• A Simple Baseline for Bayesian Uncertainty in Deep Learning

  2019cited by this paper
• Intelligent rotating machinery fault diagnosis based on deep learning using data augmentation

  2018cited by this paper
• Railway Track Circuit Fault Diagnosis Using Recurrent Neural Networks

  2017cited by this paper
• Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning

  2015cited by this paper

• No citing papers are available for this paper.