학술논문

An Online Detection Algorithm of Train Coupler Impact Based on Stacked Auto-Encoders
Document Type
Conference
Source
2021 CAA Symposium on Fault Detection, Supervision, and Safety for Technical Processes (SAFEPROCESS) Fault Detection, Supervision, and Safety for Technical Processes (SAFEPROCESS), 2021 CAA Symposium on. :1-6 Dec, 2021
Subject
Components, Circuits, Devices and Systems
Robotics and Control Systems
Signal Processing and Analysis
Deep learning
Heuristic algorithms
Signal processing algorithms
Couplers
Feature extraction
Stability analysis
Robustness
dynamic simulation
real-time detection
pattern recognition
stacked auto-encoders
Language
Abstract
In this paper, a high-precision multi-body dynamic model is first established to simulate the process of locomotive impacting the vehicles for marshalling. Then an online algorithm is designed based on semi-supervised deep learning to realize automatic coupler impact detection. The approach has the advantage of adaptive feature extraction with no need of complex signal processing technology. The robustness of the algorithm is enhanced through the fusion of multi-window information. Experimental results show that the algorithm is superior to the conventional methods in detection accuracy, stability and delay performance. The proposed algorithm provides an effective solution for online intelligent detection of coupler impact in the automatic operation of rail trains.