Abstract
Railway ballastless track is critical for maintaining track stability and extending service life. However, its effectiveness may be compromised by settlement, propagation, and scaling, thereby affecting the safe operation of trains. To address this challenge, this study developed an intelligent detection device named "SmartGrid" and proposed a machine learning method based on detection data to monitor equipment waveforms, thereby achieving intelligent ballastless track integrity detection. In machine learning, experimental data were divided into test and validation sets and fed into a CNN algorithm for learning and classification. Experimental results demonstrate that SmartGrid can accurately distinguish whether track turnouts are scaled and can be applied to large-scale detection of track turnout integrity in engineering applications.
Full Text
Preamble
Real-Time Detection of Railroad Ballast Integrity Using SmartGrid and CNN-Based Learning Algorithms
Meitao Zou¹, Kun Zeng¹,*, Qi Li¹
¹School of Civil Engineering, Tongji University, Shanghai 200092, China
Abstract
Ballasted track is essential for maintaining railway stability and extending service life. However, its effectiveness can be compromised by settlement, particle migration, and fouling, thereby jeopardizing train safety. To address this challenge, this study developed an intelligent detection device named "SmartGrid" and proposed a machine learning approach that analyzes detection data waveforms to enable automated ballast integrity assessment. The methodology employs a convolutional neural network (CNN) trained on experimental data partitioned into test and validation sets for classification tasks. Experimental results demonstrate that SmartGrid can accurately discriminate between clean and fouled ballast, validating its potential for large-scale ballast integrity monitoring in engineering applications.
Keywords: SmartGrid, railroad ballast, CNN, detection equipment