A Novel Approach to Railway Track Faults Detection Using Acoustic Analysis

Abstract

Regular inspection of railway track health is crucial for maintaining safe and reliable train operations. Factors, such as cracks, ballast issues, rail discontinuity, loose nuts and bolts, burnt wheels, superelevation, and misalignment developed on the rails due to non-maintenance, pre-emptive investigations and delayed detection, pose a grave danger and threats to the safe operation of rail transport. The traditional procedure of manually inspecting the rail track using a railway cart is both inefficient and prone to human error and biases. In a country like Pakistan where train accidents have taken many lives, it is not unusual to automate such approaches to avoid such accidents and save countless lives. This study aims at enhancing the traditional railway cart system to address these issues by introducing an automatic railway track fault detection system using acoustic analysis. In this regard, this study makes two important contributions: data collection on Pakistan railway tracks using acoustic signals and the application of various classification techniques to the collected data. Initially, three types of tracks are considered, including normal track, wheel burnt and superelevation, due to their common occurrence. Several well-known machine learning algorithms are applied such as support vector machines, logistic regression, random forest and decision tree classifier, in addition to deep learning models like multilayer perceptron and convolutional neural networks. Results suggest that acoustic data can help determine the track faults successfully. Results indicate that the best results are obtained by RF and DT with an accuracy of 97%.

Keywords: acoustic signals analysis; deep convolution neural networks; logistic regression; machine learning; railway track cracks detection; railway track inspection.

Figure 1

Samples of faults on railway tracks, (a) Destroyed patch [7]; (b) Partial crack [8]. Such faults can happen due to excessive loads, and the influence of cold and hot weather.

Figure 2

Wheel burn and ballast issue on railway track, (a) Wheel burnt issue on railway track [9]; (b) Week and expired ballast issue [10].

Figure 3

Surface and nuts and bolts problems of a railway track: (a) Damage of surface of rail head due to super elevation issue; (b) Absence of nuts and bolts [11].

Figure 4

Mechanical railway cart used for data collection. The cart is driven by the engine that is manually controlled.

Figure 5

Pictures of wheels in contact with the track. (a) Assembly of microphone on left side of the mechanical cart; (b) Assembly of microphone on right side of the mechanical cart.

Figure 6

ECM-X7BMP type microphone [28].

Figure 7

Architecture of the proposed methodology for faulty track detection comprising data collection, MFCC feature extraction, and training and testing the models.

Figure 8

Five steps to extracting MFCC features [33]. It shows the steps followed to extract MFCC features that are used for the training and testing of the machine learning models.

Figure 9

Normal, super elevation and wheel burn signals in time domain and MFCC. Mel-spectrogram shows clear difference in the signals for different faults.

Figure 10

Architecture of MLP model designed for this study. Dense refers to a fully connected layer, activation is the activation function used while the dropout layer shows the neural dropout ratio used for optimization.

Figure 11

Architecture of CNN model used for experiments in this study. Conv2D shows the 2D convolutional layer with kernel size of 2 and max pooling layer with a pool size of 2. Dropout rate for neuron drop is 0.2 indicating 20% drop for optimization.

Figure 12

Classification accuracy using different train-test splits.