Neural Network-Based Train Identification in Railway Switches and Crossings Using Accelerometer Data

Abstract

This paper aims to analyse possibilities of train type identification in railway switches and crossings (S&C) based on accelerometer data by using contemporary machine learning methods such as neural networks. That is a unique approach since trains have been only identified in a straight track. Accelerometer sensors placed around the S&C structure were the source of input data for subsequent models. Data from four S&C at different locations were considered and various neural network architectures evaluated. The research indicated the feasibility to identify trains in S&C using neural networks from accelerometer data. Models trained at one location are generally transferable to another location despite differences in geometrical parameters, substructure, and direction of passing trains. Other challenges include small dataset and speed variation of the trains that must be considered for accurate identification. Results are obtained using statistical bootstrapping and are presented in a form of confusion matrices.

Machine Learning-Based Node Characterization for Smart Grid Demand Response Flexibility Assessment

“…Commonly used machine learning methods for the classification of time-series data are Support Vector Machines (SVM) and Artificial Neural Networks (ANN), as recently reported in [30][31][32].…”

“…In the case of ANN, the subjective choice of its architecture can significantly influence both its performance as well as computational requirements. Given the relatively small size of the time series (96 data points), compared to applications that differ by two orders of magnitude, where highfrequency components have to be maintained, such as the dynamic response of railway track due to a passing train, which, if resampled to a lower resolution, looses its most important characteristics (see e.g., [30]), finding proper characterization for SVM input vector would make a little sense, since the entire time series vector can be directly processed by a more general ANN model. Some advanced time series classification techniques can be used such as Least-Squares Wavelet (LSWAVE) [33].…”

“…• CNN1 has the same architecture as the best performing model CNN in [30]. It contains one convolutional layer with 64 filters followed by max-pooling layer, fully connected hidden layer and an output layer; • CNN2 contains three convolutional layers with 128, 64 and 32 filters, respectively.…”

Multi-Layer Perceptron Neural Network Implementation as Train Type Classification

“…(18) The processing of immense quantities of data can be facilitated by machine learning techniques. Support vector machines (SVMs) (20) and Long Short Term Memory (LSTM) (21) have already been implemented for train detection monitoring in switches and crossing with 3 axis from accelerometer.…”

Condition Monitoring and Trend Analysis of Railway Turnouts Based on In-Situ Accelerometer Measurements

“…Known locomotive types included classes 150, 151, 162, 163, 350, 362 and 363, which were selected due to their mutual similarity to reduce the data variability. Machine learning models can identify concrete locomotive types provided sufficient training data [7]. Dataset II was obtained at a location in the UK around swing nose and was supplied by the University of Birmingham.…”