A Hybrid LSTM-CPS Approach for Long-Term Prediction of Train Delays in Multivariate Time Series

Abstract

In many big cities, train delays are among the most complained-about events by the public. Although various models have been proposed for train delay prediction, prior studies on both primary and secondary train delay prediction are limited in number. Recent advances in deep learning approaches and increasing availability of various data sources has created new opportunities for more efficient and accurate train delay prediction. In this study, we propose a hybrid deep learning solution by integrating long short-term memory (LSTM) and Critical Point Search (CPS). LSTM deals with long-term prediction tasks of trains’ running time and dwell time, while CPS uses predicted values with a nominal timetable to identify primary and secondary delays based on the delay causes, run-time delay, and dwell time delay. To validate the model and analyse its performance, we compare the standard LSTM with the proposed hybrid model. The results demonstrate that new variants outperform the standard LSTM, based on predicting time steps of dwell time feature. The experiment results also showed many irregularities of historical trends, which draws attention for further research.

An Interpretable Station Delay Prediction Model Based on Graph Community Neural Network and Time-Series Fuzzy Decision Tree

“…Huang et al [3] used a three-dimensional (3-D) convolutional neural network to process spatial features and a long short-term memory recurrent neural network long short-term memory (LSTM) to process time series variables to predict station delays for four railways from China and The Netherlands, respectively. Wu et al [7] integrated the long short-term memory network LSTM and the key point search technology critical point search (CPS), and proposed a hybrid LSTM mode to process the time series variables of historical train motion data to predict the running time and dwell time of the train. Huang et al [6] modeled train delay data as time series data, using two long short-term memory neural networks LSTM to capture the interactions between stations and between trains, respectively.…”

Deep learning on SPY stock prices using improved multi-head LSTM

“…Gathering real charging data from shopping centers, residential, public, and workplace charging sites, and forecasts of the aggregated charging load being in 15-min resolutions, they found that that both models produced strong predictions, with the 4 time step model (which corresponds to predicting an hour ahead) outperforming the 96 time step model (which corresponds to predict a day ahead). InWu et al (2021), they used a long short term memory model combined with Critical Point Search to produce a forecast for train delays. This multi-step model showed significant improvement when compared to standard LSTM models at predicting train delays.…”