Long Short-Term Memory Neural Network Applied to Train Dynamic Model and Speed Prediction

Zhen Li,Tao Tang,Chunhai Gao

doi:10.3390/a12080173

Abstract

The automatic train operation system is a significant component of the intelligent railway transportation. As a fundamental problem, the construction of the train dynamic model has been extensively researched using parametric approaches. The parametric based models may have poor performances due to unrealistic assumptions and changeable environments. In this paper, a long short-term memory network is carefully developed to build the train dynamic model in a nonparametric way. By optimizing the hyperparameters of the proposed model, more accurate outputs can be obtained with the same inputs of the parametric approaches. The proposed model was compared with two parametric methods using actual data. Experimental results suggest that the model performance is better than those of traditional models due to the strong learning ability. By exploring a detailed feature engineering process, the proposed long short-term memory network based algorithm was extended to predict train speed for multiple steps ahead.

Highlights

Railway transportation, an effective means to increase the efficiency of energy consumption and relieve the traffic congestion problem, has been stressed as an ideal transport mode in large cities [1]
To find an innovative way to deal with the above-mentioned task, different from the parametric approaches, we propose to apply deep learning algorithms to solve the problem of train dynamic model construction
Encouraged by the successful applications of long short-term memory (LSTM) based algorithms in the domain of transportation, we propose to employ LSTM networks for train dynamic model construction and train speed prediction, since the train operation process can be regarded as a time sequence problem

Summary

Introduction

An effective means to increase the efficiency of energy consumption and relieve the traffic congestion problem, has been stressed as an ideal transport mode in large cities [1]. To achieve safe and efficient operation, train control algorithms remain a key technical issue in the process of the development of railway systems [2]. With the help of signal devices, the train operation is accomplished by skilled drivers. This human-based train operation method lacks precise consideration, which may lead to a poor performance of energy consumption, service quality and safety [3]. Different from the manual labor based method, an automatic train operation system can provide a better operation performance by optimizing train control decisions. Many indicators that describe the running state of the train such as punctuality, riding comfort and energy efficiency can be remarkably improved by automatically adjusting the commands of train accelerating, coasting and braking process. The automatic train operation system mainly solves three arduous tasks: train dynamic model construction [7], speed profile optimization [8] and train speed control [9]

Methods

Results

Conclusion