Automatic Train Research Articles

Effects of expertise for automatic train operations

The aim was to investigate the effects of automatic speed control and expertise on train driver performance in unexpected, degraded operations in the railway domain. Research from other domains suggested increasing levels of automation to exert detrimental effects on human performance in degraded operations. In addition, research about the effect of expertise on performance in different levels of railway automation was scarce. Reaction times of 26 train drivers to critical events in a high-fidelity railway simulator featuring manual and automatic speed control were assessed. Participants showed longer reaction times in degraded operations under automatic speed control. Results on the influence of expertise were insignificant and methodologically discussed to enable further research. These results highlight performance consequences of automatic speed control in railway operations and offer first insights on how existing expertise needs to be taken into account when introducing automatic speed control into the railway domain.

RAILTOPOMODEL AND RAILML 3 IN OVERALL CONTEXT

This paper aims to provide a brief insight into the UIC RailTopomodel and railML initiative. Concerning the railML exchange format, mainly the current form of forthcoming third edition, involving especially the infrastructure schema, based on the RailTopoModel modelling principles, is taken into account. When rewarding, the comparison between railML 3 and the previous railML 2.3 version is given. At the end, the author focuses on selected issues of possible applicability of these tools for the needs of such systems as the automatic train control, automatic train operation, prediction of train etc. If essential, some extensions of the structures are proposed.

CANT EXCESS AND CANT DEFICIENCY EVALUATION IN RELATION TO TRAIN SPEED

The paper deals with the evaluation of measured speeds of the vehicles on the chosen sections of the railway network in the Czech Republic. In connection with the phased implementation of the AVV system (automatic train operation) and the European ETCS system, the dispersion of the train speeds will be reduced. Measured data were used to assess the weighted average of the cant excess or cant deficiency in relation to the rail defects. Within the framework of the research, the speeds were statistically evaluated covering more aspects like types and sorts of the trains, their operators, daytime and more.

Automatic large-scale data acquisition via crowdsourcing for crosswalk classification: A deep learning approach

Correctly identifying crosswalks is an essential task for the driving activity and mobility autonomy. Many crosswalk classification, detection and localization systems have been proposed in the literature over the years. These systems use different perspectives to tackle the crosswalk classification problem: satellite imagery, cockpit view (from the top of a car or behind the windshield), and pedestrian perspective. Most of the works in the literature are designed and evaluated using small and local datasets, i.e. datasets that present low diversity. Scaling to large datasets imposes a challenge for the annotation procedure. Moreover, there is still need for cross-database experiments in the literature because it is usually hard to collect the data in the same place and conditions of the final application. In this paper, we present a crosswalk classification system based on deep learning. For that, crowdsourcing platforms, such as OpenStreetMap and Google Street View, are exploited to enable automatic training via automatic acquisition and annotation of a large-scale database. Additionally, this work proposes a comparison study of models trained using fully-automatic data acquisition and annotation against models that were partially annotated. Cross-database experiments were also included in the experimentation to show that the proposed methods enable use with real world applications. Our results show that the model trained on the fully-automatic database achieved high overall accuracy (94.12%), and that a statistically significant improvement (to 96.30%) can be achieved by manually annotating a specific part of the database. Finally, the results of the cross-database experiments show that both models are robust to the many variations of image and scenarios, presenting a consistent behavior.

Designing train-speed trajectory with energy efficiency and service quality

ABSTRACTWith the development of automatic train operations, optimal trajectory design is significant to the performance of train operations in railway transportation systems. Considering energy efficiency and service quality, this article formulates a bi-objective train-speed trajectory optimization model to minimize simultaneously the energy consumption and travel time in an inter-station section. This article is distinct from previous studies in that more sophisticated train driving strategies characterized by the acceleration/deceleration gear, the cruising speed, and the speed-shift site are specifically considered. For obtaining an optimal train-speed trajectory which has equal satisfactory degree on both objectives, a fuzzy linear programming approach is applied to reformulate the objectives. In addition, a genetic algorithm is developed to solve the proposed train-speed trajectory optimization problem. Finally, a series of numerical experiments based on a real-world instance of Beijing–Tianjin Intercity Railway are implemented to illustrate the practicability of the proposed model as well as the effectiveness of the solution methodology.

Particle swarm optimization for generating interpretable fuzzy reinforcement learning policies

Fuzzy controllers are efficient and interpretable system controllers for continuous state and action spaces. To date, such controllers have been constructed manually or trained automatically either using expert-generated problem-specific cost functions or incorporating detailed knowledge about the optimal control strategy. Both requirements for automatic training processes are not found in most real-world reinforcement learning (RL) problems. In such applications, online learning is often prohibited for safety reasons because it requires exploration of the problem’s dynamics during policy training. We introduce a fuzzy particle swarm reinforcement learning (FPSRL) approach that can construct fuzzy RL policies solely by training parameters on world models that simulate real system dynamics. These world models are created by employing an autonomous machine learning technique that uses previously generated transition samples of a real system. To the best of our knowledge, this approach is the first to relate self-organizing fuzzy controllers to model-based batch RL. FPSRL is intended to solve problems in domains where online learning is prohibited, system dynamics are relatively easy to model from previously generated default policy transition samples, and it is expected that a relatively easily interpretable control policy exists. The efficiency of the proposed approach with problems from such domains is demonstrated using three standard RL benchmarks, i.e., mountain car, cart-pole balancing, and cart-pole swing-up. Our experimental results demonstrate high-performing, interpretable fuzzy policies.

Learning Aerial Image Segmentation From Online Maps

This study deals with semantic segmentation of high-resolution (aerial) images where a semantic class label is assigned to each pixel via supervised classification as a basis for automatic map generation. Recently, deep convolutional neural networks (CNNs) have shown impressive performance and have quickly become the de-facto standard for semantic segmentation, with the added benefit that task-specific feature design is no longer necessary. However, a major downside of deep learning methods is that they are extremely data-hungry, thus aggravating the perennial bottleneck of supervised classification, to obtain enough annotated training data. On the other hand, it has been observed that they are rather robust against noise in the training labels. This opens up the intriguing possibility to avoid annotating huge amounts of training data, and instead train the classifier from existing legacy data or crowd-sourced maps which can exhibit high levels of noise. The question addressed in this paper is: can training with large-scale, publicly available labels replace a substantial part of the manual labeling effort and still achieve sufficient performance? Such data will inevitably contain a significant portion of errors, but in return virtually unlimited quantities of it are available in larger parts of the world. We adapt a state-of-the-art CNN architecture for semantic segmentation of buildings and roads in aerial images, and compare its performance when using different training data sets, ranging from manually labeled, pixel-accurate ground truth of the same city to automatic training data derived from OpenStreetMap data from distant locations. We report our results that indicate that satisfying performance can be obtained with significantly less manual annotation effort, by exploiting noisy large-scale training data.

Model Predictive Control for Real-time Position Tracking of a Catenary-free Tram

Automatic Train (or Tram) Operation (ATO) systems have been widely studied in the last decade to guarantee energy-efficient operation. Speed profiles and timetables are derived by applying optimization methods geared towards finding the optimal trade-off between journey time, emissions and energy efficiency. An on-board controller is required to track such optimal trajectories, despite model uncertainties. Due to the non-linear dynamics and non-holonomic constraints on both inputs and outputs, the tracking of such trajectories represents a difficult task. This paper deals with the design of the tracking algorithm, for a catenary-free tram, with regenerative braking. A Model Predictive Control (MPC) approach is used to track the optimal position profile. Thanks to the ability to easily deal with multivariable systems, make use of the references and disturbances preview, and enforce time-varying input/output constraints, MPC is the perfect candidate to fulfill this task. The proposed MPC formulation is simple enough to be implemented in real-time, with a quadratic programming solver. Results show that the algorithm tracks the position reference, and guarantees punctuality and comfort, while enforcing time-varying constraints on vehicle velocity, electrical machine and friction forces. The results are also confirmed in the case of operation with model mismatch.

A robust and fault tolerant approach for automatic train stop control system design

This paper is about an advanced braking control strategy for rail transportation systems. The motivation is to improve the efficiency of the control in real-world conditions taking into account modeling errors, unknown perturbations and occurring failures. This will allow traffic frequency improvement along with overall safety and security enhancement. The strategy is based on an integrated approach, including a fault tolerant control strategy along with perturbation estimation and compensation, and a fault detection module. Fault detection, if positive, will be used to switch the controller into a “safe mode” avoiding the fault and improving control precision with respect to stopping position requirements. Simulations are provided to prove the efficiency of the proposed approach.

Propagation modeling for outdoor-to-indoor and indoor-to-indoor wireless links in high-speed train

Nowadays telecommunication companies have shown a great interest in deploying broadband mobile wireless networks in high-speed trains (HSTs) with the aim of supporting both passenger services provisioning as well as automatic train control and signaling. The train carriage, as a confined steel structure, has specific propagation characteristics, which motivates the study of the indoor-to-indoor and outdoor-to-indoor radio propagation characteristics for broadband wireless communication systems in high-speed railways, constituting the main contribution of this work. This study has been performed by means of measurements considering an actual Long Term Evolution (LTE) network deployment, as well as a portable test transmitter and different configurations of antennas and receivers at 2.4, 2.6 and 5.7GHz in a commercial high-speed rail line in Spain. The results show that radio waves incur obvious waveguiding effects inside the HST car. Moreover, for the propagation from the railway station to a mobile receiver inside the HST car, waves at higher frequencies experience less attenuation through the train carriage, by better propagating through windows. Although the railway station and train interior contain objects that induce a rich set of multipath components, the analysis of small-scale fading statistics shows that the channel still has a dominant path. Also, the LTE coverage tests for Base Transceiver Station (BTS)-Train and BTS-Mobile links were conducted and with internal and external antennas on board the train. We found that there was a strong signal penetration loss of approximately 26dB caused by the train carriage structure. The final results constitute an initial model for the propagation incurred by a relay-based communications system for fourth generation (4G) network in railways.

An overview of the ETCS braking curves

The purpose of the article is to present an overview of the ETCS braking curves according to the European specifications. Though the ETCS specifications lay down the basic principles for the braking curves and the associated information displayed to the driver, but there is still no harmonized method to compute them. In the absence of any requirement, the algorithms of the ETCS on-board suppliers lead to different braking distances for a given type of rolling stock. For cross border trains, the differences through national rules require the implementation in the ETCS on-board of several national braking curves. The main features of the braking curves have been reviewed in the article, including the following main aspects: basic principle to ensure of the train movement safety, the purpose and main types of ETCS braking curves, the input parameters for braking curve calculation, construction of the emergency brake deceleration curve, and guaranteed emergency brake deceleration. The basic principle to ensure the train movement safety is based on their separation in fixed block distance (for a conventional signalling system) or in moving block-sections (for ERTMS/ETCS level 3). To ensure these principles the ETCS onboard computer must predict the decrease of the train speed in the future, from a mathematical model of the train braking dynamics and of the track characteristics ahead. This prediction of the speed decrease versus distance is called a braking curve. The minimum interval between trains under automatic train protection system is defined by minimum movement authority that comprises odometer tolerance, driver allowance, ATP reaction times, brake application time. Braking curve predict of the train speed decrease versus distance by the ERTMS/ETCS on-board equipment, from a mathematical model of the train braking dynamics and the track characteristics ahead. ETCS supervises both the position and speed of the train to ensure they continuously remain within the allowed speed and distance limits, and if necessary it will command the intervention of the braking system to avoid any risk of the train exceeding those limits. There have been considered in the work the main ETCS braking curves and the supervision limits of the EBD braking curves, the movement authority, the end of authority, the most restrictive speed profile, the supervised location and others. Differences in braking on dry and wet rails are considered. The braking on dry rails is relatively easy to represent through a statistical model that take into account the dispersion of the braking performance. But on wet rails the physical phenomenon that occur when braking are still today extremely difficult to model. In order to overcome this difficulty, two distinct rolling stock correction factors have been considered in order to get the guaranteed emergency brake deceleration.

Second-order sliding mode controller with model reference adaptation for automatic train operation

ABSTRACTIn this paper, a new approach to model reference based adaptive second-order sliding mode control together with adaptive state feedback is presented to control the longitudinal dynamic motion of a high speed train for automatic train operation with the objective of minimal jerk travel by the passengers. The nonlinear dynamic model for the longitudinal motion of the train comprises of a locomotive and coach subsystems is constructed using multiple point-mass model by considering the forces acting on the vehicle. An adaptation scheme using Lyapunov criterion is derived to tune the controller gains by considering a linear, stable reference model that ensures the stability of the system in closed loop. The effectiveness of the controller tracking performance is tested under uncertain passenger load, coupler-draft gear parameters, propulsion resistance coefficients variations and environmental disturbances due to side wind and wet rail conditions. The results demonstrate improved tracking performance of the proposed control scheme with a least jerk under maximum parameter uncertainties when compared to constant gain second-order sliding mode control.

Accurate Human Localization for Automatic Labelling of Human from Fisheye Images

Deep learning networks like Convolutional Neural Networks (CNNs) show successful performances in many computer vision applications such as image classification, object detection, and so on. For implementation of deep learning networks in embedded system with limited processing power and memory, deep learning network may need to be simplified. However, simplified deep learning network cannot learn every possible scene. One realistic strategy for embedded deep learning network is to construct a simplified deep learning network model optimized for the scene images of the installation place. Then, automatic training will be necessitated for commercialization. In this paper, as an intermediate step toward automatic training under fisheye camera environments, we study more precise human localization in fisheye images, and propose an accurate human localization method, Automatic Ground-Truth Labelling Method (AGTLM). AGTLM first localizes candidate human object bounding boxes by utilizing GoogLeNet-LSTM approach, and after reassurance process by GoogLeNet-based CNN network, finally refines them more correctly and precisely(tightly) by applying saliency object detection technique. The performance improvement of the proposed human localization method, AGTLM with respect to accuracy and tightness is shown through several experiments.

Effects of fluency training on reading competence in primary school children: The role of prosody

Reading fluency defined as speed, accuracy, and prosody, is a critical component of reading development. The purpose of this research was to compare the efficacy of automaticity versus prosody programmes on reading comprehension. The study included 122 Spanish primary-school children (74 second and 48 fourth graders), randomly assigned to one of three groups: (a) automaticity training, which consisted of repeated reading with a focus on speed and accuracy plus phonological and orthographic awareness activities; (b) prosody training, which consisted of repeated reading with a focus on expressiveness plus prosody sensitivity activities; and a (c) ‘no treatment’ control group. Multiple measures were used to determine pre-post training performance in reading fluency—automaticity and prosody—and comprehension. Prosody training proved superior to automaticity training in promoting automaticity and prosody. Prosody and automaticity training in fourth graders resulted in superior sentence comprehension compared to controls. The importance of prosody for reading development in primary school is discussed.

Hybrid model reference adaptive second order sliding mode controller for automatic train operation

This study presents a non-linear longitudinal dynamic model of high speed trains by considering the rolling stock, train formations, rolling resistance, track environment, etc., and longitudinal controller design for Automatic Train Operation (ATO) as well. For a controller design, a novel hybrid model reference based adaptive super twisting sliding mode control scheme is proposed. The control law comprises a non-linear model reference adaptive control (MRAC) and an adaptive second order sliding mode control (SMC) in order to make the control signal continuous and to alleviate the chattering phenomenon. The adaptive gains of the proposed control law are derived based on Lyapunov approach that ensures global stability with exact tracking in finite time. A linear stable reference model is developed with the ideal train-track parameters to generate the reference trajectory. The effectiveness of the presented concept is demonstrated by introducing load variation, model uncertainties and external disturbances in the non-linear model of the train and its performance is compared with the non-linear MRAC and the second order sliding mode controller.

A flexible and robust train operation model based on expert knowledge and online adjustment

Most existing automatic train operation (ATO) models are based on different train control algorithms and aim to closely track the target velocity curve optimized offline. This kind of model easily leads to some problems, such as frequent changes of the control outputs, inflexibility of real-time adjustments, reduced riding comfort and increased energy consumption. A new data-driven train operation (DTO) model is proposed in this paper to conduct the train control by employing expert knowledge learned from experienced drivers, online optimization approach based on gradient descent, and a heuristic parking method. Rather than directly to model the target velocity curve, the DTO model alternatively uses the online and offline operation data to infer the basic control output according to the domain expert knowledge. The online adjustment is performed over the basic output to achieve stability. The proposed train operation model is evaluated in a simulation platform using the field data collected in YiZhuang Line of Beijing Subway. Compared with the curve tracking approaches, the proposed DTO model achieves significant improvements in energy consumption and riding comfort. Furthermore, the DTO model has more advantages including the flexibility of the timetable adjustments and the less operation mode conversions, that are beneficial to the service life of train operation systems. The DTO model also shows velocity trajectories and operation mode conversions similar to the one of experienced drivers, while achieving less energy consumption and smaller parking error. The robustness of the proposed algorithm is verified through numerical simulations with different system parameters, complicated velocity restrictions, diverse running times and steep gradients.

FORMALIZATION OF LOCOMOTIVE DRIVER ACTIVITY TENSION INDICATOR BASED ON THE ERGONOMIC MODEL

Purpose. A key factor contributing to safety and quality of ergatic system train-driver is intensity of locomotive crew’s work. The aim of this is formalization of locomotive driver activity tension indicator. Methodology. One of characteristics of driver activity tension is difference between time allotted to complete task, and necessary (external reserve or deficiency time). The sets of major and minor operations in management of train locomotive in different train situations were identified. Using methods of fuzzy logic, concept of materiality of of locomotive is presented in form of a set of linguistic variables. To determine function membership of elements of set the importance of of locomotive method of expert evaluations was used. Coefficient of temporary tension is presented in form of fuzzy number L-R-type. Findings. It was found value of relative number of operations of locomotive control according to distribution using parameter of importance. To determine most tensioned mode of driver ranking traffic condition according to parameter of relative amounts of important management operations was conducted. The most difficult modes are front hindrance, movement in unfavorable weather conditions and departure from station to running line. Originality. The introduction of value conventional importance of operation allowed us to more accurately describe terms of train driving. For first time presents determination of tension of driver’s in form of a unimodal fuzzy number, which will make it possible to use methods of theory of artificial intelligence to simulate activity of locomotive driver and develop intelligent control systems. Practical value. There were obtained opportunity to consider such an important quality indicator, as the tension of driver’s work in systems of automatic train control and traffic safety. Ranking onboard situations and control operations enables us to justify order of management action for improvement of algorithms of system of train driving.

Toughness of Railroad Concrete Crossties with Holes and Web Openings

Prestressed concrete sleepers (or railroad ties) are principally designed in order to carry wheel loads from the rails to the ground of railway tracks, as well as to secure rail gauge for safe train travels. Their design takes into account static and dynamic loading conditions. In spite of prestressed concrete crossties being most commonly used in railway tracks, there have always been many demands from rail engineers to improve the serviceability and functionality of concrete crossties. For example, signaling, fiber optic, equipment cables are often damaged either by ballast corners or by the tamping machine. There has been a need to re-design concrete crossties to incorporate cables internally so that they would not experience detrimental or harsh environments. Also, many concrete crossties need a retrofit for an automatic train control device and similar signaling equipment. In contrast, the effects of holes and web openings on the structural capacity of concrete crossties have not been thoroughly investigated. This paper accordingly highlights the experimental investigations into the effect of holes and web openings on the toughness and ductility of concrete crossties. The key outcome of this research is to enable a better decision making process for retrofitting prestressed concrete crossties with holes and web openings in practice.

Experimental Characterization of LTE Wireless Links in High-Speed Trains

Multimedia and data-based services experienced a nonstopping growth over the last few years. People are continuously on the move using devices to access multimedia contents or other data-based services. Due to this, railway companies are showing a great interest in deploying broadband mobile wireless networks in high-speed-trains with the aim of supporting both passenger services provisioning as well as automatic train control and signaling. Nowadays, the most widely used technology for communications between trains and the railway infrastructure is GSM for Railways (GSM-R); however, it has limited capabilities to support such advanced services. Due to its success in the mass market, Long Term Evolution (LTE) seems to be the best candidate to substitute GSM-R. In this paper, we experimentally characterize the downlink between an LTE Evolved NodeB (eNodeB) and a high-speed train in a commercial high-speed line. We consider two links: the one between the eNodeB and the antennas placed outdoors on the train roof, and the direct link between the eNodeB and a receiver inside the train. Such a characterization consists in assessing the path loss, the Signal to Noise Ratio, the K-Factor, the Power Delay Profile, the delay spread, and the Doppler Power Spectral Density.

自動列車運転装置を用いた省エネルギー運転制御手法の実証試験

自動列車運転装置を用いた省エネルギー運転制御手法の実証試験