Energy-efficient Train Research Articles

Train Control and Schedule Integrated Optimization With Reversible Substations

The Carbon Neutralization is a grand strategy to deal with global climate change. As a major energy consumer, urban rail transit (URT) system should contribute corresponding solutions. This paper proposes an energy-efficient train control and schedule integrated optimization method under DC traction power supply system (TPSS) with reversible substation (RSS). The dynamic coupling mechanism between the train movement process and the RSS working mode is analyzed and modeled. Aiming to minimize the total RSS energy consumption for the train traveling over multi-interstation with a given total trip time, the train trajectory and schedule optimization problem is integrated in the same level, and is formulated as a multi-phase optimal control problem (MOCP) to be solved by a pseudo-spectral method (PSM). As a supplement, a customized dynamic programming (DP) method for the problem is introduced to verify the validity of the proposed solution framework. Simulation results under the models of the RSS energy together with conventional rectifier substation energy and vehicle mechanical energy are presented and compared, indicating that the proposed methodology has great energy saving performance for the new-type DC railway system.

SPIDE: A purely spike-based method for training feedback spiking neural networks

Spiking neural networks (SNNs) with event-based computation are promising brain-inspired models for energy-efficient applications on neuromorphic hardware. However, most supervised SNN training methods, such as conversion from artificial neural networks or direct training with surrogate gradients, require complex computation rather than spike-based operations of spiking neurons during training. In this paper, we study spike-based implicit differentiation on the equilibrium state (SPIDE) that extends the recently proposed training method, implicit differentiation on the equilibrium state (IDE), for supervised learning with purely spike-based computation, which demonstrates the potential for energy-efficient training of SNNs. Specifically, we introduce ternary spiking neuron couples and prove that implicit differentiation can be solved by spikes based on this design, so the whole training procedure, including both forward and backward passes, is made as event-driven spike computation, and weights are updated locally with two-stage average firing rates. Then we propose to modify the reset membrane potential to reduce the approximation error of spikes. With these key components, we can train SNNs with flexible structures in a small number of time steps and with firing sparsity during training, and the theoretical estimation of energy costs demonstrates the potential for high efficiency. Meanwhile, experiments show that even with these constraints, our trained models can still achieve competitive results on MNIST, CIFAR-10, CIFAR-100, and CIFAR10-DVS.

Intelligent Energy-Efficient Train Trajectory Optimization Approach Based on Supervised Reinforcement Learning for Urban Rail Transits

Artificial intelligence of things (AIoT)-enabled intelligent automatic train operation (iATO) is an urgently needed technology to expand the capability of ATO in addressing the real-time responsiveness and dynamic online challenges to energy-efficient train trajectory optimization (TTO) and its associated ride-comfort, punctuality, and safety issues in modern urban rail transit networks. This paper proposes a three-step supervised reinforcement learning-based intelligent energy-efficient train trajectory optimization (SRL-IETTO) approach for iATO by hybrid-integrating deep reinforcement learning (DRL) and supervised learning. First, multiple objectives are formulated based on real-time train operation and systematically integrated into the RL algorithm by a binary function-based goal-directed reward design method. Second, an IETTO model is established to handle uncertain disturbances in real-time train operation and generate optimal energy-efficient train trajectories online by optimizing energy efficiency and receiving supervisory information from trajectories of pre-trained TTO models. Finally, numerical simulations are implemented to validate the effectiveness of the SRL-IETTO using in-service subway line data. The results demonstrate the superiority and improved energy saving of the proposed approach, and confirm its adaptability to online trip time adjustments within the practical running time range under uncertain disturbances with less trip time error compared to other intelligent TTO algorithms.

Linear conductance update improvement of CMOS-compatible second-order memristors for fast and energy-efficient training of a neural network using a memristor crossbar array.

Memristors are two-terminal memory devices that can change the conductance state and store analog values. Thanks to their simple structure, suitability for high-density integration, and non-volatile characteristics, memristors have been intensively studied as synapses in artificial neural network systems. Memristive synapses in neural networks have theoretically better energy efficiency compared with conventional von Neumann computing processors. However, memristor crossbar array-based neural networks usually suffer from low accuracy because of the non-ideal factors of memristors such as non-linearity and asymmetry, which prevent weights from being programmed to their targeted values. In this article, the improvement in linearity and symmetry of pulse update of a fully CMOS-compatible HfO2-based memristor is discussed, by using a second-order memristor effect with a heating pulse and a voltage divider composed of a series resistor and two diodes. We also demonstrate that the improved device characteristics enable energy-efficient and fast training of a memristor crossbar array-based neural network with high accuracy through a realistic model-based simulation. By improving the memristor device's linearity and symmetry, our results open up the possibility of a trainable memristor crossbar array-based neural network system that possesses great energy efficiency, high area efficiency, and high accuracy at the same time.

Acceleration of Deep Neural Network Training Using Field Programmable Gate Arrays.

Convolutional neural network (CNN) training often necessitates a considerable amount of computational resources. In recent years, several studies have proposed for CNN inference and training accelerators in which the FPGAs have previously demonstrated good performance and energy efficiency. To speed up the processing, CNN requires additional computational resources such as memory bandwidth, a FPGA platform resource usage, time, power consumption, and large datasets for training. They are constrained by the requirement for improved hardware acceleration to support scalability beyond existing data and model sizes. This paper proposes a procedure for energy efficient CNN training in collaboration with an FPGA-based accelerator. We employed optimizations such as quantization, which is a common model compression technique, to speed up the CNN training process. Additionally, a gradient accumulation buffer is used to ensure maximum operating efficiency while maintaining gradient descent of the learning algorithm. To validate the design, we implemented the AlexNet and VGG-16 models on an FPGA board and laptop CPU along side GPU. It achieves 203.75 GOPS on Terasic DE1 SoC with the AlexNet model and 196.50 GOPS with the VGG-16 model on Terasic DE-SoC. Our result also exhibits that the FPGA accelerators are more energy efficient than other platforms.

Collaborative optimization of energy-efficient train schedule and train circulation plan for urban rail

It is of great practical significance to save train traction energy for reducing the operation cost of urban rail transit. The energy-efficient train scheduling without combining with train circulation planning may inadvertently increase the other cost of rolling stocks, and finally lead to an increment of the total operation cost. This paper studies the integrated problem of energy-efficient train scheduling and train circulation planning for urban rail, and aims to reduce the total operation cost of rolling stocks including energy consumption. Its main challenge is to simultaneously solve three subproblems, namely the saving of train's traction energy in each rail section, the utilizing of regenerative braking energy and the optimizing of train circulation plan. We construct an optimization model to simultaneously optimize schedule and train circulation plan. Based on the designing of a strategy to create the train circulation plan for each train schedule, an efficient particle swarm algorithm is formed to solve our proposed model. The numerical experiments based on Guangzhou Metro Line 9 of China illustrate that the collaborative optimization can reduce the total operation cost of trains by 4.48% compared with the initial train schedule.

Optimization of metro vertical alignment for minimized construction costs and traction energy: A dynamic programming approach

Vertical alignment significantly affects the construction costs and train traction energy use of a metro system. Existing studies on this subject mostly optimized the vertical alignment while considering train traction energy consumption or construction costs only. There is still a lack of efficient approach for the vertical alignment optimization problem which considers both construction costs and traction energy consumption. To this end, this paper proposes a two-stage optimization program involving both the design of metro vertical alignment and train speed profile, with the objective of minimizing the total costs of construction and train traction energy. An iterative approach is proposed for solving the two-stage program, in which a dynamic programming (DP) algorithm with a backward search method is designed to seek an optimal vertical alignment given an energy-efficient train speed profile. The model and algorithm approach are tested on real-world case studies on the Line 14 of Guangzhou Metro in China. The results show that compared with the existing heuristic algorithms, the DP approach performs better in computation time and solution quality. Moreover, the optimized vertical alignments outperform that designed by experienced designers in terms of total costs of construction and traction energy consumption, with an average savings rate of 6.0%.

THETA: A High-Efficiency Training Accelerator for DNNs With Triple-Side Sparsity Exploration

Training deep neural networks (DNNs) on edge devices has attracted increasing attention in real-world applications for domain adaption and privacy protection. However, deploying DNN training on resource-limited edge devices is challenging as there are massive computations and data transportation in training. To address this issue, we propose an energy-efficient training accelerator in this work by employing a hybrid compression strategy. Here, various data redundancies are fully exploited, and the real triple-side sparsity is achieved. Hence, the computational complexity is drastically reduced with negligible accuracy loss across a range of transfer learning tasks. To facilitate triple-side zero-skipping operations during different training stages, we first present a novel sparse data representation and a triple-sparsity index matching scheme. Second, a sparse tensor processing unit (STPU) arranged in a hierarchical structure is developed, which enables a flexible dataflow to process convolutional (Conv) and fully connected (FC) layers with diverse computational patterns throughout the entire training. Third, an auxiliary processing unit (APU) is designed to execute some postprocessing operations, such as rectified linear unit (ReLU) and on-the-fly pruning. Finally, the training accelerator is implemented under Taiwan Semiconductor Manufacturing Company (TSMC) 28-nm process and evaluated on multiple benchmarks. The experimental results show that THETA achieves 7.28–22.32 tera operations per second (TOPS) and 45.24–133.70 TOPS/W in performance and energy efficiency, reducing 40– <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$72\times $ </tex-math></inline-formula> training time and 19– <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$63\times $ </tex-math></inline-formula> energy consumption over dense training, respectively. Compared with the prior art, our design offers <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$1.6\times $ </tex-math></inline-formula> throughput and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$1.9\times $ </tex-math></inline-formula> energy efficiency, respectively.

Joint optimization combining the capacity of subway on‐board energy storage device and timetable

On-board energy storage devices (OESD) and energy-efficient train timetabling (EETT) are considered two effective ways to improve the usage rate of regenerative braking energy (RBE) of subway trains. EETT is less costly but has lower ceilings, whereas OESD, although expensive, maximizes the reuse of RBE. To make the RBE usage rate of the train group reach the required target, a typical problem is how to use the OESD with the minimum capacity to reduce the cost. With this consideration, this paper proposes an optimization method that combines EETT and OESD. In this method, the joint optimization of timetable and OESD capacity allocation in case OESDs of different capacities can be carried between trains is considered to further improve the optimization. A non-linear programming model is built to minimize the total OESD capacity of the train group under the constraint of reaching target RBE usage. An efficient heuristic algorithm is designed to simultaneously obtain the optimal timetable and the matching capacity allocation scheme of OESDs. Finally, some simulation cases based on real-line data are designed to verify the effectiveness of the method.

Semi-Synchronous Federated Learning for Energy-Efficient Training and Accelerated Convergence in Cross-Silo Settings

There are situations where data relevant to machine learning problems are distributed across multiple locations that cannot share the data due to regulatory, competitiveness, or privacy reasons. Machine learning approaches that require data to be copied to a single location are hampered by the challenges of data sharing. Federated Learning (FL) is a promising approach to learn a joint model over all the available data across silos. In many cases, the sites participating in a federation have different data distributions and computational capabilities. In these heterogeneous environments existing approaches exhibit poor performance: synchronous FL protocols are communication efficient, but have slow learning convergence and high energy cost; conversely, asynchronous FL protocols have faster convergence with lower energy cost, but higher communication. In this work, we introduce a novel energy-efficient Semi-Synchronous Federated Learning protocol that mixes local models periodically with minimal idle time and fast convergence. We show through extensive experiments over established benchmark datasets in the computer-vision domain as well as in real-world biomedical settings that our approach significantly outperforms previous work in data and computationally heterogeneous environments .

Reduction of Power Consumption and CO2 Emissions as a Result of Putting into Service High-Speed Trains: Polish Case

This article presents the results of analyses of energy efficiency and environmental outcomes of putting into operation high-speed trains—first of that kind in Poland—in 2014 on upgraded 3 kV DC lines. Comparative analyses of these trains in Poland have not been carried out so far and provide a new approach to assessing the feasibility of their adoption and, in a broader context, the advisability of replacing conventional with high-speed rail transport. The analysis supports the advantages of high-speed railway (HSR) and the need to develop a network of high-speed connections using energy-efficient trains in Poland. Putting these trains into service reduced CO2 emissions, which is significant in Poland as a country relying heavily on power generated in coal power plants and complies with the assumptions of the sustainable development policies and EU Green Deal. The outcomes of our analysis can be used for transportation planning to mitigate climate change and the presented approach provides a method of quantification of environmental impacts of HS trains.

Energy-Efficient Model Predictive Train Traction Control With Incorporated Traction System Efficiency

The control system for energy-efficient train operation with the inclusion of a detailed train motion model and train traction system energy efficiency is presented in the paper. A piecewise affine train model is constructed with the parameters obtained for the electromotive train of an industrial manufacturer. The model encompasses intrinsic features of the train system such as linearized resistance force, a set of traction and braking force physical limitations and passengers comfort constraints. The resulting quadratic optimization problem is solved parametrically through dynamic programming giving the off-line precomputed optimal control law that is a function of train speed and traversed path. The on-line computed traction force profile is then tuned with respect to the traction system energy efficiency. The developed control system is evaluated on a detailed real case study scenario put together with a railway operator and the train manufacturer. The presented results show the possibility of significant energy consumption reductions achieved by energy-efficient train control.

Energy‐optimal study integrated speed trajectories, timetable and the layout of neutral sections for high‐speed railway

Speed trajectories, timetable and the layout of neutral sections are the key elements that affect train energy consumption (EC). This paper combines the microscopic energy-efficient train control (EETC), macroscopic energy-efficient train timetabling (EETT) and neutral sections location planning (NSLP), to achieve minimum EC in bi-directional high-speed railway. An integrated model is first established based on neutral sections distribution model, timetable model and a multi-phase model. Then, a double loop genetic algorithm (DLGA) is designed for matching the proposed model. The case studies are conducted by the proposed model and algorithm on the Wuhan-Yueyang rail corridor. A simple case for single-train movement in one section is primarily executed with variable neutral section location, to investigate the impact of neutral section location for train operation. The results indicate that neutral section locating at the coasting (CO) and cruising (CR) zones is the prior choice for energy conservation. Then, three integrated problems (EETC, EETC + EETT and EETC + EETT + NSLP) are carried out to obtain the energy-optimal speed trajectories, timetable and the layout of neutral sections, showing that EETC + EETT + NSLP increases the regenerative energy recycling from 14.2% to 49.1% and achieves 9.2% energy reduction compared with EETC.

Energy-efficient high-speed train rescheduling during a major disruption

Disruptions are inevitable in daily train operations, and can cause high-speed trains to deviate from their official schedules. Therefore, the efficient rescheduling of disrupted trains is critical for ensuring smooth daily operations. We sought to determine the appropriate arrival and departure times and orders of trains at each station during a disruption, and the speed profile of each train, to reduce the delay costs and energy consumption. To embed the train speed profile corresponding to energy saving into the train rescheduling problem, a space–time–speed network is applied for problem formulation. Thus, the energy-efficient train speed profile is embedded in the speed dimension in the space–time–speed hypernetwork. The detailed train speed profile between two stations is formulated as a multiple-phase optimal control model, which is solved using a pseudospectral method. Then, an integer linear programming model based on multicommodity flow is built to solve the train rescheduling problem. We decompose it via the alternating direction method of multipliers into a series of easy-to-solve shortest path subproblems. Each subproblem is then efficiently solved using a dynamic programming algorithm. Finally, the Xi’an–Chengdu high-speed railway line is used to test our model and algorithms, thereby demonstrating the trade-off between passenger-service quality and energy efficiency.

Determining the awareness and knowledge of officers towards ship energy efficiency measures

Improving energy efficiency practices on ships is crucial for both economic and environmental purposes. Both technical and operational applications are made to increase energy efficiency on ships. The success of energy efficiency applications on ships depends on the effective work of seafarers. Therefore, awareness and knowledge about the energy efficiency of seafarers should be enough. The aim of the study is to determine the awareness and knowledge level of seafarers on energy efficiency on ships and to evaluate energy efficiency practices in maritime transportation companies from the perception of seafarers. In the study, the survey method was used. As a result of the survey, it was found that the knowledge level of the seafarers on ship energy efficiency was low, almost half of them did not receive energy efficiency training but they were informed about the importance of the issue. At the end of the study, it was evaluated that the awareness and knowledge level of officers should be increased in order to improve energy efficiency on ships and suggestions on this issue were determined.

Robust energy-efficient train speed profile optimization in a scenario-based position—time—speed network

Robust energy-efficient train speed profile optimization in a scenario-based position—time—speed network

Cooperative Eco-Driving of Multi-Train Under dc Traction Network

Cooperative eco-driving of multi-train combined with dc railway system can achieve energy saving at the substation level for the metro system. This article presents an integrated optimization method for eco-driving by combining energy-efficient train control and dc traction power supply system (TPSS) model, to minimize the substation energy consumption. The coupled time-varying TPSS models for single train and multi-train are established, and a novel power flow calculation (PFC) method is proposed, to directly obtain the precise power flow distribution without iterations. Then, customized dynamic programming (DP) algorithms for specific TPSS models are developed, and optimal speed profiles of the tracking train and the preceding train can be respectively obtained by the multi-phase DP method. The former maximizes the utilization of a distributed regenerative braking energy (RBE), and the latter actively generates RBE to be fully utilized. Finally, the performance of the proposed method is analyzed by implementing the numerical experiments with field data from Guangzhou Metro Line 8, and the effectiveness is verified by comparing it with a previous study-based heuristic algorithm (HA). The simulation result indicates that the proposed method can also effectively reduce the lost energy on the catenary, suppress catenary voltage fluctuations, and precise match power during the RBE transmission phase.

A Two-Step Method for Energy-Efficient Train Operation, Timetabling, and Onboard Energy Storage Device Management

This article proposes a novel two-step approach to concurrently optimize the train operation, timetable, and energy management strategy of the onboard energy storage device (OESD) to minimize the net energy consumption for a whole urban railway line. In Step 1, approximating functions representing the minimum net energy consumption of each specific interstation operation is obtained by data fitting based on the previous research outcomes. In Step 2, the optimal running time, initial state of energy (ISOE) of OESD, train speed profiles, discharge/charge management of the OESD during each interstation journey and at each station are obtained by applying convex optimization formulated by approximating functions gained in Step 1. The method is first tested with several general train cases to show its robustness and adaptability. Then, a real-world case based on Beijing metro Yizhuang line is studied and the optimal solution is found to reduce the net energy consumption 1.04%, 2.09%, and 23.77% for a service cycle of a single train when compared to other operation scenarios, i.e., fully charged, no management, and no OESD scenario, respectively. The approach is also computationally efficient with a computational time less than 1 min, namely 38.86 s for the upline and 48.38 s for the downline, spent on finding the optimal solution.

Energy-efficient operation by cooperative control among trains: A multi-agent reinforcement learning approach

With the ever-increasing operating mileage of urban rail transit systems, a large amount of energy is consumed for train operation, and the energy-saving for train operation has become an attractive topic in the research community in recent years. In order to minimize the net energy consumption (i.e., the difference between the traction energy and the reused regenerative energy), an energy-efficient train operation method using the cooperative control approach is proposed in this paper. Firstly, a set of mathematical models for the cooperative control are formulated with considering the transmission mechanism of regenerative energy. Then, a multi-agent reinforcement learning algorithm is designed to obtain the cooperative driving strategy for trains. In this algorithm, the global Q-function is decomposed into several local Q-functions by using value function factorization method. A proposed update method is applied to update the local Q-functions according to the transmission mechanism of regenerative energy. Finally, the case studies built upon a metro line is performed to illustrate the effectiveness of the proposed method. According to the simulation results, the net energy consumption is reduced by more than 11.1% compared to the method in which the driving strategy of each train is independently optimized.

Energy Efficient Metro Train Running Time Rescheduling Model for Fully Automatic Operation Lines

AbstractReducing energy consumption without degrading the normal operation of metro trains and service quality has received increasing attention. Besides, fully automatic operation (FAO), for which...