Individual Charging Research Articles

Research on Machine Learning-Based Method for Predicting Industrial Park Electric Vehicle Charging Load

To achieve global sustainability goals and meet the urgent demands of carbon neutrality, China is continuously transforming its energy structure. In this process, electric vehicles (EVs) are playing an increasingly important role in energy transition and have become one of the primary user groups in the electricity market. Traditional load prediction algorithms have difficulty in constructing mathematical models for predicting the charging load of electric vehicles, which is characterized by high randomness, high volatility, and high spatial heterogeneity. Moreover, the predicted results often exhibit a certain degree of lag. Therefore, this study approaches the analysis from two perspectives: the overall industrial park and individual charging stations. By analyzing specific load data, the overall framework for the training dataset was established. Additionally, based on the evaluation system proposed in this study and utilizing both Multilayer Perceptron (MLP) and Long Short-Term Memory (LSTM) algorithms, a framework for machine learning-based load prediction methods was constructed to forecast electric vehicle charging loads in industrial parks. Through a case analysis, it was found that the proposed solution for the short-term prediction of the charging load in industrial park electric vehicles can achieve accurate and stable forecasting results. Specifically, in terms of data prediction for normal working days and statutory holidays, the Long Short-Term Memory (LSTM) algorithm demonstrated high accuracy, with R2 coefficients of 0.9283 and 0.9154, respectively, indicating the good interpretability of the model. In terms of weekend holiday data prediction, the Multilayer Perceptron (MLP) algorithm achieved an R2 coefficient of as high as 0.9586, significantly surpassing the LSTM algorithm’s value of 0.9415, demonstrating superior performance.

Energy management algorithm development for smart car parks including charging stations, storage, and renewable energy sources

In this study, a photovoltaic system and stationary energy storage unit integrated vehicle charging station energy management algorithm were developed using a long-short term based prediction model (LSTM). The aim of the proposed system is to develop intelligent car parks with controlled charging stations aggregated by centralized charging stations instead of individual or dispersed uncontrolled charging stations to eliminate the imbalance in the demands of charging. The system has four energy sources: grid, vehicle batteries, PV system, and the stationary battery group. By calculating the power demand for vehicles in the car park, a dynamic energy management algorithm has been developed that provides efficient use of energy resources by considering the power demand density. Moreover, the forecasting model has been created by LSTM) not only to adjust charging timing and effectively use energy sources. The model was performed by forecasting not to take energy from the grid at critical times (overloaded times). The grid load is analyzed by the energy management algorithm and run for different times of the year, creating a load profile for 16 vehicles. The results of this study show that energy demand during the critical time interval can be reduced by 8–20 % solely through load shifting before and after the critical time interval without any additional resource support. Moreover, while the integration of only the PV system supports the grid by 15–20 %, the optimal utilization of other energy sources during the relevant time frame supports the grid by 65–75 %. In addition, by collecting charging stations at a central point, energy storage capacity is increased and effective energy management is achieved. In conclusion, it is proposed that the infrastructure of such a parking system should be set up before the use of electric vehicles increases significantly.

Dynamic scheduling of electricity demand for decentralized EV charging systems

The rapid growth of electric vehicles (EVs) has brought forth new challenges to the power grid. Further, the simultaneous charging of EVs could lead to peak demand, potentially causing overloading, voltage swings, and other grid-related problems. To address these issues and lower the high energy costs faced by EV owners and grid operators, EV charging must be optimized. The study proposes an innovative strategy that utilizes decentralized charging systems to lessen the impact of EVs on the grid. A decentralized EV scheduling strategy offers scalability. Thus, making it suitable for a large EV population and it remains resilient to the dynamic arrivals of the EVs. The approach aims to balance the load on the grid and improve the effectiveness of charging operations. To achieve this, a convex optimization problem has been developed to effectively regulate the charging procedure, taking into account the distinct attributes of each EV. The mechanism operates by dividing time into several intervals. Each electric vehicle in the system autonomously adjusts its charging rate during the assigned time slots, with the goal of minimizing individual charging expenses. Moreover, the system demonstrates flexibility in deciding when to charge and discharge, allowing prioritization based on individual EV battery levels and power grid conditions. As a result, the cost analysis was conducted using the number of EVs and the average group size. A comparison of computational times between centralized and decentralized systems was undertaken to demonstrate the efficacy of the system.

How Many More Public Charging Stations Do We Need? A Data-Driven Approach Considering Charging Station Overflow Dynamics

The development of public charging infrastructure is crucial to support mass electric vehicle (EV) adoption. Although many cities worldwide have already installed an initial network of public chargers, it is often unclear whether the current supply of infrastructure is in line with demand and how many more charging stations are required to cope with future EV growth. In this sense, transactional charging data on the existing network can help answer these questions. We present a novel method that uses historical charging data as input to obtain answers to the following questions: (a) How many more chargers are required to meet future demand? and (b) Where should these new chargers be installed? By mining the individual charging behavior of EV drivers, we show that overflow dynamics can be found between charging stations. That is, when a preferred charging station is fully occupied, it is found that EV drivers divert to other charging stations nearby. Identifying these dynamics allows us to simulate the impact of a demand increase on the charging infrastructure network more accurately. We found the number of new chargers required to be significantly lower when considering overflow dynamics. Our simulations indicate that if demand is doubled, 30%–50% fewer charging points are needed compared with a situation in which overflow dynamics are neglected but the same failure rate is still maintained (i.e., percentage of failed charging sessions in the network). Determining the exact number of chargers will depend on the failure rate policymakers are willing to accept, reflecting the trade-off between charging convenience and utilization.

Spatial-temporal load forecasting of electric vehicle charging stations based on graph neural network

Accurate forecasting of the load of electric vehicle (EV) charging stations is critical for EV users to choose the optimal charging stations and ensure the safe and efficient operation of the power grid. The charging load of different charging stations in the same area is interrelated. However, forecasting the charging load of individual charging station using traditional time series methods is insufficient. To fully consider the spatial-temporal correlation between charging stations, this paper proposes a new charging load forecasting framework based on the Adaptive Spatial-temporal Graph Neural Network with Transformer (ASTNet-T). First, an adaptive graph is constructed based on the spatial relationship and historical information between charging stations, and the local spatial-temporal dependencies hidden therein are captured by the spatio-temporal convolutional network. Then, a Transformer network is introduced to capture the global spatial-temporal dependencies of charging loads and predict the future multilevel charging loads of charging stations. Finally, extensive experiments are conducted on two real-world charging load datasets. The effectiveness and robustness of the proposed algorithm are verified by experiments. In the Dundee City dataset, the MAE, MAPE, and RMSE values of the proposed model are improved by approximately 71%, 90%, and 67%, respectively, compared to the suboptimal baseline model, demonstrating that the proposed algorithm significantly improves the accuracy of load forecasting.

Machine Learning for Anomaly Detection in Electric Transportation Networks

This study introduces a sophisticated anomaly detection system based on machine learning. The system is specifically developed to enhance the dependability and safeguard the security of electric transportation networks, with a particular emphasis on the charging infrastructure for electric vehicles (EVs). Utilizing extensive datasets, the research examines several facets of charging stations, charging records, identified abnormalities, and following maintenance measures. The examination of the charging station demonstrates the system’s versatility in accommodating many charging circumstances, as seen by the range of power ratings, consumption patterns, and energy provided. Further examination of charging records provides comprehensive understanding of individual charging sessions, enabling the detection of irregularities such as atypical energy surges and extended charging durations. The machine learning system, having been trained and verified using this data, has a commendable degree of precision in identifying anomalies, as shown by the congruence between anticipated abnormalities and real results. The maintenance and repair measures carried out in reaction to identified abnormalities highlight the practical ramifications of the system, with proactive tactics utilized to reduce downtime and enhance charging station operations. The performance measures, including accuracy, recall, and F1 score, unequivocally validate the resilience of the anomaly detection system, guaranteeing precise identification while mitigating the occurrence of false positives and negatives. The seamless incorporation of machine learning into electric transportation networks, as shown by the results, not only amplifies the dependability and safeguarding of EV charging infrastructure but also establishes the system as an invaluable instrument for practical implementations. The research, in addition to offering a thorough examination of the system’s performance, elucidates forthcoming avenues for scalability, real-time monitoring, and interpretability, thereby making a valuable contribution to the wider discussion on the revolutionary capabilities of machine learning in the ever-changing realm of electric transportation.

Evaluation of hosting capacity of the power grid for electric vehicles – A case study in a Swedish residential area

The deployment of electric vehicles (EVs) is growing significantly in recent years. The increasing EV charging loads pose great stress on power grids in Sweden, as many existing power grids are not designed to host such large shares of new electric loads. Hence, studies investigating the impact of EV charging are needed. This study conducts a case study based on an existing Swedish residential power grid using real-life EV charging data to estimate the local grid hosting capacity (HC) for EVs. A combined time-series and stochastic HC assessment method is used with voltage deviation, cable loading and transformer loading as the performance indices. Uncertainty in EV charging locations and individual charging behaviour have been considered via Monte Carlo simulations. The power grid HC is analysed and compared under three charging strategies and four EV penetration levels. Study results show that a charging strategy based on low electricity prices gave lower HC due to simultaneous EV loads compared to the other two strategies: charging directly after plugging in the EV and an even charging load through the plug-in session. This implies the need for coordinated charging controls of EV fleets or diversified power tariffs to balance power on a large scale.

Single Elementary Charge Fluctuations on Nanoparticles in Aqueous Solution.

100 years ago, in 1923, the Nobel prize in physics was awarded for measurement of the unit charge. In addition to a profound impact on contemporary physics, this discovery has reshaped our understanding of charge-based interactions in chemistry and biology, ranging from oxidation and ionization to protein folding and metabolism. In a liquid, the discrete nature of the electric charge becomes prominent at the nanoscale when a charge carrier is exchanged between a molecule or a nanoparticle and the surrounding medium. However, our ability to observe the dynamics of such interactions at the level of a single elementary charge is limited due to the abundance of ions in water. Here, we report on the observation of single binding-unbinding events with elementary charge resolution at the surface of a nanoparticle suspended in water. Discrete steps in the electrical charge are revealed by analyzing the motion of optically trapped nanoparticles under the influence of an applied sinusoidal electric field. The measurements are sufficiently fast and long to observe individual (dis)charging events that occur on average every 3 s. Our results offer prospective routes for studying the dynamics of diverse chemical and biological phenomena on the nanoscale with elementary charge resolution.

Comparison of electric vehicle load forecasting across different spatial levels with incorporated uncertainty estimation

Accurate load forecasting is important to mitigate the negative impact of Electric vehicle integration into the existing grid. Previous studies mostly focus on individual or aggregated levels without specifying the impact of accuracy due to the selection of different spatial levels and lack the integration of uncertainty estimation in the forecasting models. To address these issues, this study compares the predictive performance of a Random Forest and Artificial Neural Networks at different spatial levels with 15-min resolution data across case studies (i) with 2 Electric Vehicles charging poles and 3 users, (ii) with 75 charging poles, 8 charging rails and 70 users. The outcome shows that forecasting the Electric Vehicle load of smaller case studies will require the presence or calendar information of users. Whereas in case studies with more than 10 charging piles, the features “previous week's power”, “hour of the day” and the “number of connections” can achieve similar results. The results also showed that the aggregated forecasting was more accurate than individual charging piles. Moreover, the uncertainty plot generated for a 90% prediction interval showed that the uncertainty estimates were more reliable for the case study with large numbers of Electric Vehicles.

A method for deriving battery one-way efficiencies

Batteries are becoming an important decarbonization technology because they can act as convenient energy storage in various applications. They are often part of larger, complex systems and, in order to preserve computational tractability, battery models are usually oversimplified. Majority of such battery models ignore dependency of the charging/discharging efficiency on the charging/discharging power rate and instead use a constant efficiency over the entire range of power rates. This paper presents a method for obtaining individual one-way charging and discharging efficiencies dependent on the charging/discharging power. The method consists of two parts. First, the roundtrip cycle efficiency is experimentally obtained for different pairs of charging and discharging power rates. Second, an optimization problem is solved to decompose the roundtrip efficiency into the charging and the discharging efficiency for different power rates, resulting in one-way (charging/discharging) efficiency characteristics. As a demonstration, we apply the obtained efficiency characteristics to an electric vehicle driving profile consisting of both charging and discharging stages under different power rates and experimentally validate the accuracy of the proposed method.

A genuine V2V market mechanism aiming for maximum revenue of each EV owner based on non-cooperative game model

As the number of electric vehicles (EVs) surges, EVs are expected to supply auxiliary services for power grid through the vehicle-to-grid (V2G) operation. Specifically, vehicle-to-vehicle (V2V), as a beneficial supplement of V2G, has received increasing attention owing to the capability for flexible EVs deployment and profitable individual interaction. In this paper, a genuine V2V market mechanism aiming for maximum revenue of each EV owner is proposed to construct a distributed electricity market. It reveals the dominance of EV owners from the perspective of market and allows the trade among EV owners in local distribution grid with charging points such as charging stations and auto parks. To this end, the implementation of V2V market mechanism is analyzed in detail, including the electricity price, the process and the integration with the internet of smart charging points (ISCP), a cyber-physical architecture. Then, EVs’ energy dispatching is formulated via the non-cooperative game model, in which each EV owner anticipates maximizing individual revenue by choosing feasible strategies. The existence and uniqueness of Nash Equilibrium (NE) in a non-cooperative game is proved by means of variation inequality (VI) theory. Finally, the ISCP-based optimal algorithm is developed to solve the NE efficiently, thereby maximizing reward of each EV owner. Comprehensive case studies show that the proposed V2V market mechanism can implement distributed dealing among EV owners with reducing individual charging cost and regional load variance effectively. Specially, the costs of EV1, EV8 and EV9 decrease by 31%, 36% and 34%, respectively. And V2V charging demonstrates much lower variance (10.5209) than regular charging (14.1124). Additionally, the convergence of the algorithm is also demonstrated, reaching the NE in less than 30 iterations for each EV.

A Comprehensive Data Analysis of Electric Vehicle User Behaviors Toward Unlocking Vehicle-to-Grid Potential

Electric vehicles (EVs) improve the power grid by increasing intermittent renewable energy consumption and providing financial support to EV users via vehicle-to-grid (V2G) integration. While estimating these advantages, a number of studies have neglected to consider the effect of driving and charging behavior patterns on their results. This article provides a framework that systematically evaluates EV driving and charging behaviors to improve charge management in the light of recent standards and advancements. In addition, the collected data on driving habits are analyzed in order to provide a consistent and usable dataset. By evaluating the individual and simultaneous charging demand characteristics, the V2G potential is further explored. Moreover, managerial recommendations for EV charging management are offered by improving the time step using the Bootstrap approach for more precise results than lower resolution. It is also addressed that the simultaneous use of a limited number of EVs required minimum time. According to the findings of this study, daily travel habits have a crucial influence in defining seasonal and individual charging demands. In order to continue with EV charging-related assessments with a confidence interval of more than 95%, the findings suggest that time steps of lower than ten minutes must be used. In addition, the purpose of this study is to assist researchers from academia and business with further information as they build initiatives linked to EV charging infrastructure and real-time charging management standards that account environmental aspects.

Charging demand prediction in Beijing based on real-world electric vehicle data

The accurate estimation and prediction of charging demand play an essential role in charging infrastructure planning, power grid laying and efficient operations. In this paper, three-month real-world travel and charging records of 25,489 electric passenger vehicles in Beijing are utilized to quantitatively study the travel patterns and charging behaviors of electric vehicles and support charging demand prediction. A “multi-level & multi-dimension” travel and charging characteristic parameters extraction framework is proposed with >60 electric vehicle behavior characteristic parameters extracted from time, spatial, and energy dimensions. Then a two-step GMM and K-means-based clustering model is proposed to cluster the electric vehicle users into six categories with different usage habits, attributes, and characteristics. A trip-chain-simulation-based charging demand prediction model is further established to generate a week-period continuous activities simulation for specific category vehicles, in which the interdependences of travel and charging characteristic parameters are considered to make the trip chain simulation closer to reality. The city charging demands are aggregated from individual charging prediction results at a spatial resolution of 0.46 km and time resolution of 15 min quantitatively; the spatiotemporal aggregation results show that the proposed model can realize high accuracy prediction for real-world charging demands. The results are expected to provide the foundation for research on charging infrastructure planning and the impact of charging behaviors on the grid load.

A Comparison of the Use of Pontryagin’s Maximum Principle and Reinforcement Learning Techniques for the Optimal Charging of Lithium-Ion Batteries

Battery systems are one of the most important components for the development of flexible energy storage for future applications. These comprise energy storage in both the mobility sector and stationary applications. To ensure the safe operation of multiple battery cells connected in series and parallel in a battery pack, it is essential to implement state of charge (SOC) equalization strategies. Generally, two fundamentally different approaches can be distinguished. On the one hand, these are passive approaches for SOC equalization that are based on including additional Ohmic resistors in a battery back over which equalization currents flow as long as the correspondingly connected cells have different voltages. Despite the simple implementation of such equalization circuits, they have a major drawback, namely wasting stored energy to perform the SOC equalization. This waste of energy goes along with Ohmic heat production, which leads to the necessity of additional cooling for batteries with large power densities. On the other hand, active SOC equalization approaches have been investigated, which allow for an independent charging of the individual cells. Especially, this latter approach has big potential to be more energy efficient. In addition, the potential for a reduction of Ohmic heat production may contribute to extending the lifetime of battery cells. To perform the individual charging of battery cells in an energetically optimal manner, this paper provides a comparison of closed-form optimization approaches on the basis of Pontryagin’s maximum principle and approaches for reinforcement learning. Especially, their accuracy and applicability for the implementation of optimal online cell charging strategies are investigated.

Understanding charging dynamics of fully-electrified taxi services using large-scale trajectory data

An accurate understanding of “when, where, and why” of charging activities is crucial for the optimal planning and operation of E-shared mobility services. In this study, we leverage a unique trajectory of a city-wide fully electrified taxi fleet in Shenzhen, China, and we present one of the first studies to investigate the charging behavioral dynamics of a fully electrified shared mobility system from both system-level and individual driver perspectives. The electric taxi (ET) trajectory data contain detailed travel information of over 20,000 ETs over one month period. By combing the trajectory and charging infrastructure data, we reveal remarkable regularities in infrastructure utilization, temporal and spatial charging dynamics as well as individual driver-level charging preferences. Specifically, we report that both temporal and spatial distributions of system-level charging activities present strong within-day and daily regularities, and most charging activities are induced by drivers’ shift schedules. Further, with 425 charging stations, we observe that the drivers show strong preferences over a small subset of charging stations, and the power-law distribution can well characterize the charging frequency at each charging station. Finally, we show that drivers’ shift schedules also dominate the individual charging behavior, and there are strikingly stable charging patterns at the individual level. The results and findings of our study represent lessons and insights that may be carried over to the planning and operation of E-shared mobility in other cities and deliver important justifications for future studies on the modeling of E-shared mobility services.

Simulation-Based Evaluation of Charging Infrastructure Concepts: The Park and Ride Case

In this study, a framework regarding park and ride facilities is presented and demonstrated to evaluate different approaches of charging concepts. The innovation in this study is that the framework can be used to evaluate arbitrary conductive charging concepts on a detailed level and on the basis of real usage data. Thus, the results can be broken down to the level of individual charging events and charging points. Among other factors, the study considers the expected growth in electric vehicles, the construction and operating costs for the investigated charging infrastructure, and the impact of heterogeneous electric vehicle fleets with different vehicle-specific charging powers. Since both technological and economic perspectives are considered in the framework, the study is relevant for all decision makers involved in the development and operation of charging infrastructure. The results in the investigated case of park and ride facilities show a high potential for cost-efficient low-power charging concepts. Thus, significantly higher energy volumes could be transmitted and better economic results could be achieved by the investigated low-power approaches. Especially for heterogeneous electric vehicle fleets, the number of available charging points appears to be more important than the charging power of the individual charging points in this case.

Optimal en-route charging station locations for electric vehicles: A new modeling perspective and a comparative evaluation of network-based and metanetwork-based approaches

In many regions across the world, the distribution density of electricity-charging stations is yet sufficient for taming the range anxiety concern of electric vehicle drivers. In such driving circumstances, making detours to find alternative charging opportunities is often an operational remedy to accomplish long-distance trips. To reduce the frequency and extra mileage of detouring, public charging infrastructures should be coordinately sited and constructed in a regional network. With the goal of minimizing the possible detour mileage, this paper describes an optimal charging station location problem for intercity highway networks and develops two approaches for modeling and solving it. Specifically, two mixed linear integer programming models are constructed in different network modeling paradigms: The first one is formed and solved on the basis of the original node-link network topology, with newly introduced integer variables such as path and subpath activation indicators to quantify the utilization status of paths and subpaths, respectively. The classic branch-and-bound algorithm encapsulating a bi-criteria label-correcting algorithm is designed for solving this network-based model. The other one is formulated and solved relying on the station-subpath metanetwork topology, as the result of a two-phase process that decomposes an individual routing and charging decision into two parts. The two phases are conducted on the node-link network and station-subpath metanetwork levels, respectively, where the first phase is done by solving a network-based, distance-constrained minimum-cost path problem, while the second phase collapses to a metanetwork-based, simple minimum-cost path problem. The findings we obtained through conducting this research are twofold: First, analytical and computational studies clearly identify the effectiveness of the two modeling and solution approaches, while the metanetwork-based approach exhibits its appealing computing advantage for solving problems of large size; second, the application of the solution approaches for a real-world network instance reveals how budget limit and distance limit impact the charging station locations, individual routing and charging decisions, and network cost and accessibility levels.

Explicit demand response potential in electric vehicle charging networks: Event-based simulation based on the multivariate copula procedure

This paper proposes a novel combined event-based simulation model for assessing the explicit demand response potential of electric vehicle (EV) charging networks. The model utilizes different multivariate copulas in generation of realistic artificial charging events that effectively retain the complex dependency structures and parameter distributions of real data important for accurate demand response simulation. A deterministic model is used to estimate the maximal explicit demand response potential of individual charging events based on technical requirements of the frequency containment reserve for disturbance situations (FCR-D) market. The proposed model achieved a mean absolute percentage error (MAPE) of 3.27% when considering averaged daily dispatchable FCR-D potentials, and a MAPE of 4.65% in prediction of dispatchable FCR-D potential with one workweek of data. The results and methodology have been verified and validated with real life data and through comparison with a previous non-copula application for EV FCR profile estimation which it outperformed. The combined event-based simulation model can boost active participation of EVs in power network balancing and is suitable for use in various practical and theoretical applications.

Empirical charging behavior of plug-in hybrid electric vehicles

Plug-in hybrid electric vehicles (PHEV) offer greenhouse gas emission reduction in car usage if charged frequently and driven mainly on electricity. However, little is known about the actual charging behavior of PHEV owners. Here, we investigate the daily charging of 10,488 Chevrolet Volt PHEV driven on a total of 4.3 million total driving days in the US and Canada. We propose a new method to detect the frequency of individual charging behavior from the daily utility factor and daily distance travelled. Our results show that no charging overnight occurs typically on 3–7% of the driving days per user and additional charging happens on 20–26% of the driving days. We also analyze the relation between charging frequency and utility factor for different user groups and days. Our results show that the utility factor should not be used as the only measure of environmental performance of PHEVs.

Forecasting Charging Point Occupancy Using Supervised Learning Algorithms

The prediction of charging point occupancy enables electric vehicle users to better plan their charging processes and thus promotes the acceptance of electromobility. The study uses Adaptive Charging Network data to investigate a public and a workplace site for predicting individual charging station occupancy as well as overall site occupancy. Predicting individual charging point occupancy is formulated as a classification problem, while predicting total occupancy is formulated as a regression problem. The effects of different feature sets on the predictions are investigated, as well as whether a model trained on data of all charging points per site performs better than one trained on the data of a specific charging point. Reviewed studies so far, however, have failed to compare these two approaches to benchmarks, to use more than one algorithm, or to consider more than one site. Therefore, the following supervised machine-learning algorithms were applied for both tasks: linear and logistic regression, k-nearest neighbor, random forest, and XGBoost. Further, the model results are compared to three different naïve approaches which provide a robust benchmark, and the two training approaches were applied to two different sites. By adding features, the prediction quality can be increased considerably, which resulted in some models performing better than the naïve approaches. In general, models trained on data of all charging points of a site perform slightly better on median than models trained on individual charging points. In certain cases, however, individually trained models achieve the best results, while charging points with very low relative charging point occupancy can benefit from a model that has been trained on all data.