CityCharge: Advanced modeling of urban electric vehicle charging demand patterns

Demand forecasting plays a critical role in supply chain management, enabling suppliers, manufacturers, and retailers to synchronise operations and enhance overall efficiency. Despite extensive research on time series forecast model selection, choosing the most appropriate forecasting model for a given time series remains a complex challenge, particularly in volatile and uncertain environments. The increasing availability of data and the emergence of new forecasting methods have introduced greater complexity, making automated model selection essential for improving forecasting accuracy and decision-making in supply chain operations. This study proposes an automated demand forecast model selection framework that integrates a broad range of statistical and machine learning models. A key feature of the framework is the optimisation of hyperparameters across all models, ensuring each method is fine-tuned for optimal performance. The approach is validated on the M3 monthly dataset, where it outperforms all previously submitted methods, demonstrating significant improvements in forecast accuracy. Additionally, the methodology is tested in a real-world supply chain setting, further showcasing its effectiveness in handling complex and dynamic demand patterns. By enhancing forecast accuracy and reducing the reliance on manual model selection, this research provides an efficient decision support system for supply chain demand forecasting in fast-changing supply chain environments.

Using grid-level credit card transactions, this study explains the spatial distribution of restaurant demand on Jeju Island and quantitatively tests how dimensional-topological properties of road networks affect demand patterns. Rather than being determined solely by unit-level attributes, restaurant demand is shaped by broader spatial contexts, such as network connectivity, route-configuration efficiency, and the legibility of the surrounding environment. These spatial factors constrain tourists’ movement and decision-making, influencing the frequency and patterns of commercial activity. In particular, road networks function as core infrastructure that structures travelers’ paths and influences accessibility to and choice among establishments. Demand does not arise merely from proximity; rather, the efficiency of movement and the clarity of routes condition the likelihood of visitation, making both the structural dimension and the topological arrangement of road networks central to explaining traveler behavior. To mitigate sensitivity to outliers and data non-normality, the study employs robust regression, thereby enhancing the reliability of the results. The findings indicate that restaurant demand responds more strongly to route legibility and network efficiency than to simple distance or physical accessibility. This implies that road networks are not mere conduits but structural determinants of behavioral radius and spatial cognition. The legibility and coherence of the built environment likewise emerge as explanatory factors in consumer decision-making. Grounded in spatial interaction theory, the study elucidates the spatial mechanisms of restaurant demand and provides empirical foundations for commercial district strategy and urban planning in tourist destinations. The findings suggest that integrating traveler mobility with spatial features enables more effective market management and policy intervention.

Ecosystems provide us with countless benefits, such as material resources, regulation of environmental processes, and opportunities for recreation. These benefits, known as ecosystem services (ES), support our daily welfare and well-being. ES arise from ecological, sociocultural, and economic interactions, and are influenced by both ecosystems’ capacity to provide services and society’s demand for them. ES are unevenly distributed across space, and their supply and demand change over time. Understanding the patterns of ES provision and consumption facilitates the evaluation of their sustainable use. Therefore, comprehensive assessments of ES production and consumption across spatial and temporal scales are essential to deepen our understanding of the ES concept and its role in natural resource management. In this thesis, I exemplify the use of the ES framework by assessing the spatial and temporal patterns of ES potential, supply, and demand. Overall, I demonstrate how to select and interpret indicators of ES potential, supply, and demand and address them using spatial and statistical methods. I study the provisioning services of forests (wood resources) and the cultural services provided by urban and peri-urban areas through three separate case studies. Each article examines ES aspects across distinct scales, ranging from continental to local. Two articles are at the European level, one of which also includes a temporal scale, and one is at the urban level. The results regarding wood ES show that the potential, supply, and demand for wood have all increased across Europe. Compared to demand, Europe has a substantial supply surplus, and the analysis of mismatches between the supply and demand indicates that, on average, Europeans have good spatial accessibility to wood resources. However, the growing trend of exploiting wood ES might affect the state of forest ecosystems and their capacity to provide high-quality ES other than wood. The findings regarding cultural ES suggest that subjective spatial characteristics of green spaces, such as perceived accessibility, play a bigger role in more frequent interactions with nature than the biophysical features of these spaces or the consumption of cultural ES itself. This indicates that urban residents demand better access to green spaces in order to fully enjoy and recognize the capacity of urban ecosystems to deliver high-quality cultural ES within close proximity to their homes. My thesis exemplifies the application of the ES framework in ES mapping, incorporates ES spatial flow into supply and demand mismatch evaluation, and highlights the importance of subjective human needs and perceptions regarding ES demand as vital parts of the ES framework. The evaluation of the distribution and trends in the potential, supply, and demand of the provisioning ES of wood, alongside the produced maps, supports resource monitoring of European forests. The same applies to the maps of wood ES supply–demand mismatches, which integrate the ES spatial flow through spatial accessibility analysis. These results can inform European forest management strategies, providing spatial insights into wood potential, supply, and demand, and their mismatches. Furthermore, the evaluation of the characteristics of green spaces’ use patterns emphasizes the importance of spatial perceptions in interactions with urban and peri-urban nature. This information can be communicated to decision-makers in the studied cities and used to enhance access to green spaces that provide vital cultural ES for urban populations.

Investigating the spatio-temporal differentiation patterns and driving factors of ecosystem services (ESs) supply and demand is of great significance for early warning of ecosystem imbalance risks and identifying regional natural resource supply–demand conflicts. This study takes the typical coal-grain overlapping area (CGOA) in Eastern China as the research object, dividing it into mining townships (MT) and non-mining townships (NMT) for comparative analysis. By integrating the InVEST model, ESs supply–demand ratio (ESDR) index, four-quadrant model, and the XGBoost-SHAP algorithm, the study systematically reveals the spatiotemporal differentiation characteristics and driving mechanisms of ESs supply and demand from 2000 to 2020. The results indicated that: (1) grain production (GP) service maintained a continuous supply–demand surplus, with the ESDR of NMT areas surpassing that of MT areas in 2020. The ESDR of water yield (WY) service was significantly influenced by interannual fluctuations in supply, showing deficits in multiple years. The decline in carbon sequestration (CS) service and sharp increase in carbon emissions led to a continuous decrease in the ESDR of CS service, with MT areas facing a higher risk of carbon deficit. (2) The spatial heterogeneity of ESs supply and demand was significant, with GP and CS services exhibiting a typical urban-rural dual spatial structure, and the overall region was dominated by the Type II ESs supply–demand matching (ESDM) pattern. The ESDR of WY service generally decreases from Southeast to Northwest across the region. with the Type IV ESDM pattern dominating in most years. (3) Human activities are the core driving force shaping the supply–demand patterns of ESs. Among these, land use intensity exhibits a nonlinear effect, high population density demonstrates an inhibitory effect, and MT areas are more significantly affected by coal mining subsidence. Natural environmental factors primarily drive WY service. The research findings can provide a scientific reference for the coordinated allocation of regional natural resources and the sustainable development of the human–land system.

In the modern era, demand forecasting enhances the decision-making tasks of industries for controlling production planning and reducing inventory costs. However, the dynamic nature of the fashion and apparel retail industry necessitates precise demand forecasting to optimize supply chain operations and meet customer expectations. Consequently, this research proposes the Formicary Zebra Optimization-Based Distributed Attention-Guided Convolutional Recurrent Neural Network (FZ-DACR) model for improving the demand forecasting. In the proposed approach, the combination of the Formicary Zebra Optimization and Distributed Attention mechanism enabled deep learning architectures to assist in capturing the complex patterns of the retail sales data. Specifically, the neural networks, including convolutional neural networks (CNNs) and recurrent neural networks (RNNs), facilitate extracting the local features and temporal dependencies to analyze the volatile demand patterns. Furthermore, the proposed model integrates visual and textual data to enhance forecasting accuracy. By leveraging the adaptive optimization capabilities of the Formicary Zebra Algorithm, the proposed model effectively extracts features from product images and historical sales data while addressing the complexities of volatile demand patterns. Based on extensive experimental analysis of the proposed model using diverse datasets, the FZ-DACR model achieves superior performance, with minimum error values including MAE of 1.34, MSE of 4.7, RMS of 2.17, and R2 of 93.3% using the DRESS dataset. Moreover, the findings highlight the ability of the proposed model in managing the fluctuating trends and supporting inventory and pricing strategies effectively. This innovative approach has significant implications for retailers, enabling more agile supply chains and improved decision making in a highly competitive market.

Inventory management is a crucial factor in retail operations as it influences cost efficiency, sales continuity, and customer satisfaction. In small-scale retail businesses, inventory planning is often performed manually, increasing the risk of overstock and stockout conditions. This study aims to develop a web-based inventory forecasting information system using the Weighted Moving Average (WMA) method to support effective inventory planning. The system integrates item data management, sales transaction recording, and demand forecasting within a single platform. The WMA method is applied to 12 months of historical monthly sales data using a three-period forecasting window with an optimized weight configuration of 5–1–7 to emphasize recent demand patterns. Forecasting accuracy is evaluated using Mean Absolute Deviation (MAD), Mean Squared Error (MSE), and Mean Absolute Percentage Error (MAPE). A case study conducted at Toko Tita shows that the WMA method outperforms the Simple Moving Average method by producing lower MAD and MAPE values, indicating better responsiveness to short-term demand fluctuations. The results demonstrate that the proposed system provides reliable quantitative information to support inventory procurement decisions, reduces manual calculation errors, and improves operational efficiency. Although forecasting errors increase during extreme demand changes, the system is practical and effective for daily inventory management in small retail businesses.

Pedestrian crossing facilities in urban areas are often positioned without adequate consideration of actual movement patterns, leading to low utilization rates, widespread informal crossings, and increased pedestrian-vehicle conflict risks. Traditional manual observation methods for assessing crossing behavior are time-consuming, subjective, and unable to capture continuous spatial-temporal movement dynamics at scale. This study aims to develop and evaluate an automated framework for extracting pedestrian trajectories and assessing the alignment between pedestrian desire lines and existing zebra crossing infrastructure. The methodology integrates YOLO11 fine-tuned object detection with ByteTrack multi-object tracking to process unmanned aerial vehicle (UAV) video data collected at an urban intersection in Surabaya, Indonesia. Pedestrian-vehicle conflict severity was quantified using Time-to-Collision (TTC)-based surrogate safety indicators, including Time Exposed to Time-to-Collision (TET) and Time Integrated Time-to-Collision (TIT). The results reveal substantial heterogeneity in crossing behavior, with distinct spatial clustering of informal crossing hotspots located away from the designated zebra crossing. Asymmetric yet bidirectional pedestrian demand patterns were observed across the study area. Based on trajectory-derived evidence, the study recommends strategic relocation of the zebra crossing approximately 135 meters south to better accommodate natural pedestrian flow and reduce vehicular traffic exposure. These findings demonstrate that deep learning-based trajectory analysis offers a practical, objective, and scalable approach for evidence-based pedestrian infrastructure planning, particularly applicable to rapidly urbanizing contexts in developing countries where conventional assessment resources are limited.

Analysis of the Future Supply and Demand Pattern of Global Rare Earth Resources

This paper addresses the critical need for efficient energy management in healthcare facilities, where fluctuating energy demands pose challenges to both operational reliability and sustainability objectives. Traditional energy management approaches often fall short in healthcare settings, resulting in inefficiencies and increased operational costs. To address this gap, the paper explores AI-driven methods for demand forecasting and load balancing and proposes an integrated framework combining Long Short-Term Memory (LSTM) networks, a genetic algorithm (GA), and SHAP (Shapley Additive Explanations), specifically tailored for healthcare energy management. While LSTM has been widely applied in time-series forecasting, its use for healthcare energy demand prediction remains relatively underexplored. In this study, LSTM is shown to significantly outperform conventional forecasting models, including ARIMA and Prophet, in capturing complex and non-linear demand patterns. Experimental results demonstrate that the LSTM model achieved a Mean Absolute Error (MAE) of 21.69, a Root Mean Square Error (RMSE) of 29.96, and an R2 of approximately 0.98, compared to Prophet (MAE: 59.78, RMSE: 81.22, R2 ≈ 0.86) and ARIMA (MAE: 87.73, RMSE: 125.22, R2 ≈ 0.66), confirming its superior predictive performance. The genetic algorithm is employed both to support forecasting optimisation and to enhance load balancing strategies, enabling adaptive energy allocation under dynamic operating conditions. Furthermore, SHAP analysis is used to provide interpretable, within-model insights into feature contributions, improving transparency and trust in AI-driven energy decision-making. Overall, the proposed LSTM–GA–SHAP framework improves forecasting accuracy, supports efficient energy utilisation, and contributes to sustainability in healthcare environments. Future work will explore real-time deployment and further integration with reinforcement learning to enable continuous optimisation.

E-commerce logistics networks are under increasing pressure to strike a balance between operational efficiency and service quality under unstable patterns of demand. The classical planning models treat facility location and vehicle routing as a set of independent optimization problems, which do not reflect important interdependencies between strategic network design and operational dispatch decisions. These decision layers are combined in a two-stage stochastic programming framework with distribution facilities in place before demand is realized, and operational flexibility is retained by taking recourse actions of dynamic routing decisions. The former is the determination of facility activation and the initial demand zone assignments in the face of uncertainty, and the latter is the adaptive construction of vehicle routes as the customer orders become known. Rolling horizon heuristics allow path replanning within operating windows, which can handle real-time arrivals of orders without the need to add too much computational load. Geospatial road network data provides realistic estimates of distances that accommodate real driving routes and city topography. The validation of simulation has shown that the efficiency in transportation, cost reduction, and responsiveness of service are significantly improved as compared to traditional centralized or static planning strategies. The combined framework is specifically useful where third-party logistics providers, online grocery delivery businesses, and business-to-consumer retailers are dealing with uncertain demand and managing distributed customer bases with strong demand expectations in terms of delivery time.

Building scalable user interfaces for high-demand systems requires a comprehensive approach that integrates architectural design principles, data management strategies, asynchronous processing patterns, and adaptive optimization techniques. This article examines the critical factors that enable user interfaces to maintain peak performance under varying loads and user volumes, addressing the challenge of preventing the interface layer from becoming a system-wide bottleneck. Through analysis of component-based architectures, distributed state management, and RESTful design patterns, the article establishes foundational principles for creating interfaces inherently prepared for growth. The article explores efficient data handling mechanisms, including pagination, lazy loading, virtual scrolling, and differential rendering, that transform data-heavy interfaces into streamlined experiences. Asynchronous processing and non-blocking operations are examined as essential techniques for maintaining responsiveness during resource-intensive processes, with particular attention to event-driven architectures, web workers, and optimistic update patterns. Performance optimization strategies for variable load conditions are investigated, including multi-layer caching, resource prioritization, adaptive quality reduction, connection pooling, and request batching. By integrating empirical research findings with practical implementation approaches, this article provides a holistic framework for understanding how theoretical principles of human-computer interaction translate into tangible design decisions that support scalability, ensuring consistent user experiences whether serving small user groups or massive concurrent populations across diverse usage scenarios and fluctuating demand patterns.

This article studies the item storage reassignment problem (ISRP) in the robotic mobile fulfilment system (RMFS). As the demand pattern evolves, the item storage assignment should be adapted accordingly to avoid potential workload. Hence, the item storage reassignment becomes an indispensable process of warehouse operations. This paper proposes an Interchange-Guided Sequencing Algorithm (IGSA) that identifies the interchange dependencies between item groups and constructs the corresponding group interchange sequence. Computational experiments demonstrate the effectiveness of the proposed algorithm and further analyse when reassignment should be performed under different item varieties and warehouse capacities.

Accurate production forecasting is essential for micro, small, and medium enterprises (MSMEs) to support effective production planning, inventory control, and decision-making. This study evaluates the performance of the Simple Moving Average (SMA) and Weighted Moving Average (WMA) methods in forecasting tela-tela production demand at MSME X using different historical period lengths. Production data from November 2023 to October 2024 were analyzed, and forecasting accuracy was assessed using the Mean Absolute Percentage Error (MAPE). The results indicate that forecasting accuracy varies depending on both the length of the moving average period and the weighting scheme applied. The WMA model with a 4-period window (n = 4) achieved the highest accuracy, producing the lowest MAPE value of 8.36%, which is classified as highly accurate. The SMA model with n = 4 also demonstrated good performance, with a MAPE value of 14.40%. Meanwhile, models employing longer historical periods (n = 8) yielded MAPE values of 16.20% for WMA and 19.82% for SMA, both falling within the good forecasting performance category but exhibiting lower responsiveness to recent demand changes. These findings highlight that shorter historical periods, when combined with appropriate weighting, can more effectively capture recent demand patterns in dynamic production environments. Accordingly, the WMA method with a 4-period window is recommended for MSME X as a reliable and accurate approach to support production planning, optimize resource allocation, and reduce the risk of overproduction or stock shortages.

Emergency supplies allocation is a critical task in post-disaster response, as ineffective or delayed decisions can directly lead to increased human suffering and loss of life. In practice, emergency managers must make rapid allocation decisions over multiple periods under incomplete information and highly unpredictable demand, making robust and adaptive decision support essential. However, existing allocation approaches face several challenges: (1) Those traditional approaches rely heavily on predefined uncertainty sets or probabilistic models, and are inherently static, making them unsuitable for multi-period, dynamically allocation problems; and (2) while reinforcement learning (RL) technique is inherently suitable for dynamic decision-making, most existing RL-base approaches assume fixed demand, making them unable to cope with the non-stationary demand patterns seen in real disasters. To address these challenges, we first establish a multi-period and multi-objective emergency supplies allocation problem with demand uncertainty and then formulate it as a two-player zero-sum Markov game (TZMG). Demand uncertainty is modeled through an adversary rather than predefined uncertainty sets. We then propose RESA, a novel RL framework that uses adversarial training to learn robust allocation policies. In addition, RESA introduces a combinatorial action representation and reward clipping methods to handle high-dimensional allocations and nonlinear objectives. Building on RESA, we develop RESA_PPO by employing proximal policy optimization as its policy optimizer. Experiment results with realistic post-disaster data show that RESA_PPO achieves near-optimal performance, with an average gap of only 3.7% in terms of the objective value of the formulated problem, from the theoretical optimum derived by exact solvers. Moreover, RESA_PPO outperforms all baseline methods, including heuristic and standard RL methods, by at least 5.25% on average.

Accurate pharmaceutical demand forecasting is essential to ensure timely drug availability, reduce inventory costs, and improve operational efficiency in healthcare supply chains. However, existing statistical, machine learning, and deep learning approaches often struggle to capture the nonlinear and dynamic demand patterns arising from drug substitutions, comorbidity treatments, and seasonal disease fluctuations. To address this challenge, we propose KG-GCN-LSTM, a novel hybrid model that integrates a pharmaceutical knowledge graph (KG) with deep learning techniques. A clipped Graph Convolutional Network (GCN) is employed to extract feature representations from both the historical demand of the target drug and the related drugs encoded in the knowledge graph. The outputs of the GCN are subsequently processed by a Long Short-Term Memory (LSTM) network to capture temporal dynamics in drug demand. Experiments on real-world pharmacy sales data demonstrate that KG-GCN-LSTM consistently outperforms established benchmarks-including ARIMA, SVR, XGBoost, RNN, CNN-LSTM, TimeMixer and NBEATS, achieving a 3.62% reduction in Symmetric Mean Absolute Percentage Error (SMAPE) relative to NBEATS, while delivering performance comparable to the state-of-the-art TimeMixer. These results highlight the effectiveness of knowledge graph-enhanced deep learning in improving the accuracy and robustness of pharmaceutical demand forecasting, which can support data-driven decision-making in healthcare supply chain management.

Real-time AI-based inventory optimization is transforming large enterprises by enhancing efficiency, reducing costs, and meeting customer demands effectively. SAP solutions provide a robust platform for implementing AI-driven inventory management, enabling organizations to make data-driven decisions quickly. This study explores the integration of AI with SAP for inventory optimization, highlighting benefits such as real-time demand forecasting, automated replenishment, and inventory reduction. Leveraging predictive analytics, machine learning models, and IoT data, the AI-based approach aligns inventory levels with actual demand patterns, optimizing stock levels and minimizing wastage. The research discusses methodologies, tools, and implementation strategies, and analyzes real-world results from large enterprises to present a comprehensive view of the impact and potential of real-time AI-based inventory optimization in SAP environments.

Multi-scale Spatio-temporal graph neural network for enhanced water demand forecasting.

Decomposition of Water Demand Patterns Using Skewed Gaussian Distributions for Behavioral Insights and Operational Planning

Energy management (EM) for grid-isolated electric vehicle (EV) charging with microgrids (MGs) focuses on optimising energy consumption and integrating renewable sources to ensure efficient power distribution. Challenges such as high operational costs from advanced energy storage solutions and reduced efficiency due to power fluctuations and demand mismatches can hinder performance. To address these issues, a hybrid approach combining Genghis Khan Shark Optimiser (GKSO) and Similarity Navigated Graph Neural Network (SNGNN), termed GKSO-SNGNN, is proposed. The purpose of this method is to make the best use of renewable energy sources (RESs), reduce energy losses and operating expenses, and increase the overall system efficiency. GKSO optimises energy distribution, resource allocation and charging schedules, while SNGNN predicts energy demand patterns and charging station (CS) utilisation. Compared with algorithms like Gannet Optimisation Algorithm-Tree Hierarchical Deep Convolutional Neural Network (GOA-THDCNN), Deep Neural Network (DNN), Red Panda Optimisation-Adaptive Drop Block-Enhanced Generative Adversarial Networks (RPO-ADGAN), Young's Double-Slit Experiment Optimiser-THDCNN (YDSE-THDCNN) and Golden Jackal Optimisation-Attention Pyramid Convolutional Neural Network (GJO-APCNN) in MATLAB simulations, the GKSO-SNGNN approach achieves 98.7% efficiency, reduces computation time by 23% on average and lowers operational cost to 1502 cents per run, demonstrating both faster processing and improved cost-efficiency.