Search Results (7,348)

Global warming is recognized as a climate crisis that extends beyond a mere increase in the Earth’s temperature, triggering rapid and widespread climatic changes worldwide. In particular, the frequency and intensity of extreme rainfall events have increased in Korea and the Association of Southeast Asian Nations (ASEAN) region, leading to a significant increase in flood damage. The growing number of large-scale hydrological disasters underscores the urgent need for accurate and rapid flood-forecasting systems that can support disaster preparedness and mitigation. Compared with conventional physics-based forecasting systems, artificial intelligence (AI) models can provide faster predictions using limited observational data. In this study, a river water-level prediction model was constructed using real-time observation data and a long short-term memory (LSTM) algorithm, which is a recurrent neural network-based deep learning approach suitable for hydrological time-series forecasting. A repeated k-fold cross-validation technique was applied to enhance model generalization and prevent overfitting. In addition, water-level differencing was employed to convert nonstationary water-level data into stationary time-series inputs, thereby improving the prediction stability. Water-level observation stations in the Philippines, Indonesia, and the Republic of Korea were selected as study sites, and the model performance was evaluated at each location. The differenced LSTM model achieved a root mean square error of 0.13 m, coefficient of determination (R²) of 0.866, Nash–Sutcliffe efficiency (NSE) of 0.844, and Kling–Gupta efficiency of 0.893, thus outperforming the non-differenced baseline by approximately 17%. The repeated k-fold validation approach was particularly effective when the training data period was short or the number of input variables was limited. These results confirm that ensuring temporal stationarity and applying repeated cross-validation can significantly enhance the predictive accuracy of real-time flood forecasting. The proposed framework exhibits strong potential for implementation in regional early warning systems across data-limited flood-prone areas in the ASEAN region. Ongoing studies that apply and verify this approach in diverse hydrological contexts are expected to further improve and expand AI-based flood prediction models. Full article

The electro-hydraulic hybrid system has emerged as a critical technology in new energy vehicles, owing to the remarkable power density and efficient energy regeneration capabilities of hydraulic technology, coupled with the high energy density of electric power. This system effectively enhances vehicle range and battery life. We developed an energy management strategy (EMS) for the electro-hydraulic hybrid system (EHHS) to ensure smooth energy conversion, while ensuring the full utilization of electrical and hydraulic energy within a reasonable and efficient range. To enhance the system’s overall performance, it is imperative to address pivotal technologies, including power coupling and energy management. In this research, the structure of an electro-hydraulic hybrid vehicle (EHHV) is classified, compared and discussed. The application of existing EHHVs is studied. Subsequently, an analysis and summary are conducted on the current status and development trends of EMSs and collaborative operation control strategies (COCSs), and a novel mechanical-electro-hydraulic power-coupled system (MEHPCS) is put forward that successfully converts mechanical, electrical, and hydraulic energy in performance. Simultaneously, other applications of the system are forecasted. Finally, some suggestions for the electro-hydraulic hybrid systems’ future development are made. This study can promote the development of sustainable transportation technologies. The system integrates mechanical engineering, control theory, and environmental science, enabling interdisciplinary methodological innovation. In addition, relevant studies provide data support for policy makers by quantifying energy consumption indicators. Full article

Long-term traffic flow prediction (LTFP) is crucial for intelligent transportation systems but remains challenging due to complex spatiotemporal dependencies and multi-scale temporal patterns. While recent models like Autoformer have introduced decomposition techniques, they often lack tailored mechanisms for traffic data′s unique characteristics, such as strong periodicity and long-range spatial correlations. To address this gap, we propose STLLformer, a novel spatiotemporal Transformer that establishes a seasonal-dominated, multi-component collaborative forecasting paradigm. Unlike existing approaches that merely combine decomposition with graph networks, STLLformer features: (1) a dedicated encoder–decoder architecture for separate yet synergistic modeling of trend, seasonal, and residual components; (2) a seasonal-driven autocorrelation mechanism that purely captures cyclical patterns by filtering out trend and noise interference; and (3) a low-rank graph convolutional module specifically designed to capture dynamic, long-range spatial dependencies in road networks. Experiments on two real-world traffic datasets (PEMSD8 and HHY) demonstrate that STLLformer outperforms strong baseline methods (including LSTGCN, LSTM, and ARIMA), achieving an average improvement of over 10% in MAE and RMSE (e.g., on PEMSD8 for 6-h prediction, MAE drops from 36.87 to 30.34), with statistical significance (p < 0.01). This work provides a more refined and effective decomposition-fusion solution for traffic forecasting, which holds practical promise for enhancing urban traffic management and alleviating congestion. Full article

The growing integration of plug-in electric vehicles (PEVs) into microgrids presents both challenges and opportunities, particularly through vehicle-to-grid (V2G) services. This paper proposes a dynamic horizon optimization (DHO) framework with adaptive pricing for real-time scheduling of PEVs in a renewable-powered microgrid. The system integrates solar and wind energy, V2G capabilities, and time-of-use (ToU) tariffs. The DHO strategy dynamically adjusts control horizons based on forecasted load, generation, and electricity prices, while considering battery health. A PEV-specific pricing scheme couples ToU tariffs with system marginal prices. Case studies on a microgrid with four heterogeneous EV charging stations show that the proposed method reduces peak load by 23.5%, lowers charging cost by 12.6%, and increases average final SoC by 12.5%. Additionally, it achieves a 6.2% reduction in carbon emissions and enables V2G revenue while considering battery longevity. Full article

Age-related macular degeneration (AMD) is a leading cause of irreversible blindness, yet current tools for forecasting treatment outcomes remain limited by either the opacity of deep learning or the rigidity of rule-based systems. To address this gap, we propose a hybrid neuro-symbolic and large language model (LLM) framework that combines mechanistic disease knowledge with multimodal ophthalmic data for explainable AMD treatment prognosis. In a pilot cohort of ten surgically managed AMD patients (six men, four women; mean age 67.8 ± 6.3 years), we collected 30 structured clinical documents and 100 paired imaging series (optical coherence tomography, fundus fluorescein angiography, scanning laser ophthalmoscopy, and ocular/superficial B-scan ultrasonography). Texts were semantically annotated and mapped to standardized ontologies, while images underwent rigorous DICOM-based quality control, lesion segmentation, and quantitative biomarker extraction. A domain-specific ophthalmic knowledge graph encoded causal disease and treatment relationships, enabling neuro-symbolic reasoning to constrain and guide neural feature learning. An LLM fine-tuned on ophthalmology literature and electronic health records ingested structured biomarkers and longitudinal clinical narratives through multimodal clinical-profile prompts, producing natural-language risk explanations with explicit evidence citations. On an independent test set, the hybrid model achieved AUROC 0.94 ± 0.03, AUPRC 0.92 ± 0.04, and a Brier score of 0.07, significantly outperforming purely neural and classical Cox regression baselines (p ≤ 0.01). Explainability metrics showed that >85% of predictions were supported by high-confidence knowledge-graph rules, and >90% of generated narratives accurately cited key biomarkers. A detailed case study demonstrated real-time, individualized risk stratification—for example, predicting an >70% probability of requiring three or more anti-VEGF injections within 12 months and a ~45% risk of chronic macular edema if therapy lapsed—with predictions matching the observed clinical course. These results highlight the framework’s ability to integrate multimodal evidence, provide transparent causal reasoning, and support personalized treatment planning. While limited by single-center scope and short-term follow-up, this work establishes a scalable, privacy-aware, and regulator-ready template for explainable, next-generation decision support in AMD management, with potential for expansion to larger, device-diverse cohorts and other complex retinal diseases. Full article

Under the dual carbon goals, microgrids face significant challenges in managing multi-energy flow coupling and maintaining operational robustness with high renewable energy penetration. This paper proposes a novel dynamic pressure-aware spatiotemporal optimization dispatch strategy. The strategy is centered on intelligent energy storage and enables proactive energy allocation for critical pressure moments. We designed and validated the strategy under an ideal benchmark scenario with perfect foresight of the operational cycle. This approach demonstrates its maximum potential for spatiotemporal coordination. On this basis, we propose a Multi-Objective Self-Adaptive Hybrid Enzyme Optimization (MOSHEO) algorithm. The algorithm introduces segmented perturbation initialization, nonlinear search mechanisms, and multi-source fusion strategies. These enhancements improve the algorithm’s global exploration and convergence performance. Specifically, in the ZDT3 test, the IGD metric improved by 7.7% and the SP metric was optimized by 63.4%, while the best HV value of 0.28037 was achieved in the UF4 test. Comprehensive case studies validate the effectiveness of the proposed approach under this ideal setting. Under normal conditions, the strategy successfully eliminates power and thermal deficits of 1120.00 kW and 124.46 kW, respectively, at 19:00. It achieves this through optimal quota allocation, which involved allocating 468.19 kW of electricity at 13:00 and 65.78 kW of thermal energy at 18:00. Under extreme weather, the strategy effectively converts 95.87 kW of electricity to thermal energy at 18:00. This conversion addresses a 444.46 kW thermal deficit. Furthermore, the implementation reduces microgrid cluster trading imbalances from 1300 kW to zero for electricity and from 400 kW to 176.34 kW for thermal energy, significantly enhancing system economics and multi-energy coordination efficiency. This research provides valuable insights and methodological support for advanced microgrid optimization by establishing a performance benchmark, with future work focusing on integration with forecasting techniques. Full article

Reliable load forecasting is crucial for ensuring optimal dispatch, grid security, and cost efficiency. To address limitations in prediction accuracy and generalization, this paper proposes a hybrid model, GRU-MHSAM-ResNet, which integrates a gated recurrent unit (GRU), multi-head self-attention (MHSAM), and a residual network (ResNet)block. Firstly, GRU is employed as a deep temporal encoder to extract features from historical load data, offering a simpler structure than long short-term memory (LSTM). Then, the MHSAM is used to generate adaptive representations by weighting input features, thereby strengthening the key features. Finally, the features are processed by fully connected layers, while a ResNet block is added to mitigate gradient vanishing and explosion, thus improving prediction accuracy. The experimental results on actual load datasets from systems in China, Australia, and Malaysia demonstrate that the proposed GRU-MHSAM-ResNet model exhibits superior predictive accuracy to compared models, including the CBR model and the LSTM-ResNet model. On the three datasets, the proposed model achieved a mean absolute percentage error (MAPE) of 1.65% (China), 5.52% (Australia), and 1.57% (Malaysia), representing a significant improvement over the other models. Furthermore, in five repeated experiments on the Malaysian dataset, it exhibited lower error fluctuation and greater result stability compared to the benchmark LSTM-ResNet model. Therefore, the proposed model provides a new forecasting method for power system dispatch, exhibiting high accuracy and generalization ability. Full article

With rising energy demand and the need for sustainable waste treatment, anaerobic digestion (AD) has emerged as a key technology for converting organic residues into renewable energy. However, predicting methane yield in full-scale facilities remains challenging due to the complexity of AD processes, the variability of feedstocks, and the impracticality of frequent biochemical methane potential (BMP) testing. In this study, we developed a simple, data-driven approach to forecast methane production in a commercial-scale digester co-digesting manure and food waste. The model employs weekly cumulative BMP of feedstock mixtures, calculated from literature values, as the explanatory variable. The model achieved an R² of 0.70 and a forecast mean absolute percentage error (MAPE) of 7.4, indicating its potential for full-scale AD prediction. Importantly, the analysis revealed a long-run equilibrium between BMP and methane yield, with deviations corrected within roughly one month—closely matching the system’s hydraulic retention time. These findings demonstrate that literature-based BMP values can be used to reliably predict methane yield in operating AD systems, offering a low-cost and scalable tool to support decision-making in waste management and biogas plant operations. Full article

Air pollution forecasting remains a critical challenge for urban public health management, with traditional approaches struggling to balance accuracy and interpretability. This study introduces a novel PM₁₀ forecasting framework combining physics-informed feature engineering with interpretable ensemble fusion using the Choquet integral, the first application of this non-linear aggregation operator for air quality forecasting. Using hourly data from 11 monitoring stations in Budapest (2021–2023), we developed four specialized feature sets capturing distinct atmospheric processes: short-term dynamics, long-term patterns, meteorological drivers, and anomaly detection. We evaluated machine learning models including Random Forest variants (RF), Gradient Boosting (GBR), Support Vector Regression (SVR), K-Nearest Neighbors (KNN), and Long Short-Term Memory (LSTM) architectures across six identified pollution regimes. Results revealed the critical importance of feature engineering over architectural complexity. While sophisticated models failed when trained on raw data, the KNN model with 5-dimensional anomaly features achieved exceptional performance, representing an 86.7% improvement over direct meteorological input models. Regime-specific modeling proved essential, with GBR-Regime outperforming GBR-Stable by a remarkable effect size. For ensemble fusion, we compared the novel Choquet integral approach against conventional methods (mean, median, Bayesian Model Averaging, stacking). The Choquet integral achieved near-equivalent performance to state-of-the-art stacking while providing complete mathematical interpretability through interaction coefficients. Analysis revealed predominantly redundant interactions among models, demonstrating that sophisticated fusion must prevent information over-counting rather than merely combining predictions. Station-specific interaction patterns showed selective synergy exploitation at complex urban locations while maintaining redundancy management at simpler sites. This work establishes that combining domain-informed feature engineering with interpretable Choquet integral aggregation can match black-box ensemble performance while maintaining the transparency essential for operational deployment and regulatory compliance in air quality management systems. Full article

This study introduces an IoT-enabled smart greenhouse system tailored for rice cultivation and designed as a controlled experimental platform to evaluate fertilizer application methods. Traditional greenhouse farming often struggles with unpredictable weather, pest infestations, and inefficient resource use. To overcome these challenges, the proposed system optimizes environmental conditions and enables precise monitoring and control through the Thingsboard IoT platform, which tracks temperature, humidity, and sunlight intensity in real time. The cultivation process involved Inceptisol soil preparation, slurrying, fertilization, seeding, transplantation, and continuous monitoring. The novelty lies in its dual-purpose design, enabling both cultivation and structured agronomic experimentation under identical environmental conditions. The system enables both rice cultivation and comparative testing of nano-silica fertilizer applied via root (soil) and foliar (leaf) methods. Automated temperature control (maintaining 20–36.5 °C) and humidity regulation (10–85% RH) with a mist blower response time under 5 s ensured consistent conditions. Sensor accuracy was validated with deviations of 0.4% (±0.11 °C) and 0.77% (±0.5% RH). Compared to conventional setups, this system achieved superior environmental stability and control precision, improving experimental reproducibility. Its integration potential with machine learning models opens new possibilities for forecasting plant responses based on historical data. Overall, the study demonstrates how advanced technology can enhance agricultural precision, sustainability, and research reliability. Full article

As demand for heat and power continues to grow, production planning of a combined heat and power (CHP) system becomes one of the most crucial optimization problems. Due to the fluctuations in demand and production costs of heat and power, it is necessary to quickly solve the production planning problem of the contemporary CHP system. In this paper, we propose a Hybrid Time series Informed neural Network (HYTIN) in which, a deep learning-based planner for CHP production planning predicts production levels for heat and power for each time step. Specifically, HYTIN supports inventory-aware decisions by utilizing a long short-term memory network for heat production and a convolutional neural network for power production. To verify the effectiveness of the proposed method, we build ten independent test datasets of 1200 h each with feasible initial states and common limits. Experimentation results demonstrate that HYTIN achieves lower operation cost than the other baseline methods considered in this paper while maintaining quick inference time, suggesting the viability of HYTIN when constructing production plans under dynamic variations in demand in CHP systems. Full article

Renewable energy sources provide environmental and economic benefits by replacing conventional energy sources. In Korea, photovoltaic (PV) systems are increasingly deployed in apartment complexes and residential buildings. In self-consumption PV systems, surplus generation exceeding local demand often leads to a reverse power flow. This phenomenon becomes more frequent in microgrid environments where multiple distributed energy resources are interconnected. Accordingly, inverter control strategies based on generation forecasting have emerged as critical challenges. In this paper, we propose an on-device artificial intelligence model for inverter control that integrates net power forecasting with time-series neural networks. Two novel forecasting methods were proposed and introduced: Prediction-to-Prediction (P–P) and Net-Power Prediction (N–P). Various neural network models were trained and evaluated using multiple performance metrics. A novel threshold adjustment mechanism based on the mean absolute error was designed for inverter control. The control scenarios were analyzed by comparing the actual power losses with the forecast-based power losses, and the energy savings were quantified by adjusting the correction factor. The proposed forecasting methods achieved a reduction of approximately 40–70% in energy losses compared with the actual loss levels. The threshold adjustment strategy enhances flexibility in balancing the number of on/off switching events and the power loss, contributing to improved energy efficiency and system stability. Full article

Guizhou Province in China exhibits a distinctive agroforestry mosaic landscape with frequent human activity in forested areas. This region experiences recurrent forest fires, characterized by significant difficulties in suppression and high risks. Research on the prediction of forest fire occurrences holds crucial practical significance in terms of enhancing regional forest fire prevention capabilities. However, the current fire risk forecasting methods in the area consider only meteorological factors, neglecting firebrands and fuel conditions, which results in deviations between forecasted and actual fire occurrences. Therefore, this study proposes a novel fire occurrence prediction method that utilizes the ignition component (IC) from the National Fire Danger Rating System (NFDRS) to characterize the weather–fuel complex while integrating the firebrand occurrence probability to construct a predictive model. The applicability and accuracy of this method are also evaluated. The results show that, firstly, the probability of at least one daily forest fire occurrence in the study area can be expressed as a nonlinear function based on the IC. Secondly, as time progresses, the correlation between the forest fire occurrence probability and the IC shows a decreasing trend, although the differences across different time spans are not statistically significant. Thirdly, when a 5-year time span is adopted, the error in calculating the forest fire occurrence probability based on the IC is significantly lower than at other time spans. Finally, a predictive model for the forest fire occurrence probability based on the IC is established, where P = (100*IC)/(4.06 + IC), with a mean absolute error (MAE) of 4.83% and mean relative error (MRE) of 14.87%. Based on this research, the IC enables the calculation of forest fire occurrence probabilities, assessment of fire risk ratings, and guidance for fire preparedness and planning. This work also provides theoretical support and a methodological reference for conducting forest fire probability studies in other regions. Full article

The Dvorak technique has represented a fundamental tool for understanding the power of tropical cyclones based on their shape and geometric evolution. However, it should be noted that the Dvorak technique is purely morphological in nature and was developed for wind, not precipitation. The role of shape methods in precipitation prediction remains uncertain, particularly in the context of modern multi-sensor capabilities. This uncertainty forms the motivation for the present study. In an attempt to enrich Dvorak’s technique, this study proposes a novel hypothesis. This study tests the hypothesis that higher precipitation intensity is associated with more organized cloud-system morphology, as captured by simple geometric descriptors and indicative of dynamically coherent convection. A total of 3419 cloud-system objects (after size filter) were utilized to establish geometric relationships in each of them. For the case study of Hurricane Patricia over the Mexican coast in 2015, 3858 geometric shapes were processed. The cloud-system morphology was derived from geostationary imagery (GOES-13) and collocated with satellite precipitation estimates in order to isolate intense-rainfall objects (>50 mm/h). For each object, simple geometric descriptors were computed, and shape variability was summarised via Principal Component Analysis (PCA). The present study sought to evaluate the associations with rain-rate metrics (mean, mode, maximum) using rank correlations and k-means clustering. Furthermore, sensitivity analyses were conducted on the rain threshold and minimum object size. A Shape Descriptor: ratio between perimeter and diameter was identified as a promising tool to enhance early prediction models of extreme rainfall, contributing to enhanced meteorological risk management. The study indicates that cloud shape can serve as a valuable indicator in the classification and forecasting of intense cloud systems. Full article

The growing complexity of modern building systems requires advanced monitoring frameworks to improve fault detection, energy efficiency, and operational resilience. Digital Twin (DT) technology, which integrates real-time data with virtual models of physical systems, has emerged as a promising enabler for predictive diagnostics. Despite growing interest, key challenges remain, including the neglect of short- and long-term forecasting across different scenarios, insufficiently robust data preparation, and the rare validation of models on multi-zone buildings over extended test periods. To address these gaps, this study presents a comprehensive DT-enabled framework for predictive monitoring and anomaly detection, validated in a multi-zone educational building in Rhode Island, USA, using a full year of operational data for validation. The proposed framework integrates a robust data processing pipeline and a comparative analysis of machine learning models, including LSTM, RNN, GRU, ANN, XGBoost, and RF, to forecast short-term (1 h) and long-term (24 h) indoor temperature variations. The LSTM model consistently outperformed other methods, achieving R² > 0.98 and RMSE < 0.55 °C for all tested rooms. For real-time anomaly detection, we applied the hybrid LSTM–Interquartile Range (IQR) method on one-step-ahead residuals, which successfully identified anomalous deviations from expected patterns. The model’s predictions remained within a ±1 °C error margin for over 90% of the test data, providing reliable forecasting up to 16 h ahead. This study contributes a validated, generalizable DT methodology that addresses key research gaps, offering practical tools for predictive maintenance and operational optimization in complex building environments. Full article