查看更多>>摘要：Accurate short-term load forecasting (STLF) is crucial for the operational stability and efficiency of modern energy systems, particularly in markets experiencing increasing complexity. Traditional statistical methods, despite their robustness, often struggle to model multifrequency seasonality. This paper introduces a novel forecasting model, the decomposed Fourier ARIMA (FARIMA), designed to address multifrequency seasonal patterns in time series data with a specific application in this study to electricity consumption data. The FARIMA model combines Fourier decomposition to isolate periodic components, polynomial regression to capture trends, and ARIMA to model the residuals. The study evaluates optimal training periods and benchmarks FARIMA's performance against mature traditional methods, specifically SARIMA and Holt-Winters, which are widely used in short-term load forecasting. Using Greek energy market data, FARIMA consistently outperformed SARIMA and Holt-Winters across different setups. For a one-year training period, it achieved a MAPE that was 1.59 percentage points lower than SARIMA's and 1.44 percentage points lower than Holt-Winters'. For a six-month training period, FARIMA achieved a MAPE that was 0.17 percentage points lower than SARIMA's and 3.33 percentage points lower than Holt-Winters'. Additionally, FARIMA demonstrated significant computational efficiency, achieving a runtime reduction of 98.5 % compared to SARIMA in both setups, due to its ability to simplify residual signals and use less complex ARIMA parameters. Results demonstrate FARIMA's superior forecasting accuracy, specifically for one-week-ahead predictions, with lower Mean Absolute Percentage Error (MAPE) and Root Mean Squared Error (RMSE) compared to conventional models. The FARIMA model bridges gaps in traditional forecasting by effectively capturing multifrequency patterns. While its practical implications primarily focus on short-term load forecasting, for which it was specifically developed, FARIMA is fundamentally a new time series model. As such, it holds potential for application across a wide range of domains involving multifrequency time series data.
查看更多>>摘要：The adoption of renewable energy sources requires solutions to reduce the temporal mismatch between power demand and supply. Buildings’ energy demand shifting can mitigate the problem. The objective of this paper is to quantify the time-dependent energy flexibility provided by the building’s thermal mass. This flexibility is highly dynamic, as it is influenced by temperature levels, environmental conditions and various disturbances. This time-evolving evaluation is preceded by a preliminary assessment of the energy flexibility, through the calculation of the available storage capacity and storage efficiency, aimed at evaluating the performance of the heat storage process in the building’s thermal mass. Flexibility is quantified for two heating systems: radiator and floor heating. The results show that the case with a floor heating system recovers 50% of the heat stored in the thermal mass after one day, while the case with a radiator takes 3 days to achieve 50% heat recovery. Finally, time-dependent flexibility is almost constant for the radiator case and the average flexible power is up to 12.3 W∕m~2, while for the floor heating system it is variable along the day and the maximum flexible power varies between 24.6 W∕m~2 at midday and 86.2 W∕m~2 at night.
查看更多>>摘要：State of health (SOH) estimation is essential to improve reliability of battery. However, it is a challenging issue to develop an effective health prognostics model due to the lack of run-to-failure data in the industrial scenarios. In this paper, a prognosability regularized generative adversarial network (ProGAN) is proposed to implement SOH estimation of battery. Firstly, a prognosability regulator is proposed in ProGAN to generate the samples with good trendability over time. Secondly, a multi-level wavelet decomposition long short-term memory (MwLSTM) is proposed to generate samples from low and high frequency components. Finally, an SOH estimator is constructed based on the synthetic samples. Compared to existing generative models, ProGAN introduces a novel prognosability regulator that enhances the trendability of generated samples. Additionally, a MwLSTM is developed by considering both low and high frequency components. Three battery SOH estimation cases are adopted to verify the effectiveness of ProGAN. The experimental results indicate that ProGAN can not only generate the samples with high consistency in space (i.e., data distribution), but also have good trendability over time (i.e., prognosability). Moreover, ProGAN has a better performance in both data augmentation and SOH estimation than other generative models.
查看更多>>摘要：Extended Locational Marginal Pricing (eLMP) integrates start-up costs into market clearing, enhancing participant incentives compared to classical Locational Marginal Pricing (LMP). Despite eLMP’s operational complexity, we propose a two-stage transfer learning strategy that overcomes perceived prediction challenges. First, an LMP predictor is trained by decomposing historical data into distinct modes via Variational Mode Decomposition (VMD). Three customized deep learning architectures and an improved loss function optimize training for each mode, with predictions aggregated nonlinearly through a learning ensemble. Second, the pre-trained model transfers to eLMP by decomposing new pricing data into analogous modes, testing mode compatibility, and selectively fine-tuning mismatched modes alongside the ensemble. Experiments on three Midcontinent ISO (MISO) sites demonstrate that adjusting at most one mode achieves more than 95% prediction accuracy across datasets, surpassing direct eLMP forecasting by more than 50% in accuracy with minimal computational overhead. Theoretical error bounds align with empirical results, confirming nearoptimal transfer efficiency. This strategy bridges LMP and eLMP forecasting frameworks, proving that eLMP’s complexity can be practically addressed without exhaustive retraining. By reusing mode-specific knowledge from LMP and enabling targeted fine-tuning, our approach reduces deployment costs while maintaining high generalizability, offering a scalable solution for future electricity market designs.
查看更多>>摘要：The wireless power transfer (WPT) technology has significantly transformed the transportation industry over the past decade, particularly in applications like electric vehicles. However, one major challenge remains: maintaining optimal energy transfer efficiency despite misalignment between the charging coils. This study investigates the impact of misalignment on energy coupling between coils and proposes a low-cost solution to address the issue. Three coil geometries-Square, Pentagon, and Double-D-are analyzed to determine the most efficient and stable configuration under misalignment conditions. The results indicate that the pentagon coil achieves over 99 % energy transfer efficiency, with tolerances of up to 15◦ angular and 2 cm horizontal misalignment. Furthermore, the pentagon coil exhibits a high degree of stability, with only a 5 % variation in the coupling coefficient under misaligned conditions, making it a promising solution for robust WPT systems in transportation.
查看更多>>摘要：Thermoacoustic engines (TAEs) are promising energy conversion technologies due to their absence of moving parts, flexibility, and environmental friendliness. The driver of such an engine is the thermoacoustic core (TAC). In this study, we propose a framework that integrates CFD simulations, a surrogate model based on an artificial neural network (ANN), and deep reinforcement learning (DRL) to optimize the channel shape in the TAC of a supercritical CO_2 TAE. CFD simulations generate a dataset for the surrogate model. The surrogate model demonstrates exceptional generalization capability (R~2 = 0.992) and computational efficiency (within 3.8 ms per prediction), enabling fast reward evaluation during the DRL optimization. The TD3 algorithm is employed to explore the continuous design space. The optimized channel achieves a pressure amplitude of 0.663 Mpa, an 8.51% improvement compared to the original straight channel, which can be attributed to the enhanced heat transfer matching between the hot heat exchanger and the ambient one. This study demonstrates the potential of combining ANN-based surrogate models with DRL for optimizing thermoacoustic devices. The proposed framework is adaptable for optimizing other thermal systems and casts light on integrating artificial intelligence with physical modeling for engineering optimization.
查看更多>>摘要：The lean-premixed swirl combustor with slope confinement (SC) exhibits significant potential in attenuating thermoacoustic instability compared with traditional dump confinement (DC). In this study, we scrutinize the stability origin from the perspectives of flow stability and complex network analysis. First, the dominant temporal and absolute growth rates of SC are respectively reduced by 68.0% and 86.6% compared to DC. These reductions quantitatively demonstrate the superior capability of SC in suppressing temporal and absolute instability. Then, we propose a backflow ratio to distinguish the absolute instabilities in two cases. The stability enhancement in SC is characterized by a backflow ratio less than a critical value of 2, which stems from a less pronounced flow recirculation through eliminating the corner flow. Additionally, the flow complex network changes from a dual-ring structure in the DC case to a straight-chain structure in the SC case. The weighted closeness centrality measurement implies that the corner recirculation zone (CRZ) is responsible for self-excited flow oscillation. The network without CRZ features more difficult disturbance propagation, which is further conducive to mitigating instability.
Marco GerolaFrancesco CecinatoJacco K. HaasnootPhilip J. Vardon...
135990.1-135990.16页
查看更多>>摘要：Energy quay walls (EQWs) are an innovative type of energy geostructure (EG), capable of exchanging thermal energy with both soil and open water. In this work, a validated 3D finite element numerical model is employed to conduct a parametric analysis aimed at identifying the most important design- and site-dependent parameters for optimising EQW energy performance. The Taguchi Experimental Design statistical method is employed to explore the parameter space for two types of heat exchanger loops used in EQW installations: loops incorporated into the structural elements and add-on panels. The most influential design parameter on the energy performance is shown to be the number of U-loops, which can significantly improve the energy yield (up to ∼50%). The effects of reduced inlet temperature (up to ∼35%), enlarged pipe cross-sectional area (up to ∼26%) and increased heat exchanger fluid velocity (up to ∼20%) are also significant for the EQW thermal performance. Among site-specific factors, the presence of a deep water body (up to ∼100%) with high temperature (up to ∼62%) is confirmed essential for achieving high energetic performance, while a high open water flow velocity (up to ∼32%) and elevated soil thermal conductivity (up to ∼23%) are influential in the short-term thermal output.
查看更多>>摘要：Deep reinforcement learning is a potential method for online energy management of electric vehicles. However, deep reinforcement learning can suffer from limited training data and underfitting/overfitting characteristics, leading to a compromised vehicle economy in online applications. This paper proposes an online updating approach for deep reinforcement learning-based energy management in electric vehicles with battery/supercapacitor hybrid energy management systems. Firstly, Jensen-Shannon Divergence is introduced to compare the pre-training and online velocity transfer probability surfaces quantitatively and to trigger the online strategy updating when exceeding the threshold. Secondly, existing online driving data is optimized through dynamic programming to construct the optimal state/action dataset, which will be used for actornetwork updating. Thirdly, a soft updating method is further proposed to ensure smooth strategy updating for satisfactory generalization ability based on the weighted fusion of multiple networks. Results under Dallas realworld driving cycles validate that the proposed online updating method can reduce the battery degradation and vehicle operation cost by 5.3-10.2% and 4.5-9.0%, respectively, compared with the strategy before updating. This study provides a meaningful attempt at the online updating of deep reinforcement learning-based energy management strategies, which can be easily extended to different vehicle types and driving conditions.
查看更多>>摘要：Condition monitoring are critical for ensuring the long-term stability and efficiency of equipment operations. In particular, under extreme conditions, the number of sensors is often severely limited, resulting in extremely sparse sensor data. This scarcity renders it challenging to obtain interpretable high-dimensional physical information in real-time. Many methods for condition monitoring predominantly rely on sensor data analysis, such as nonlinear fitting, which often lack physical interpretability. Hyper projection-based reduced order models (HPROMs) incorporating the physics, provide strong physical interpretability and high dimensional physical field real-time computing capability. However, HPROMs strictly adhere to forward calculation procedures because of approximation process of intrusive operators. To address these challenges, this paper introduces a novel physicssensing framework (PSF) driven by a non-intrusive, inverse, hyper projection-based reduced order model (NIIHPROM) with extremely sparse sensor data. The NII-HPROM circumvents the approximation process of intrusive operators, enabling direct inverse computation of physical fields at hyper-reduced speeds. Moreover, the PSF incorporates a reliability evaluation system (RES), a physical noise-filtering system (PNFS), and an abnormal condition identification system (ACIS), not only offering a comprehensive solution but also ensuring reliable evaluation, noise filtering, and fault identification for high-temperature components. In this study, the PSF is applied to an industrial high-temperature component, the rotor, using only two sensors to achieve rapid inverse nonlinear temperature field calculations, which are 579.6 times faster than HPROM forward iterative computations and 17,500 times faster than full-order model forward iterative calculations.
NETL
NSTL
Elsevier