Pub Date : 2026-06-01Epub Date: 2026-01-24DOI: 10.1016/j.epsr.2026.112778
Sami Astan , Kavan Fatehi , Amin Hajizadeh
The paper introduces CCS-CoLearn, an integrated cooperative learning framework for the predictive and equitable operation of shared carbon capture and storage (CCS) infrastructures. The methodology unifies probabilistic deep-learning forecasting with game-theoretic optimisation to coordinate multiple industrial emitters linked to a common transport-and-storage system. A Transformer–BiLSTM ensemble generates multi-quantile forecasts of CO₂ inflows, parasitic electrical loads, and market prices, which feed a Nash-bargaining coordination layer ensuring fairness and Pareto efficiency among participants. The system is implemented within a digital-twin simulation of the Liverpool Bay CCS network, coupled to a reduced Great Britain power-grid model. Scenario analyses for 2030 and 2050 show that CCS-CoLearn achieves up to 18 % total-cost reduction, fairness index > 0.9, and 10 % higher CO₂ abatement compared with non-cooperative or rule-based baselines. The framework demonstrates that intelligent coordination can substitute for physical over-capacity, providing a scalable pathway for data-driven and cooperative management of CCS clusters under the UK Net-Zero 2050 strategy.
{"title":"CCS-CoLearn: A cooperative learning framework for shared carbon capture and storage infrastructure — Application to Liverpool bay and the North-West UK cluster","authors":"Sami Astan , Kavan Fatehi , Amin Hajizadeh","doi":"10.1016/j.epsr.2026.112778","DOIUrl":"10.1016/j.epsr.2026.112778","url":null,"abstract":"<div><div>The paper introduces CCS-CoLearn, an integrated cooperative learning framework for the predictive and equitable operation of shared carbon capture and storage (CCS) infrastructures. The methodology unifies probabilistic deep-learning forecasting with game-theoretic optimisation to coordinate multiple industrial emitters linked to a common transport-and-storage system. A Transformer–BiLSTM ensemble generates multi-quantile forecasts of CO₂ inflows, parasitic electrical loads, and market prices, which feed a Nash-bargaining coordination layer ensuring fairness and Pareto efficiency among participants. The system is implemented within a digital-twin simulation of the Liverpool Bay CCS network, coupled to a reduced Great Britain power-grid model. Scenario analyses for 2030 and 2050 show that CCS-CoLearn achieves up to 18 % total-cost reduction, fairness index > 0.9, and 10 % higher CO₂ abatement compared with non-cooperative or rule-based baselines. The framework demonstrates that intelligent coordination can substitute for physical over-capacity, providing a scalable pathway for data-driven and cooperative management of CCS clusters under the UK Net-Zero 2050 strategy.</div></div>","PeriodicalId":50547,"journal":{"name":"Electric Power Systems Research","volume":"255 ","pages":"Article 112778"},"PeriodicalIF":4.2,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146039175","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-06-01Epub Date: 2026-02-04DOI: 10.1016/j.epsr.2026.112812
Osama E. Gouda , Gomaa F.A. Osman
In this paper the influences of unbalanced and faulty underground distribution cables on their neighboring pipelines are investigated. The studied cases include unbalances of the cable phases in current amplitudes, in phase shifts and in both of them. The impact of the induced potential on the pipelines has been highlighted in terms of the degree of unbalance, the composition of backfill soils surrounding the cables and pipelines as well as the separation distance between cables and pipelines. In addition, the impacts of homogeneous or heterogeneous soil and the cable laying depth on the induced voltage are investigated. The model simulation is executed by the use of the 2D-FEM with COMSOL multi-physics program. Significant increase in the induced voltages affecting buried metal pipes near the cable path is observed when there is imbalance in cable phases. The induced voltage reaches 32.24, 28.84, and 19.6 times the permissible value in cases of unbalanced 33 kV, 11 kV and 0.4 kV respectively with sandy soil as backfill material. It is found that the induced voltage depends on the soil composition, the space between the cables and PLs, in addition to the unbalance of cable phases. The safe distances between distribution cables and the pipelines have been examined and recommendations are made for safe distances. Installing cables inside PVC ducts mitigates the induced voltage by about 33.81% to 32% of its value. The present paper results are useful for workers in the installations of electrical cables and metal pipelines.
{"title":"A study on the impacts of unbalanced AC underground distribution cables on their neighboring metal pipes","authors":"Osama E. Gouda , Gomaa F.A. Osman","doi":"10.1016/j.epsr.2026.112812","DOIUrl":"10.1016/j.epsr.2026.112812","url":null,"abstract":"<div><div>In this paper the influences of unbalanced and faulty underground distribution cables on their neighboring pipelines are investigated. The studied cases include unbalances of the cable phases in current amplitudes, in phase shifts and in both of them. The impact of the induced potential on the pipelines has been highlighted in terms of the degree of unbalance, the composition of backfill soils surrounding the cables and pipelines as well as the separation distance between cables and pipelines. In addition, the impacts of homogeneous or heterogeneous soil and the cable laying depth on the induced voltage are investigated. The model simulation is executed by the use of the 2D-FEM with COMSOL multi-physics program. Significant increase in the induced voltages affecting buried metal pipes near the cable path is observed when there is imbalance in cable phases. The induced voltage reaches 32.24, 28.84, and 19.6 times the permissible value in cases of unbalanced 33 kV, 11 kV and 0.4 kV respectively with sandy soil as backfill material. It is found that the induced voltage depends on the soil composition, the space between the cables and PLs, in addition to the unbalance of cable phases. The safe distances between distribution cables and the pipelines have been examined and recommendations are made for safe distances. Installing cables inside PVC ducts mitigates the induced voltage by about 33.81% to 32% of its value. The present paper results are useful for workers in the installations of electrical cables and metal pipelines.</div></div>","PeriodicalId":50547,"journal":{"name":"Electric Power Systems Research","volume":"255 ","pages":"Article 112812"},"PeriodicalIF":4.2,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146189628","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
On‑line dissolved gas analysis (DGA) of transformer oil is essential for diagnosing incipient insulation faults and ensuring the operational reliability of power transformers. However, long‑term operation leads to sensor drift and reduced sensitivity, which eventually result in inadequate calibration and degraded diagnostic accuracy. To address these issues, this study proposes a comprehensive field‑deployable calibration platform based on precisely prepared standard oil samples. An integrated preparation‑calibration platform is developed, incorporating controlled single‑component gas injection, constant‑temperature and constant‑pressure oil‑gas equilibrium, automated multi‑level concentration switching, and pipeline self‑cleaning.
The proposed platform enables accurate preparation of dissolved‑gas reference samples with high linearity (R² ≥ 0.99). A complete on-site calibration workflow is established and validated on a 220 kV hydropower transformer. Based on comparative calibration using the prepared standard oil samples, results show that the tested commercial on‑line DGA device exhibits substantial deviations, indicating systematic accuracy limitations under field operating conditions. Statistical analysis including standard deviation, coefficient of variation, and error‑distribution plots further confirms poor repeatability of the device. A qualitative post‑calibration improvement trend is also provided to illustrate the expected correction behavior. The proposed methodology provides a full‑chain evaluation framework and a practical, standardized solution for field calibration of on‑line DGA systems, forming a methodological basis for large‑scale deployment of condition-based maintenance (CBM) strategies in smart‑grid applications.
{"title":"A field‑deployable calibration platform for on‑line dissolved gas monitoring systems in power transformer oil","authors":"Yue Ma, Xiaofeng Ma, Ronghui Wang, Jianhua Li, Kai Guan, Xiaofeng Chen","doi":"10.1016/j.epsr.2026.112782","DOIUrl":"10.1016/j.epsr.2026.112782","url":null,"abstract":"<div><div>On‑line dissolved gas analysis (DGA) of transformer oil is essential for diagnosing incipient insulation faults and ensuring the operational reliability of power transformers. However, long‑term operation leads to sensor drift and reduced sensitivity, which eventually result in inadequate calibration and degraded diagnostic accuracy. To address these issues, this study proposes a comprehensive field‑deployable calibration platform based on precisely prepared standard oil samples. An integrated preparation‑calibration platform is developed, incorporating controlled single‑component gas injection, constant‑temperature and constant‑pressure oil‑gas equilibrium, automated multi‑level concentration switching, and pipeline self‑cleaning.</div><div>The proposed platform enables accurate preparation of dissolved‑gas reference samples with high linearity (R² ≥ 0.99). A complete on-site calibration workflow is established and validated on a 220 kV hydropower transformer. Based on comparative calibration using the prepared standard oil samples, results show that the tested commercial on‑line DGA device exhibits substantial deviations, indicating systematic accuracy limitations under field operating conditions. Statistical analysis including standard deviation, coefficient of variation, and error‑distribution plots further confirms poor repeatability of the device. A qualitative post‑calibration improvement trend is also provided to illustrate the expected correction behavior. The proposed methodology provides a full‑chain evaluation framework and a practical, standardized solution for field calibration of on‑line DGA systems, forming a methodological basis for large‑scale deployment of condition-based maintenance (CBM) strategies in smart‑grid applications.</div></div>","PeriodicalId":50547,"journal":{"name":"Electric Power Systems Research","volume":"255 ","pages":"Article 112782"},"PeriodicalIF":4.2,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146189657","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-06-01Epub Date: 2026-02-10DOI: 10.1016/j.epsr.2026.112787
Barbara Pereira, Laura C.S. Pires, Leandro L. Morais, Silverio Visacro
In this work, the authors present experimental results of a new methodology based on on-site measurements to determine the frequency dependence of electrical parameters of soil, in the typical range of frequency components of lightning currents. A differential aspect of the methodology consists of the use of a short cylindrical tube as electrode. The study validated the theoretical developments previously carried out by the authors, which indicated the conditions required for adopting this specific electrode geometry in measurements. Notably, it addressed how to preserve a negligible inductive effect to allow accurately determining the appropriate geometrical factor. The work included experimental results and their derived curves of resistivity and permittivity as functions of frequency in low and high resistivity soils.
{"title":"Methodology for determining the electrical parameters of soil in the typical range of frequency components of lightning currents by means of measurements using a short cylindrical electrode","authors":"Barbara Pereira, Laura C.S. Pires, Leandro L. Morais, Silverio Visacro","doi":"10.1016/j.epsr.2026.112787","DOIUrl":"10.1016/j.epsr.2026.112787","url":null,"abstract":"<div><div>In this work, the authors present experimental results of a new methodology based on on-site measurements to determine the frequency dependence of electrical parameters of soil, in the typical range of frequency components of lightning currents. A differential aspect of the methodology consists of the use of a short cylindrical tube as electrode. The study validated the theoretical developments previously carried out by the authors, which indicated the conditions required for adopting this specific electrode geometry in measurements. Notably, it addressed how to preserve a negligible inductive effect to allow accurately determining the appropriate geometrical factor. The work included experimental results and their derived curves of resistivity and permittivity as functions of frequency in low and high resistivity soils.</div></div>","PeriodicalId":50547,"journal":{"name":"Electric Power Systems Research","volume":"255 ","pages":"Article 112787"},"PeriodicalIF":4.2,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146189599","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Line-commutated-converter based high voltage direct current (LCCHVDC) systems are susceptible to commutation failure (CF) under AC faults, posing a severe threat to power grid stability. Under asymmetric faults, different commutation processes (CP) exhibit varying CF risks due to significant differences in commutation voltages. This paper analyzes the impact of advanced firing on commutation processes and the CF risk for individual CPs under asymmetric faults, pointing out that for low-risk CPs, the benefit of reduced reactive power consumption without advanced firing outweighs the benefit of implementing advanced firing to ensure successful commutation. Thus, a novel method is proposed to enhance CF resistance capability by optimizing advanced firing application. To verify the effectiveness of the proposed method, it is applied to both direct and indirect advanced firing control approaches. Simulations are conducted in PSCAD/EMTDC using the CIGRE benchmark model and a dual-infeed HVDC model. The results of waveforms and Commutation Failure Immunity Index (CFII) comprehensively demonstrate that the proposed method effectively mitigates CF while maintaining good applicability across diverse operational scenarios.
{"title":"An optimization method for commutation failure mitigation under asymmetric faults in LCC-HVDC","authors":"Renlong Zhu , Shuowei Chen , Xin Tang , Xingyu Shi , Peng Guo , Wen Wang , Zhiming Guo , Yufei Yue","doi":"10.1016/j.epsr.2026.112801","DOIUrl":"10.1016/j.epsr.2026.112801","url":null,"abstract":"<div><div>Line-commutated-converter based high voltage direct current (LCC<img>HVDC) systems are susceptible to commutation failure (CF) under AC faults, posing a severe threat to power grid stability. Under asymmetric faults, different commutation processes (CP) exhibit varying CF risks due to significant differences in commutation voltages. This paper analyzes the impact of advanced firing on commutation processes and the CF risk for individual CPs under asymmetric faults, pointing out that for low-risk CPs, the benefit of reduced reactive power consumption without advanced firing outweighs the benefit of implementing advanced firing to ensure successful commutation. Thus, a novel method is proposed to enhance CF resistance capability by optimizing advanced firing application. To verify the effectiveness of the proposed method, it is applied to both direct and indirect advanced firing control approaches. Simulations are conducted in PSCAD/EMTDC using the CIGRE benchmark model and a dual-infeed HVDC model. The results of waveforms and Commutation Failure Immunity Index (CFII) comprehensively demonstrate that the proposed method effectively mitigates CF while maintaining good applicability across diverse operational scenarios.</div></div>","PeriodicalId":50547,"journal":{"name":"Electric Power Systems Research","volume":"255 ","pages":"Article 112801"},"PeriodicalIF":4.2,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146189649","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-06-01Epub Date: 2026-02-05DOI: 10.1016/j.epsr.2026.112807
Eduardo Caro, Jesús Juan
Accuracy metrics such as MAE, RMSE and WAPE are widely used to evaluate forecasting models in energy systems, yet they are commonly interpreted as fixed values despite being computed from stochastic, temporally correlated errors. This practice leads to an underestimation of the statistical variability and sampling dispersion of these accuracy metrics.
This paper introduces a probabilistic framework for analyzing accuracy metrics under temporal dependence. Using Taylor-based approximations and the covariance structure of dependent errors, we derive closed-form expressions for the mean and variance of MAE and WAPE. Two practical methods for constructing confidence intervals are proposed: (i) a theoretical approach based on the autocorrelation function of absolute errors, and (ii) an aggregation-based method that reduces short-term dependence through weekly averaging.
Monte Carlo simulations validate the proposed intervals and quantify the impact of different dependence patterns. A ten-year case study of hourly electricity demand in Spain shows that standard methods underestimate error variability, while the proposed ones correctly detect significant changes in forecast performance.
{"title":"Probabilistic analysis of electricity load forecasting errors","authors":"Eduardo Caro, Jesús Juan","doi":"10.1016/j.epsr.2026.112807","DOIUrl":"10.1016/j.epsr.2026.112807","url":null,"abstract":"<div><div>Accuracy metrics such as MAE, RMSE and WAPE are widely used to evaluate forecasting models in energy systems, yet they are commonly interpreted as fixed values despite being computed from stochastic, temporally correlated errors. This practice leads to an underestimation of the statistical variability and sampling dispersion of these accuracy metrics.</div><div>This paper introduces a probabilistic framework for analyzing accuracy metrics under temporal dependence. Using Taylor-based approximations and the covariance structure of dependent errors, we derive closed-form expressions for the mean and variance of MAE and WAPE. Two practical methods for constructing confidence intervals are proposed: (i) a theoretical approach based on the autocorrelation function of absolute errors, and (ii) an aggregation-based method that reduces short-term dependence through weekly averaging.</div><div>Monte Carlo simulations validate the proposed intervals and quantify the impact of different dependence patterns. A ten-year case study of hourly electricity demand in Spain shows that standard methods underestimate error variability, while the proposed ones correctly detect significant changes in forecast performance.</div></div>","PeriodicalId":50547,"journal":{"name":"Electric Power Systems Research","volume":"255 ","pages":"Article 112807"},"PeriodicalIF":4.2,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146189655","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-06-01Epub Date: 2026-02-12DOI: 10.1016/j.epsr.2026.112822
Miaomiao Ma , Rui Zhang , Xingyao Guo , Guobin Fu , Yujie Ding
The planning of wind solar hydrogen storage systems (WS-HSS) is challenging due to source-side uncertainties and inadequate load-side regulation, often leading to over-investment and unreliable supply. Therefore, this paper proposes a two-stage robust optimization (RO) method for capacity configuration of WS-HSS to balance the economic efficiency and robustness of the system. First, source-side uncertainty is fully captured by modeling its fluctuation range with a polyhedral uncertainty set. Then, demand response (DR) is introduced to reduce capacity requirements for critical equipment and the total costs. Subsequently, a two-stage RO model is constructed to minimize the total cost while ensuring robustness of the system. To solve the model efficiently, a column-and-constraint generation (CCG) algorithm is employed to dynamically search key scenarios representing uncertainty, thereby improving computational efficiency. Case study demonstrates that the proposed approach achieves a balance between cost-effectiveness and robustness, reducing the capacity of the energy storage system by approximately 20% and lowering the total cost by 7.08% of WS-HSS.
{"title":"Two-stage robust optimal capacity configuration method for wind solar hydrogen storage system considering source-side uncertainty","authors":"Miaomiao Ma , Rui Zhang , Xingyao Guo , Guobin Fu , Yujie Ding","doi":"10.1016/j.epsr.2026.112822","DOIUrl":"10.1016/j.epsr.2026.112822","url":null,"abstract":"<div><div>The planning of wind solar hydrogen storage systems (WS-HSS) is challenging due to source-side uncertainties and inadequate load-side regulation, often leading to over-investment and unreliable supply. Therefore, this paper proposes a two-stage robust optimization (RO) method for capacity configuration of WS-HSS to balance the economic efficiency and robustness of the system. First, source-side uncertainty is fully captured by modeling its fluctuation range with a polyhedral uncertainty set. Then, demand response (DR) is introduced to reduce capacity requirements for critical equipment and the total costs. Subsequently, a two-stage RO model is constructed to minimize the total cost while ensuring robustness of the system. To solve the model efficiently, a column-and-constraint generation (C<span><math><mo>&</mo></math></span>CG) algorithm is employed to dynamically search key scenarios representing uncertainty, thereby improving computational efficiency. Case study demonstrates that the proposed approach achieves a balance between cost-effectiveness and robustness, reducing the capacity of the energy storage system by approximately 20% and lowering the total cost by 7.08% of WS-HSS.</div></div>","PeriodicalId":50547,"journal":{"name":"Electric Power Systems Research","volume":"255 ","pages":"Article 112822"},"PeriodicalIF":4.2,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146189660","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-06-01Epub Date: 2026-02-04DOI: 10.1016/j.epsr.2026.112811
Gederson Alvaro da Cruz , Sérgio Haffner , Mariana Resener
We propose a mixed-integer linear programming (MILP) model for assessing the reliability of active distribution networks under both normal and contingency conditions. The objective function combines system operation costs with reliability costs. The model integrates self-healing mechanisms, including topology adjustments, system restoration via microgrids, and tie lines. Additionally, a linear formulation is developed to represent reward-penalty schemes and assess reliability costs. To demonstrate the applicability of the proposed model, tests are conducted on a 12-node system and a 136-node system, the latter based on real data. The results illustrate that the model provides a reliable diagnosis of the network and supports operators in decision-making to enhance network performance under various operational scenarios. A sensitivity analysis considering different failure rates, distributed generation capacities, and reward–penalty scheme coefficients is also presented, providing insights into their impact on system reliability and operational performance.
{"title":"MILP model for reliability assessment of active distribution networks with microgrids","authors":"Gederson Alvaro da Cruz , Sérgio Haffner , Mariana Resener","doi":"10.1016/j.epsr.2026.112811","DOIUrl":"10.1016/j.epsr.2026.112811","url":null,"abstract":"<div><div>We propose a mixed-integer linear programming (MILP) model for assessing the reliability of active distribution networks under both normal and contingency conditions. The objective function combines system operation costs with reliability costs. The model integrates self-healing mechanisms, including topology adjustments, system restoration via microgrids, and tie lines. Additionally, a linear formulation is developed to represent reward-penalty schemes and assess reliability costs. To demonstrate the applicability of the proposed model, tests are conducted on a 12-node system and a 136-node system, the latter based on real data. The results illustrate that the model provides a reliable diagnosis of the network and supports operators in decision-making to enhance network performance under various operational scenarios. A sensitivity analysis considering different failure rates, distributed generation capacities, and reward–penalty scheme coefficients is also presented, providing insights into their impact on system reliability and operational performance.</div></div>","PeriodicalId":50547,"journal":{"name":"Electric Power Systems Research","volume":"255 ","pages":"Article 112811"},"PeriodicalIF":4.2,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146189658","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-06-01Epub Date: 2026-02-12DOI: 10.1016/j.epsr.2026.112837
Zengxi Feng , Zhenghao Zhu , Hang Ge , Lang Han , Anjun Zhao , Wei Quan , Xiao Xue
Forecasting building energy consumption is essential for improving energy efficiency and enabling energy scheduling. However, existing energy consumption prediction models face two main issues: (1) Although some building energy consumption prediction models have demonstrated good predictive performance, they may have limitations when confronted with the complex nonlinear and time-dependent characteristics of building load data. (2) Existing studies mainly focus on comparing the performance of different prediction models, with limited attention given to the impact of optimizing combinations of hyperparameters within the models. Therefore, this study proposes a CNN-BiLSTM-Attention hybrid prediction model optimized by the improved Black Kite Algorithm (IBKA). The improved IBKA is then applied to optimize the hyperparameters under different combinations. Simulation results indicate that the building energy consumption prediction model optimized by IBKA algorithm outperforms models optimized by other algorithms, such as PSO or BWO. The study further demonstrates that optimizing six hyperparameters, including learning rate, the number of bidirectional long short-term (BiLSTM) hidden layers, and the maximum number of iterations, leads to optimal prediction accuracy. On Dataset 1, compared with models optimized by other algorithms, the proposed model achieves lower root mean square error (RMSE) and mean absolute error (MAE) values, with an R² of 98.74% under the same six-hyperparameter configuration. In addition, on Dataset 2, the proposed model also exhibits good prediction performance for the test days from summer, transitional, and winter seasons, indicating reliable prediction capability.
{"title":"A day-ahead building load forecasting method based on IBKA-CNN-BiLSTM-Attention model","authors":"Zengxi Feng , Zhenghao Zhu , Hang Ge , Lang Han , Anjun Zhao , Wei Quan , Xiao Xue","doi":"10.1016/j.epsr.2026.112837","DOIUrl":"10.1016/j.epsr.2026.112837","url":null,"abstract":"<div><div>Forecasting building energy consumption is essential for improving energy efficiency and enabling energy scheduling. However, existing energy consumption prediction models face two main issues: (1) Although some building energy consumption prediction models have demonstrated good predictive performance, they may have limitations when confronted with the complex nonlinear and time-dependent characteristics of building load data. (2) Existing studies mainly focus on comparing the performance of different prediction models, with limited attention given to the impact of optimizing combinations of hyperparameters within the models. Therefore, this study proposes a CNN-BiLSTM-Attention hybrid prediction model optimized by the improved Black Kite Algorithm (IBKA). The improved IBKA is then applied to optimize the hyperparameters under different combinations. Simulation results indicate that the building energy consumption prediction model optimized by IBKA algorithm outperforms models optimized by other algorithms, such as PSO or BWO. The study further demonstrates that optimizing six hyperparameters, including learning rate, the number of bidirectional long short-term (BiLSTM) hidden layers, and the maximum number of iterations, leads to optimal prediction accuracy. On Dataset 1, compared with models optimized by other algorithms, the proposed model achieves lower root mean square error (RMSE) and mean absolute error (MAE) values, with an R² of 98.74% under the same six-hyperparameter configuration. In addition, on Dataset 2, the proposed model also exhibits good prediction performance for the test days from summer, transitional, and winter seasons, indicating reliable prediction capability.</div></div>","PeriodicalId":50547,"journal":{"name":"Electric Power Systems Research","volume":"255 ","pages":"Article 112837"},"PeriodicalIF":4.2,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146189602","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-06-01Epub Date: 2026-02-09DOI: 10.1016/j.epsr.2026.112766
Wen He , Rui Ma , Jiangbo Sha , Dongge Zhu , Shuang Zhang
Due to the high penetration of renewable energy sources in distribution networks, the problem of voltage limit violations has become increasingly prominent. Conventional methods relying on physical models and comprehensive measurement data struggle to address the challenges posed by the stochastic fluctuations of photovoltaics (PV) and frequent topology changes. In response, a hybrid provenance model integrating Graph Convolutional Networks with Causal reasoning (GCN+Causal) is proposed. The model characterizes time-varying coupling relationships among nodes using a dynamic adjacency matrix, incorporates counterfactual reasoning to quantify causal effects along fault propagation paths via double machine learning, and utilizes a spatio-temporal GCN-based feature encoder to handle non-Euclidean characteristics of electrical measurement data for feature extraction, thereby overcoming the limitations of traditional correlation-based analysis. In simulation tests on an IEEE node system, the proposed method improves the comprehensive performance of voltage anomaly localization to 93.2 % (a 21.5 % increase over conventional state estimation methods) and reduces the false alarm rate to 4.3 %. The experimental results demonstrate that the model accuracy remains at 87.6 % even with 30 % missing measurements and controls traceability time within 500 ms. The research outcome provides a novel analytical tool for enhancing the security of active distribution networks, and its causal interpretability facilitates the development of precise control strategies.
{"title":"Voltage out-of-limit traceability simulation model of distribution network based on graph convolutional network and causal reasoning","authors":"Wen He , Rui Ma , Jiangbo Sha , Dongge Zhu , Shuang Zhang","doi":"10.1016/j.epsr.2026.112766","DOIUrl":"10.1016/j.epsr.2026.112766","url":null,"abstract":"<div><div>Due to the high penetration of renewable energy sources in distribution networks, the problem of voltage limit violations has become increasingly prominent. Conventional methods relying on physical models and comprehensive measurement data struggle to address the challenges posed by the stochastic fluctuations of photovoltaics (PV) and frequent topology changes. In response, a hybrid provenance model integrating Graph Convolutional Networks with Causal reasoning (GCN+Causal) is proposed. The model characterizes time-varying coupling relationships among nodes using a dynamic adjacency matrix, incorporates counterfactual reasoning to quantify causal effects along fault propagation paths via double machine learning, and utilizes a spatio-temporal GCN-based feature encoder to handle non-Euclidean characteristics of electrical measurement data for feature extraction, thereby overcoming the limitations of traditional correlation-based analysis. In simulation tests on an IEEE node system, the proposed method improves the comprehensive performance of voltage anomaly localization to 93.2 % (a 21.5 % increase over conventional state estimation methods) and reduces the false alarm rate to 4.3 %. The experimental results demonstrate that the model accuracy remains at 87.6 % even with 30 % missing measurements and controls traceability time within 500 ms. The research outcome provides a novel analytical tool for enhancing the security of active distribution networks, and its causal interpretability facilitates the development of precise control strategies.</div></div>","PeriodicalId":50547,"journal":{"name":"Electric Power Systems Research","volume":"255 ","pages":"Article 112766"},"PeriodicalIF":4.2,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146189627","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号