List of IssuesDocument Types
Flexible Regenerative Heating System Retrofit and Thermal Economic Performance of Ultra-Supercritical Coal-Fired Units under Deep Peak Shaving AI Introduction
Abstract：ObjectivesIn deep peak shaving and wide-load operation of ultra-supercritical coal-fired units, insufficient secondary-air heating, redistribution of regenerative heat, and deterioration of thermal economic performance may occur. To address these issues, this study evaluates flexible regenerative heating system reconstruction and secondary-air heating retrofit schemes.MethodsA 1 000 MW ultra-supercritical coal-fired unit is selected as the research object, and thermal calculation models of the original system and two retrofit schemes are established based on EBSILON thermal simulation software. The variation patterns of key parameters including secondary-air temperature, exhaust gas temperature, steam flow rate, heat duty, heat rate, and coal consumption are analyzed in the low-load range of 15% turbine heat acceptance (THA) to 40% THA and the medium- and high-load range of 50% THA to 100% THA.ResultsIn the original system, the secondary-air temperature continuously decreases in the low-load range, while the exhaust gas temperature increases, accelerating the deterioration of thermal economic performance. In the low-load range, superheated steam is used to entrain cold reheat steam for supplementary heating, increasing the secondary-air temperature by 11.787-33.876 ℃, with a coal consumption improvement rate of 2.662%-10.694%, although the boiler efficiency decreases and the exhaust gas temperature increases. In the medium- and high-load range, third-stage extraction steam is used for supplementary heating, increasing the secondary-air temperature by 10.266-12.331 ℃. However, the unit heat rate and coal consumption increase slightly because part of the feedwater regenerative heat is used for secondary-air heating.ConclusionsThe two retrofit schemes achieve secondary-air temperature regulation and regenerative heat redistribution through different thermal system reconstruction paths, which is of great significance for improving the wide-load operation adaptability of ultra-supercritical units and supporting deep peak shaving retrofit. The findings can provide a reference for secondary-air heating system optimization and zonal operation strategy formulation.Keywords：ultra-supercritical coal-fired units;thermal system reconstruction;flexible regenerative heating;deep peak shaving;wide-load operation;thermal economic performance15|12|0
<HTML><L-PDF><WORD><Meta-XML>Updated：2026-09-30A Control Strategy for Grid-Forming Inverters in Grids With a Wide Range of Short-Circuit Ratios AI Introduction
Abstract：ObjectivesTo address the low-frequency oscillation issue in grid-forming inverters caused by voltage-source characteristic conflicts under strong-grid conditions, a current compensation control strategy applicable to grids with a wide range of short-circuit ratios is proposed, aiming to enhance the grid-connection stability and robustness of inverters under different grid strengths.MethodsThe proposed strategy retains the classic dual-closed-loop control architecture while introducing a current feedforward compensation loop. Through a control-domain compensation mechanism, it actively reshapes the frequency characteristics of the inverter's closed-loop output admittance without altering the physical impedance of the main circuit. A closed-loop impedance model of the grid-connected system is established, and frequency-domain analysis and characteristic root analysis are used to reveal the wideband damping reshaping mechanism of the proposed strategy.ResultsThe proposed strategy can effectively eliminate low-frequency power oscillations under strong grid conditions while maintaining sufficient phase stability margins under weak-grid conditions. Compared with virtual impedance control, the proposed strategy overcomes the dependency on the grid short-circuit ratio and avoids further deterioration of weak-grid strength caused by the superposition of equivalent control impedance, thereby exhibiting strong grid adaptability.ConclusionsThe proposed strategy achieves adaptive damping through admittance reshaping, ensuring the stable operation of grid-forming inverters across a wide range of short-circuit ratios, significantly enhancing the scenario adaptability of renewable energy grid-connected systems.Keywords：grid-forming inverter;current compensation control;admittance reshaping;wideband oscillation;short-circuit ratio;weak grid;small-signal stability;impedance modeling12|9|0<HTML><L-PDF><WORD><Meta-XML>Updated：2026-09-30Performance Analysis of Thermal System of Indirect Air-Cooled Units in Coal-Fired Power Plants During Variable Load Operation AI Introduction
Abstract：ObjectivesTo address the limited understanding of the coupling between thermal-system performance and cold-end parameters in indirect air-cooled units, this study takes a 660 MW ultra-supercritical coal-fired unit in Qinghai as the research object. It develops a thermodynamic-cycle simulation model using EBSILON software, and the model accuracy is verified using data from valve wide open (VWO) condition, typical turbine heat acceptance (THA) condition, and variable conditions.MethodsUnder constant-pressure and combined sliding-pressure operation modes, steam consumption, heat rate, and the exergy loss coefficients of regenerative heaters and the condenser are calculated and compared. The responses of unit output and condenser heat load to different back pressures and coal feed rates are further analyzed.ResultsThe results show that, as load increases, the heat rate decreases overall under both operation modes, while steam consumption first decreases and then increases. Compared with constant-pressure operation, combined sliding-pressure operation reduces steam consumption by about 1.63%-5.31% and heat rate by about 1.41%-4.54%, with the improvements in economic performance concentrated in the medium- and low-load range. The condenser exergy loss coefficient decreases with increasing load, and its proportion is about 3-6 times the sum of those of all regenerative heaters, making it a key component in cold-end optimization. At a fixed coal feed rate, increasing back pressure reduces unit output, with more significant losses occurring in the high-load range.ConclusionsThe findings provide a reference for variable-load operation mode selection, cold-end optimization, and energy saving and consumption reduction in indirect air-cooled coal-fired units.Keywords：coal-fired power generation;thermal system;variable load operation;exergy analysis;exergy loss coefficient;energy saving and consumption reduction13|13|0<HTML><L-PDF><WORD><Meta-XML>Updated：2026-09-30A Forecasting Method for Frequency Regulation Capacity Demand Interval Based on Chance-Constrained Optimization AI Introduction
Abstract：ObjectivesTo achieve the “dual carbon”goals, new energy sources are gradually becoming the primary contributors to both installed capacity and electricity generation in power systems. However, due to their strong volatility, the frequency fluctuations in the power system are increasing, leading to a higher demand for frequency regulation capacity. Therefore, this paper proposes a forecasting method for the frequency regulation capacity demand interval based on chance-constrained optimization.MethodsThe improved complete ensemble empirical mode decomposition with adoptive noise technique is used to process various factors related to frequency regulation capacity fluctuations, including new energy sources, load, frequency regulation capacity, and meteorological data. A convolutional neural network-bidirectional long short-term memory with self-attenion model is established to predict the point value of frequency regulation capacity demand during scheduling periods. Adaptive Bandwidth Kernel Density Estimation is employed to determine the interval of frequency regulation capacity demand, yielding a 95% confidence interval for the frequency regulation capacity demand. Based on actual unit scheduling, the Monte Carlo simulation method is applied to optimize the upper bound of the demand interval using chance-constrained optimization, thereby aligning the upper bound more closely with actual values.ResultsThe predicted interval of frequency regulation capacity demand meets actual regulation requirements. Additionally, the average interval width is reduced by 11%-19% compared to other methods.ConclusionsThe proposed method effectively addresses the challenge of forecasting frequency regulation capacity demand in power systems, thereby enhancing the flexibility, stability, and economic efficiency of power grid scheduling.Keywords：new energy;new power system;frequency regulation capacity demand;interval forecasting;power grid dispatching;chance-constrained optimization;empirical mode decomposition7|4|0<HTML><L-PDF><WORD><Meta-XML>Updated：2026-09-30Smoke and Fire Detection Algorithm for Hydropower Stations Based on Deep Learning Fusion Model AI Introduction
Abstract：ObjectivesHydropower stations have complex environments and high fire risks. Traditional smoke and fire detection methods face challenges such as missed detection of small targets, low accuracy in long-range recognition, and high false alarm rates in complex dynamic environments. Therefore, a deep learning fusion model based on YOLOv8 and Transformer is proposed.MethodsThe model combines YOLOv8's powerful local target localization capability with Transformer's advantage in global context modeling, and a feature recalibration gating mechanism is introduced to achieve dynamic complementary fusion between the two. Features at different levels are dynamically evaluated and weighted for fusion to enhance the semantic representation of smoke and fire targets at varying scales. Furthermore, correlation analysis between historical frame motion features and current frame spatial features effectively distinguishes the continuous motion patterns of real smoke and fire from environmental interference. An end-to-end training framework incorporating targeted data augmentation strategies is adopted, and the model is optimized using a comprehensive loss function. Finally, experimental verification is conducted on a self-built dataset covering multiple complex scenarios of hydropower stations.ResultsThe proposed algorithm achieves a small-target detection mAP@0.5 of 0.698. In long-range recognition tasks, its accuracy degradation is significantly lower than that of other comparison models.ConclusionsThe proposed algorithm performs better than existing mainstream methods in small-target detection, long-range recognition, and dynamic interference suppression, and can provide effective technical support for intelligent monitoring of smoke and fire in hydropower stations.Keywords：smoke and fire detection;hydropower station;deep learning fusion model;Transformer;YOLOv8;weighted fusion;loss function;feature recalibration gating mechanism (FRGM)0|0|0<HTML><L-PDF><WORD><Meta-XML>Updated：2026-09-03- Abstract：ObjectivesIndustrial parks are currently major sources of carbon emissions, and their transition to zero carbon serves as a key carrier for achieving China’s national “dual carbon” strategic goals. The development of zero-carbon parks currently faces numerous challenges, including technology integration, investment costs, institutional coordination, standard accounting, and policy implementation. Therefore, it is necessary to systematically review the corresponding policy frameworks, standard systems, and technical pathways to overcome the existing bottlenecks.MethodsStarting from the analysis of zero-carbon park development pathways, this study first discusses their connotations, development stages, and “source–grid–load–storage–carbon sink” emission reduction system, reviews the three stages of policy evolution, and proposes a three-in-one policy framework comprising national top-level design, local supporting measures and implementation, and standard system support. Subsequently, five typical demonstration project models are summarized, and four core technologies—zero-carbon energy supply, multi-energy complementarity, resource recycling, and smart energy and carbon management—are analyzed. Finally, the differences in technical pathways for the three development stages—low-carbon, near-zero-carbon, and zero-carbon parks—are compared. Common bottlenecks revealed in engineering practice and corresponding improvement pathways are examined.ResultsPolicies related to zero-carbon parks in China have evolved from local spontaneous exploration into a systematic national strategy. Demonstration projects can be categorized into five typical models. Technical standards have established a preliminary system architecture comprising four distinct categories. Core technologies are evolving from improvements in individual energy efficiency toward multi-energy integrated development. However, engineering practices still face three common bottlenecks. These problems arise from three underlying contradictions: a disconnect between planning and design and actual operation, low utilization of energy storage capacity, and difficulties in implementing policies for direct green power connection and microgrid operation.ConclusionsThe development of zero-carbon parks in the future requires coordinated efforts in five directions: technology integration and system intelligence, clean energy supply and multi-energy integration, digital carbon asset management and mature market mechanisms, a unified standard system and multi-party collaborative innovation, and policy coordination and institutional innovation. Only through coordinated progress across all five directions can zero-carbon parks achieve the transition from being merely constructed to being well-developed.Keywords：zero-carbon parks;policy system;demonstration projects;technical standards;technical system18|1|0<HTML><L-PDF><WORD><Meta-XML>Updated：2026-09-03
Analysis and Prediction of Energy Losses in AC/DC Conversion Processes for China’s New Power Systems AI Introduction
Abstract：ObjectivesThe energy supply and electricity consumption structures in China are currently undergoing a profound transformation. The integration of a high proportion of new energy into the grid and the widespread adoption of DC power consumption devices have brought new challenges to the operation of the new power systems. The energy consumption changes across the source-grid-load chain in the power system are still unclear. Therefore, this study analyzes and predicts energy losses during AC/DC conversion processes in new power systems to provide theoretical support for improving quality and efficiency in the power system transition.MethodsBased on the development characteristics of new power systems, this study is conducted from the perspective of the entire source-grid-load chain. It focuses on the AC/DC conversion losses from new energy grid integration on the power generation side, rectification losses from DC devices on the consumption side, and additional losses caused by replacing AC lines with DC lines. Additionally, two development scenarios for AC/DC hybrid systems—natural DC growth and wind-solar DC matching—are constructed for comparative analysis and loss prediction.ResultsIt is estimated that the total AC/DC conversion losses across the whole society in China reached 744.7 billion kW⋅h in 2023, accounting for 7.9% of the total national electricity generation in that year. Transitioning to an AC/DC hybrid system can effectively reduce the energy losses across the entire chain of the new power system. Under the wind-solar DC matching scenario in 2030 and 2050, national electricity availability is expected to increase by 7.9% and 14.1%, respectively, demonstrating significant effectiveness in energy loss reduction and efficiency improvement.ConclusionsPromoting the transition from the traditional AC-dominated power grid to an AC/DC hybrid system with a higher proportion of DC can effectively adapt to the industry trends of new energy becoming the dominant energy source and the continuous growth of DC electricity loads. It is a key pathway for reducing energy losses across the entire power system chain and enhancing the overall operational efficiency of the power system.Keywords：new power system;new energy;AC/DC conversion;energy loss;analysis and prediction4|0|0<HTML><L-PDF><WORD><Meta-XML>Updated：2026-08-31- Abstract：ObjectivesVirtual power plants, by aggregating distributed energy resources and participating in power balance regulation, have become a key technological pathway for enhancing the flexibility and security resilience of power systems. However, their practical operation still faces challenges of high-dimensional uncertainty and complex decision-making. Artificial intelligence (AI), with its data processing and intelligent decision-making capabilities, is becoming a key means of supporting the refined operation and intelligent upgrading of virtual power plants, such as state awareness and dynamic trading. Although related research has received widespread attention, existing studies are relatively fragmented and lack systematic review. Therefore, this study reviews the key technologies of AI-driven optimal operation of virtual power plants.MethodsFirst, this review introduces the basic characteristics of AI technology, elaborates on the core features and operational models of virtual power plants, and discusses the advantages of applying AI to optimal operation of virtual power plants. Second, the principles and research advances of AI applications are systematically examined across three dimensions: resource state awareness and data processing, intelligent optimal scheduling and coordinated control, and electricity market trading mechanisms.ResultsKey issues and difficulties in existing methodologies are analyzed, particularly concerning heterogeneous resource modeling and uncertainty characterization. Finally, the core challenges faced by AI application in virtual power plants are analyzed, and future development trends are outlined, including high trustworthiness and explainability, end-edge-cloud collaboration and distributed intelligence, and general-purpose AI and autonomous intelligent systems.ConclusionsAI can enhance the capabilities of virtual power plants in uncertainty characterization and real-time decision-making in the areas of resource awareness, coordinated scheduling, and market trading. In the future, its engineering application needs to be driven by trustworthy AI and end-edge-cloud collaboration.Keywords：virtual power plant;artificial intelligence (AI);distributed energy resources;coordinated scheduling;electricity market;optimal operation3|0|0<HTML><L-PDF><WORD><Meta-XML>Updated：2026-08-31
- Abstract：ObjectivesConcentrated photovoltaic/thermal systems provide a new approach for achieving efficient and comprehensive year-round utilization of solar energy. However, in practical applications, the non-uniform luminous flux falling on the receiver surface after reflection by the concentrator causes the receiver to be unevenly heated, thereby affecting the overall thermal and electrical performance of the system. Therefore, this study analyzes the comprehensive influence of non-uniform luminous flux on the performance of concentrated photovoltaic/thermal systems through numerical simulation methods, aiming to facilitate rational system design.MethodsThe Monte Carlo ray tracing method and the finite volume method are combined to construct an opto-electro-thermal coupling model of a concentrated photovoltaic/thermal system with a compound parabolic concentrator. By adjusting the position of the concentrating photovoltaic receiver to the plane of optimal uniformity, the non-uniformity of luminous flux distribution on the receiver surface is effectively reduced. By analyzing the influence of the solar spectrum and glass transmittance variations with wavelength, the characteristics of the influence of the average and actual luminous flux distributions on the thermal and electrical performance of the system are compared.ResultsWhen the incident solar irradiance is 1 000 W/m2 and the angle of incidence is 0°, the thermal efficiency and electrical efficiency of the system are 79.28% and 14.53%, respectively. As the angle of incidence increases, the system’s thermal efficiency decreases significantly. When the angle of incidence increases to 15°, the concentrated solar irradiance received by the photovoltaic modules falls below the original incident irradiance without concentration, and the system loses its concentrating effect.ConclusionsNon-uniform luminous flux has little influence on the average outlet temperature of the cooling fluid and the average temperature of the photovoltaic panel. An increase in the incidence angle of solar irradiance enhances the fluctuations in the luminous flux distribution on the receiver surface and gradually produces peaks, which is not conducive to improving the overall performance of the system.Keywords：solar energy;concentrated photovoltaic/thermal system;Monte Carlo ray tracing method;thermal and electrical performance;non-uniform luminous flux3|2|0<HTML><L-PDF><WORD><Meta-XML>Updated：2026-08-31
- Abstract：ObjectivesDriven by the “dual carbon” goals, to address the challenges of limited hydrogen energy recycling pathways and insufficient load scheduling potential in virtual power plants, a two-stage optimal scheduling strategy for virtual power plants considering diversified hydrogen energy utilization and excess-carbon demand response is proposed.MethodsAt the technical level, hydrogen fuel cell vehicles are aggregated as flexible energy conversion units of the virtual power plant, and pathways for diversified hydrogen utilization and blended fuel supply for gas-fired units are constructed. At the mechanism level, a two-stage excess-carbon demand response mechanism is designed. In the re-scheduling stage, excess-carbon triggering conditions are determined in real time, and a dynamic carbon penalty electricity price is generated to guide flexible loads to shift toward a low-carbon state.ResultsThe proposed strategy can create a closed-loop hydrogen demand environment within the virtual power plant. The aggregated scheduling of hydrogen fuel cell vehicles enhances renewable energy accommodation capability and effectively smooths intraday supply-demand fluctuations of the virtual power plant. Compared with the traditional time-of-use pricing mechanism, the two-stage excess-carbon demand response mechanism exhibits superior low-carbon economic performance in terms of carbon emission reduction, mitigation of load peak-valley differences, and improvement of system operation stability.ConclusionsThe proposed coupling mechanism of hydrogen energy recycling and excess-carbon demand response can profoundly expand pathways for multi-level hydrogen utilization and broaden the carbon reduction space on both the source and load sides, thereby achieving simultaneous improvement in the economic and low-carbon performance of the virtual power plant.Keywords：virtual power plant;hydrogen energy;hydrogen fuel cell vehicle;carbon emission reduction;demand response;low-carbon economic scheduling4|0|0<HTML><L-PDF><WORD><Meta-XML>Updated：2026-08-31
Research on Support Mechanism of Photovoltaic Cluster Ancillary Service Modes for System Electro-Carbon Operational Efficiency AI Introduction
Abstract：ObjectivesUnder the background of the development of new power systems, a system efficiency measurement framework considering photovoltaic cluster ancillary services (CAS) is designed to clarify the support mechanism of CAS on the system electricity-carbon operational efficiency.MethodsBased on the power grid carbon flow theory, a system carbon flow calculation method under the CAS mode and a cluster classification approach integrating carbon flow indicators are proposed. An efficiency evaluation model is developed with photovoltaic power output as the input indicator and electrical quantities and carbon flow as the output indicators, to analyze the coupling mechanism among multi-dimensional indicators of the system, and simulation verification is carried out on the improved IEEE 14-bus system.ResultsCAS can optimize the system electricity-carbon indicators. The all-time line loss rate is 4.220%-4.955%, the maximum voltage deviation is 6.000%-6.200%, the frequency deviation during peak and valley periods of the system is controlled within ±0.080 Hz, the carbon emission factor is reduced by about 16.500%, and the average node carbon potential fluctuation is reduced.ConclusionsCAS can improve key system electro-carbon indicators, reduce the system line loss rate, voltage deviation, and frequency deviation, lower the carbon emission factor and the average node carbon potential fluctuation, thus improving the energy-saving and carbon-reduction effect of the system. The findings can provide references for the formulation of low-carbon operation optimization strategies of new power systems and the allocation of new energy ancillary service resources.Keywords：photovoltaic cluster;ancillary services;carbon emission flow;cluster partitioning;energy saving and carbon reduction;low-carbon operation24|39|0<HTML><L-PDF><WORD><Meta-XML>Updated：2026-07-20- Abstract：ObjectivesUnder the national strategic goals of carbon peak and carbon neutrality, coal-fired power units frequently participate in deep peak-shaving operations at low loads, during which uneven flow distribution and combustion bias in the boiler heating surfaces cause thermal deviations, thereby increasing the risk of tube bursts due to water wall overheating. Conventional wall temperature monitoring methods, based on the enthalpy change of the working fluid, exhibit hysteresis and have difficulty in differentiating between inherent temperature distribution patterns and abnormal operational deviations. To address this, a thermal deviation discrimination method based on historical data and Euclidean distance is proposed for a 660 MW ultra-supercritical opposed-firing boiler.MethodsThrough analyzing the wall temperature of the main heating surfaces of the boiler, the operating load range is divided into intervals and a benchmark temperature database is constructed. Characteristic temperatures are extracted to eliminate the influence of overall temperature fluctuations. Furthermore, a thermal deviation index is established to quantify real-time deviations, and the method is applied in practice.ResultsSignificant temperature differences exist among the tube panels of the heating surfaces, and this is especially apparent in the high-temperature reheater. The fluctuations in wall temperature consist of two parts: one is the inherent deviation arising from the system’s own characteristics, and the other is the fluctuating deviation brought about by adjustments in the operating mode.ConclusionsThe proposed method successfully identifies both the inherent deviations of the equipment and the operational fluctuating deviations, and it provides a real-time discrimination basis for boiler safety early-warning systems, ensuring operational safety under deep peak-shaving conditions.Keywords：coal-fired power generation;ultra-supercritical boiler;deep peak-shaving;wall temperature analysis;Euclidean distance;thermal deviation34|23|0<HTML><L-PDF><WORD><Meta-XML>Updated：2026-06-30
Research on Meteorological Causes and Risk Distribution of Wind Turbine Blade Icing Based on Interpretable Machine Learning AI Introduction
Abstract：ObjectivesTo address the engineering needs for accurate identification and early warning of wind turbine blade icing risks, this study systematically analyzes the meteorological causes of blade icing and develops an engineering-applicable risk assessment scheme.MethodsLong-term icing observation records from wind farms in northern Guangxi are integrated with ERA5 multi-dimensional reanalysis meteorological data. An interpretable XGBoost-based machine learning model is constructed for icing decision-making. The SHAP method is used to quantitatively analyze the contributions of key meteorological factors. Icing criteria are refined via decision tree rule extraction and physical scenario clustering.ResultsThe constructed icing decision-making model achieves an area under the curve (AUC) of 0.964 4, with a recall rate of 97% for icing events. Near-surface temperature, low-level relative humidity, and maximum atmospheric temperature are identified as the three core influencing factors. Two parameterized icing criteria are proposed: strong low-temperature and high-humidity type and inversion-enhanced type, covering 93.7% of positive icing samples.ConclusionsThis study reveals the nonlinear coupling effects of multiple meteorological factors on wind turbine blade icing. The derived parameterized criterion scheme and regional risk distribution results can provide reliable technical support for wind farm planning, accurate icing early warning, and efficient operation of anti-icing and de-icing systems.Keywords：wind power generation;wind turbine;blade icing;interpretable machine learning;meteorological factors;parameterized criteria21|111|0<HTML><L-PDF><WORD><Meta-XML>Updated：2026-06-30- Abstract：[Objectives] This paper focuses on the soot and coking monitoring of boiler heating surfaces, in order to deal with the problem that the conventional fixed-interval soot-blowing methods of boiler are no longer adequate under such dynamic conditions under the background of flexible deep peak shaving, often leading to over-blowing or under-blowing. [Methods]This study takes the low-temperature reheater of a 660 MW ultra-supercritical coal-fired boiler as the research object and develops a time-series and multi-factor ash deposition monitoring model based on a Backpropagation Neural Network (BPNN). The actual heat transfer rate is calculated from the working fluid parameters at the heating surface inlet and outlet, while a theoretical heat transfer prediction model for a clean heating surface is established by integrating time-series data and multiple influencing factors. [Results]The model achieves a root mean square error of 5635.33 and a MAPE of 0.62%, significantly outperforming traditional models. [Conclusions]Without requiring additional sensors or measurement points, the proposed method fully leverages historical data to enable online monitoring of ash deposition severity, providing reliable data support for intelligent soot-blowing in boilers.Keywords：超超临界;燃煤机组;反向传播神经网络(BPNN);吹灰;积灰结焦;在线监测;受热面31|10|0<HTML><L-PDF><Meta-XML>Updated：2026-05-29
- Abstract：[Objecttives] To enhance the stability of renewable energy systems and achieve accurate photovoltaic power prediction, this study proposes a hybrid prediction model integrating a multi-scale temporal convolutional network (MS-TCN), a gated position-aware transformer encoder (GPATE), and a bidirectional gated recurrent unit (BiGRU). [Methods] The MS-TCN module is first employed to extract local features from multi-dimensional time series data. Then, the GPATE module is introduced, in which the traditional feedforward network is replaced by a position-aware multilayer perceptron, augmented with a gating mechanism to suppress noise and enhance the model's ability to capture long-term dependencies and global contextual information. Finally, the feature representation of MS-TCN and GPATE is fused, and BiGRU is used to construct bidirectional temporal dependencies to further strengthen the modeling ability of complex temporal dynamic features. [Results] Compared with the Transformer model, the proposed model achieves significant improvements of 42.87% and 29.37% in RMSE and MAE, respectively. [Conclusions] Through structural innovation, the proposed model effectively improves prediction accuracy and provides reliable support for power system scheduling under high penetration of renewable energy.Keywords：renewable energy, photovoltaic, power prediction, multi-scale temporal convolutional network (MS-TCN), gated position-aware transformer encoder (GPATE), feature fusion, structural innovation24|8|0<HTML><L-PDF><Meta-XML>Updated：2026-05-29
- Abstract： [Objecttives] A comprehensive solution is proposed to address the issues of insufficient accuracy in nuclear power plant (NPP) models for large-scale power grid stability simulation-leading to distorted dynamic characteristics—and the lack of coordination mechanisms between grid-related protections and the grid. [Methods] First, a high-precision nonlinear model of the NPP power system is developed, covering the reactor core, primary circuit, and secondary circuit. Key algorithms are introduced, including steady-state equilibrium adjustment and variable initialization under off rated conditions, a power interface method based on working fluid enthalpy drop, and a multi rate hybrid numerical integration scheme. Furthermore, a model order reduction method combining balanced truncation and singular perturbation theory is proposed, which significantly improves the computational efficiency of electro-mechanical transient simulation for large scale grids. Based on the developed model, the interaction mechanism between nuclear power plants and the grid is elucidated. For typical fault scenarios such as over frequency, under frequency, and under voltage, coordinated control strategies between grid related protections and grid safety automatic devices are presented. [Results] Simulation results verify the accuracy of the model and the effectiveness of the strategies, providing a powerful tool and research reference for studying the dynamic characteristics of NPPs and their interaction and coordinated operation with the power grid. [Conclusions] The research results can provide powerful simulation tools and research reference for safety and stability analysis and control decision-making of nuclear power units connected to large power grids.Keywords：pressurized water reactor(PWR);nuclear power plant (NPP);dynamic simulation;modeling;coordination between units and grids18|5|0<HTML><L-PDF><Meta-XML>Updated：2026-05-29
- Abstract： [Objectives] Against the backdrop of high-proportion renewable energy integration, the uncertainty in wind and solar power output poses challenges to the safe and stable operation of the power grid. Compressed air energy storage (CAES), with its high flexibility and adjustable rotational speed characteristics, can rapidly respond to grid power commands and effectively suppress fluctuations. However, traditional control strategies struggle to adapt to the multi-variable and complex operating conditions of CAES, resulting in suboptimal power tracking performance. To address this issue, a power tracking control strategy for CAES based on neural network PID is proposed. [Methods] First, a turbine-side dynamic model is constructed to characterize the dynamic relationships between reactive power and excitation voltage, as well as between active power and mass flow rate. Subsequently, a control framework based on neural network PID is designed to achieve precise power tracking control without overshoot through online adaptive parameter tuning. Finally, the response performance of this strategy in power tracking is validated through MATLAB/Simulink simulation. [Results] Compared to traditional PID control, dynamic active power deviation decreases by 22.7% and reactive power deviation reduces to 0.66% under continuous load fluctuations.[Conclusions] The proposed strategy achieves rapid, overshoot-free tracking of power command signals, providing a practical power tracking control solution for power grids with high-proportion renewable energy.Keywords：可再生能源;电网;压缩空气储能;功率跟踪;神经网络;PID控制29|12|0<HTML><L-PDF><Meta-XML>Updated：2026-05-15
