Table of Content

25 March 2025, Volume 47 Issue 3

Previous Issue   

Load Optimization and Control New Power System Scheduling based on AI Load Modeling and Potential Analysis

Load Optimization and Control

Review of research on situation awareness in resilient power systems

ZHANG Lu, LU Anshan, HUA Yang, LIU Tianhao, ZHANG Dongdong

2025, 47(3):  1-14.  doi:10.3969/j.issn.2097-0706.2025.03.001

Asbtract ( 55 )   HTML ( 3)   PDF (1426KB) ( 40 )  

Figures and Tables | References | Related Articles | Metrics

As global energy demand continues to rise， achieving a green and low-carbon energy transition has become the core task of energy systems. Ensuring the stability and security of power systems is particularly critical during this transition. Facing the challenges of natural disasters and human factors， the development of resilient construction in power systems under extreme events is of paramount importance. In this review， the definition， characteristic curves， resilience assessment methods of resilient power systems and the overall framework of cyber-physical power systems（CPPS） are explored in depth. Based on a review of the literature， the situation awareness（SA） technology in resilient power systems is comprehensively analyzed， especially grid SA and network SA. Combined with research results of the coupling relationship of CPPS， the SA framework of CPPS is constructed， and the collaborative SA of information side and physical side is studied to improve the resilience of power system in the face of climate change and energy challenges. The future research focus is also proposed.

Quantifying method for buildings' demand response potential applied to market access condition determination

DI Liang, DONG Jie, YAN Xinyue, ZHEN Cheng, TIAN Zhe, NIU Jide

2025, 47(3):  15-22.  doi:10.3969/j.issn.2097-0706.2025.03.002

Asbtract ( 44 )   HTML ( 2)   PDF (1197KB) ( 19 )  

Figures and Tables | References | Related Articles | Metrics

Buildings are important potential resources for electricity demand response （DR）.But the features of the power grid including lack of demand response data， diversity of physical structures and building envelopes， and complexity of consumption behaviors and meteorological conditions make it difficult to accurately and quickly calculate the DR potential of buildings，hindering the development of the DR market. Most traditional potential quantification methods are applicable to a single building. They can hardly support the judgement on the market access conditions of diverse potential users， or select the qualified target users with a review on the reliability of the declared information. To solve the problems above， a data-driven method based on the quantification of buildings' DR potential is proposed. Firstly， the input and output variables are set by considering the factors influencing the building DR potential and the requirements for the declared information. Then， the tool chain called EnergyPlus-JEPlus-Eppy is used to generate a complete dataset， improving the model generalization. Finally， a data-driven modelling approach is used to develop a potential quantification model to achieve rapid quantification of DR potential. The effectiveness of the method is verified by a typical office building in Nanjing by choosing global temperature regulation as the peak-shaving means. The results of the case show that the proposed model can keep the mean squared error （MSE） of the predicted potential within 0.009 6 with a coefficient of determination（R²） higher than 0.9. The method enables accurate quantification on DR potential for different buildings in various scenarios.

Non-intrusive load identification for public buildings based on MSCNN-BiGRU-MLP model

YANG Lijie, DENG Zhenyu, CHEN Zuoshuang, HUANG Chao, JIANG Meihui, ZHU Hongyu

2025, 47(3):  23-31.  doi:10.3969/j.issn.2097-0706.2025.03.003

Asbtract ( 37 )   HTML ( 1)   PDF (1218KB) ( 16 )  

Figures and Tables | References | Related Articles | Metrics

In the energy management of public buildings， load identification plays a critical role in optimizing energy utilization and reducing energy consumption. Traditional load monitoring methods are primarily intrusive， relying on hardware equipment or macro-level load characteristics， which fail to meet the refined management requirements of modern intelligent buildings and smart cities. To address the challenges posed by the diversity and uncertainty of public building loads， a non-intrusive load identification method was proposed based on multi-scale convolutional neural network （MSCNN）， bidirectional gated recurrent unit（BiGRU）， and multilayer perceptron（MLP）. The model integrated voltage-current（V-I） trajectory features， power features， and harmonic features to achieve classification and identification of typical socket-based loads in public buildings. MSCNN was employed to extract V-I trajectory features， capturing stable and "fingerprint-like" characteristics of equipment during operation. BiGRU was utilized for time-series modeling of power and harmonic features， revealing the dynamic characteristics of load signals. MLP was then applied to classify the fused features. Experiments on various common public building loads validated the effectiveness of the proposed model. The results showed that the MSCNN-BiGRU-MLP model achieved a load identification accuracy of 0.917 1， accurately identifying load types and maintaining high robustness under dynamic feature changes and high-frequency noise conditions.

New Power System Scheduling based on AI

Review on the application of artificial intelligence in data mining and wind and photovoltaic power forecasting

ZHANG Dongdong, SHAN Linke, LIU Tianhao

2025, 47(3):  32-46.  doi:10.3969/j.issn.2097-0706.2025.03.004

Asbtract ( 89 )   HTML ( 5)   PDF (1386KB) ( 53 )  

Figures and Tables | References | Related Articles | Metrics

As global demand for renewable energy continues to surge， efficiently and intelligently managing and forecasting renewable energy generation has become a pivotal research objective in energy sector. Applications of artificial intelligence （AI） technologies in the multi-dimensional data processing and intelligent forecasting of renewable energy generation are explored， focusing on its role in handling complex and highly variable data. First，the role of multi-dimensional feature mining techniques in processing wind and solar energy generation data from the perspective of meteorological conditions and spatiotemporal features is studied. Subsequently， a systematic analysis on intelligent forecasting techniques applied across different spatiotemporal scales and scenarios is offered， with particular emphasis on its usage in machine learning and deep learning models. These models have gained significant attention for their outstanding performance in dealing with nonlinear and high-dimensional data. Thorough reviews on the latest research findings demonstrate the substantial benefits of these AI technologies in enhancing the accuracy and efficiency of wind and solar energy generation forecasts. Additionally， it delves into the strengths and limitations of existing technologies and their development directions， particularly emphasizing the importance of improving the robustness， real-time processing capabilities， and adaptability of intelligent forecasting models in various scenarios. This study provides theoretical insights and practical guidance for advancing the development of renewable energy.

Key technologies for load forecasting in new power systems and their applications in diverse scenario

ZHANG Dongdong, LI Fangning, LIU Tianhao

2025, 47(3):  47-61.  doi:10.3969/j.issn.2097-0706.2025.03.005

Asbtract ( 84 )   HTML ( 8)   PDF (1269KB) ( 44 )  

Figures and Tables | References | Related Articles | Metrics

To achieve the goal of "dual carbon"， the new power system is transitioning towards greening， intelligence， and diversity. Load forecasting is crucial for ensuring the safe， economic， and reliable operation of the new power system. While traditional statistical methods perform well in forecasting load data with clear patterns， the high proportion of renewable energy and the stochastic user load in new power systems pose significant challenges to these methods. Artificial intelligence technologies， particularly machine learning and deep learning， have become research hotspots due to their advantages in dealing with complex data and extracting patterns， effectively improving the accuracy and robustness of load forecasting. In this context， load forecasting methods based on mathematical and statistical principles are reviewed and their limitations are discussed in this study. The latest advancements in applications of AI techniques in load forecasting are summarized， and the characteristics of traditional machine learning， deep learning， and hybrid forecasting models are analysed. Technical challenges of load forecasting and key applications under these five scenarios are summarized and discussed：regional system-level load forecasting， net load forecasting under high proportion of renewable energy scenarios，integrated energy system load forecasting in multi-type heterogeneous energy complementary scenarios， building load forecasting， and electric vehicle load forecasting. The future directions of load forecasting technologies are forecasted.

Detection and repair of abnormal load data of public buildings based on MDLOF-iForest and M-KNN-Slope

LIU Yining, CHEN Baian, DU Pengcheng, LIN Xiaogang, JIANG Meihui

2025, 47(3):  62-72.  doi:10.3969/j.issn.2097-0706.2025.03.006

Asbtract ( 34 )   HTML ( 1)   PDF (1149KB) ( 15 )  

Figures and Tables | References | Related Articles | Metrics

In research on energy consumption of public buildings， making the detection and repair of abnormal load data an indispensable part of data processing. To address the limitations of existing methods， a method based on the Mahalanobis distance-based local outlier factor-isolation forest （MDLOF-iForest） algorithm and the modified K-nearest neighbors-slope （M-KNN-Slope） algorithm was proposed.The MDLOF-iForest algorithm incorporated Mahalanobis distance into the traditional local outlier factor algorithm， improving the models ability to perceive correlations between data features. Meanwhile， by combining the advantages of MDLOF algorithm and iForest algorithm， it enabled rapid and accurate detection of abnormal data. The M-KNN-Slope algorithm used neighbors with similar load trend line characteristics of abnormal data and normal data to obtain the weighted average values of similar trend line slopes， completing the repair of abnormal data and reducing reliance on sample data. Verification was conducted using load data from an office public building and a commercial public building in Nanning， from August to November 2024. The results showed that approximately 90% of the repaired data had a difference of less than 10% compared to the correct data. Compared with conventional algorithms，the M-KNN-Slope algorithm could obtain more data with errors within 5%. Extreme gradient boosting， long short-term memory network， backpropagation neural network， and support vector machine were used to predict the data before and after repair. The root mean square values decreased by 5.02% to 17.83%， and the absolute mean errors decreased by 2.44% to 13.34%.

Load Modeling and Potential Analysis

Intra-day multi-time scale rolling optimization scheduling of mine integrated energy system considering integrated demand response

JIANG Meihui, XU Zhenjiang, NIU Tongke, ZHU Hongyu, LI Xiang

2025, 47(3):  73-83.  doi:10.3969/j.issn.2097-0706.2025.03.007

Asbtract ( 44 )   HTML ( 3)   PDF (1122KB) ( 20 )  

Figures and Tables | References | Related Articles | Metrics

To address the challenges of high energy consumption， low resource utilization， and significant uncertainty in source-load demand in mining areas， this study established a mine integrated energy system （MIES） based on coal mines. The system incorporated a demand response mechanism and a multi-time scale rolling optimization strategy to enhance economic operational efficiency. Firstly， a refined integrated demand response model was constructed for power and thermal loads on the demand side to maximize the schedulability and flexibility of various loads in mining areas. Subsequently， integrating a multi-time scale optimization strategy， an intra-day multi-time scale optimization model was developed， aiming to minimize deviations from the day-ahead plan. The model set each optimization cycle to 4 h and performed optimization every 15 min， using this rolling optimization strategy to effectively adjust the day-ahead scheduling plan. Finally， simulations and comparative analyses of different case studies demonstrated the effectiveness of the integrated demand response and multi-time scale optimization strategies in promoting low-carbon operation and managing uncertainty in MIES.

Linear fitting model for wind power curves based on density correction

ZHU Dongjie, LYU Kunye, SONG Changhong, JIANG Meihui, LI Zhijiu

2025, 47(3):  84-91.  doi:10.3969/j.issn.2097-0706.2025.03.008

Asbtract ( 36 )   HTML ( 3)   PDF (1691KB) ( 20 )  

Figures and Tables | References | Related Articles | Metrics

To address the issue that traditional wind power curve models fail to fully consider the effect of meteorological factors， resulting in reduced model accuracy， a new linear fitting model for wind power curves named LI-DASW is proposed with the inclusion of density correction. A calculation model for air density was developed based on meteorological factors such as temperature， pressure， and humidity， as well as a density-corrected wind speed strategy. It reflected the effect of meteorological changes on the wind power curve while maintaining the model's single-input and single-output characteristics. The original dataset was replaced with the first moment to reduce redundant calculations and improve modeling efficiency. A linear interpolation model was constructed using the first moment as interpolation points， effectively avoiding the Runge's phenomenon caused by higher-order polynomial fitting and enhancing the model's adaptability. The case study analysis results of two wind farms demonstrated that the LI-DASW model significantly outperformed traditional methods. Compared to the Bin method， the model's root mean square error（RMSE） reduced by 14.42% and 10.16%， and the mean absolute error（MAE） decreased by 15.63% and 9.48%， respectively. Compared to the polynomial method， the RMSE decreased by 20.33% and 7.66%， and the MAE improved by 18.15% and 8.06%. Compared to the linear interpolation method， the reductions in RMSE and MAE remained stable between 6.19% to 7.37%. Additionally， the modeling efficiency improved by over 84.81% compared to the polynomial model.

Three-phase unbalance optimization of distribution network considering load demand response

CUI Zaiyue, YANG Yang, WANG Lidi

2025, 47(3):  92-101.  doi:10.3969/j.issn.2097-0706.2025.03.009

Asbtract ( 44 )   HTML ( 2)   PDF (1938KB) ( 16 )  

Figures and Tables | References | Related Articles | Metrics

The increasing penetration of new energy generation， controllable loads， and energy storage in the distribution network has posed challenges to the safe and stable operation of power grids， making it more difficult to analyze power flow optimization and the three-phase imbalance issue. The concepts of load demand response and phase-switchable loads were introduced to construct a node load demand response model considering phase switching. By utilizing the characteristics of shiftable， interruptible， and phase-switchable loads， the power distribution among phase nodes was improved. At the same time， a node energy storage operation model was constructed to further improve the operation indexes of the three-phase distribution network. In the Matlab environment， the model solving variables were defined， the convergence accuracy and other parameters were set， the CPLEX solver was invoked to solve the optimal power flow model of the three-phase distribution network， and different scenarios were set for comparative analysis. The data analysis showed that the multi-factor flexible adjustment could effectively reduce the three-phase imbalance indexes of the distribution network in the IEEE 33 example system.