%0 Journal Article %T 基于多元时序的虚拟电厂负荷基线与潜力评估
Evaluation of Load Baseline and Potential of Virtual Power Plants Based on Multivariate Time Series %A 殷少波 %A 王思成 %J Modeling and Simulation %P 525-536 %@ 2324-870X %D 2025 %I Hans Publishing %R 10.12677/mos.2025.144306 %X 为应对高比例可再生能源并网导致的电力系统灵活性不足问题，本研究提出一种面向虚拟电厂(VPP)的负荷基线估计与可调潜力评估方法。通过整合用户侧可调负荷，构建融合用电行为特征、环境变量与历史数据的多元时间序列模型，解决中小型用户负荷分散性带来的基线估计难题，并结合需求弹性系数建立调节潜力评估框架。基于华东地区数据的实验表明，相较于传统预测模型，所提方法显著提高了负荷基线预测精度与稳定性，尤其在负荷高峰时段表现出更强的鲁棒性。最后，进一步分析揭示了系统可调潜力的时变分布规律，通过精细化时序分析验证了实际负荷与基线差异的动态特征。研究成果为系统运营商制定差异化需求响应(DR)策略提供理论依据。
To address the flexibility challenges in power systems caused by high-penetration renewable energy integration, this study proposes a load baseline estimation and adjustable potential assessment methodology for virtual power plants (VPPs). By aggregating adjustable loads from user-side resources and constructing a multivariate time series model that integrates electricity consumption behavior, environmental variables, and historical data, the proposed approach resolves the baseline estimation challenges arising from the dispersed load characteristics of small-to-medium distributed users. A demand elasticity coefficient-based framework is further established to evaluate adjustable potential. Experimental validation using operational data from East China demonstrates that the proposed methodology significantly enhances the accuracy and stability of load baseline prediction, particularly exhibiting superior robustness during peak load periods compared to conventional forecasting models. Furthermore, time-varying distribution patterns of adjustable potential are revealed through refined temporal analysis, which validates the dynamic characteristics of deviations between actual and baseline loads. The research outcomes provide a theoretical foundation for system operators to formulate differentiated demand response (DR) strategies, thereby optimizing the dynamic scheduling of distributed resources and supporting the low-carbon transition of power systems. %K 虚拟电厂， %K 梯度提升回归树， %K 负荷基线估计， %K 需求响应
Virtual Power Plant %K Gradient Boosting Regression Tree %K Load Baseline Estimation %K Demand Response %U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=112075