摘要
为提高含有车网互动(vehicle-to-grid,V2G)的储能台区负荷重过载风险评估预警与治理管控的能力,提出一种基于负荷预测的储能台区重过载预警方法,同时结合预警结果构建分时电价与V2G有序充放电参与台区重过载治理。首先,提出一种基于扩展时序注意力与多时间尺度交叉注意力改进Transformer的短期负荷预测模型,利用时序卷积与注意力计算结合,提高局部特征与短期时间依赖的捕捉能力,引入交叉注意力机制实现多时间窗口的时序卷积进行特征扩展与融合关联,在Transformer处理长期依赖的基础上增强精细化时间窗口特征信息提取,提升负荷时间序列中的周期趋势与局部波动的提取精度,分别对台区用电负荷和电动汽车充电负荷进行精确预测。然后,基于负荷预测结果,划分重过载阈值并预警,基于蒙特卡洛方法模拟电动车充放电行为,并根据重过载时段优化求解分时电价的峰谷时段,引导建立电动汽车V2G有序充放电模式,对可能出现的台区重过载状况进行治理。最后,基于实际台区数据的测试结果,表明所提方法在台区重过载风险评估的准确性和治理方案的有效性方面表现出显著优势。
In order to improve the ability of load heavy overload risk assessment warning and governance control of energy storage stations containing V2G,a heavy overload warning method based on load prediction is proposed for energy storage stations,and at the same time combining the warning results to construct time-sharing tariffs and V2G orderly charging and discharging to participate in the heavy overload governance of the stations.Firstly,a short-term load prediction model based on extended temporal attention and multi-timescale cross-attention is proposed to improve Transformer,which utilizes the combination of temporal convolution and attention computation to improve the ability to capture local features and short-term temporal dependencies,and introduces the cross-attention mechanism to realize the temporal convolution of multiple time windows for feature expansion and fusion association,which enhances the fine-grained prediction based on the long-term dependencies in Transformer.On the basis of long-term dependency processing,we enhance the extraction of refined time window feature information,improve the extraction accuracy of periodic trend and local fluctuation in the load time series,and accurately predict the power consumption load and EV charging load of the station area respectively;based on the load prediction results,we classify the heavy overload thresholds and provide early warnings,and simulate the charging and discharging behaviors of EVs based on the Monte Carlo method,and optimize the solution of the peak and valley times of the timeshare tariffs based on the heavy overload periods,so as to guide the power supply to the peak and valley times of the electricity tariff.The peak and valley periods of the tariff are optimized according to the heavy overload periods,which guides the establishment of V2G orderly charging and discharging modes for EV and manages the possible heavy overload conditions in the station area.The experimental results based on the actual station data show that the method in this paper shows significant advantages in terms of the accuracy of the risk assessment of heavy overload and the effectiveness of the management scheme in the station area.
作者
沈润
程晴
陈业策
卢志辉
莫新
吴莉琳
何海鹏
陈奇智
SHEN Run;CHENG Qing;CHEN Yece;LU Zhihui;MO Xin;WU Lilin;HE Haipeng;CHEN Qizhi(Zhanjiang Wuchuan Power Supply Bureau of Guangdong Power Grid Co.,Ltd.,Zhanjiang 524500,China)
出处
《供用电》
北大核心
2025年第6期22-30,共9页
Distribution & Utilization
基金
国家自然科学青年基金项目(52307199)
广东电网有限责任公司科技项目(0308002023030101WC00157)。
