| 引用本文: | 谭伟涛,姚冰峰,郭大琦,马闯,麻吕斌,王朝亮,林振智.基于特征辨识和变分自编码器网络的工商业空调负荷辨识[J].电力自动化设备,2024,44(12):61-68. |
| TAN Weitao,YAO Bingfeng,GUO Daqi,MA Chuang,MA Lübin,WANG Chaoliang,LIN Zhenzhi.Identification of industrial and commercial air conditioning load based on feature recognition and variational auto-encoder network[J].Electric Power Automation Equipment,2024,44(12):61-68. |
|
| 摘要: |
| 空调负荷功率的准确计算是实现其需求侧管理的关键，为此，提出基于负荷曲线特征辨识和变分自编码器网络的工商业用户空调负荷辨识方法。针对用户的连续日负荷曲线，提出基于局部加权线性拟合和快速动态时间规整的负荷曲线形态相似度度量方法，以实现对负荷曲线形态特征的度量。提出基于点排序的聚类结构辨识算法的日负荷序列特征辨识方法，以实现对负荷曲线的分类。针对同一特征类型下的用户日负荷序列，提出基于变分自编码器网络的空调负荷辨识算法，以实现空调负荷功率的准确计算。以浙江某市的加工制造业和商业写字楼宇用户负荷数据验证本文所提方法的有效性。算例仿真结果表明，所提方法可以在无需电表高频采样数据、无须预先获取用户的用电设备信息和用电行为信息的条件下准确辨识用户空调负荷功率，为量化空调负荷参与需求响应的可调潜力提供了基础。 |
| 关键词: 空调负荷 工商业用户 负荷辨识 局部加权线性拟合 OPTICS算法 变分自编码器网络 |
| DOI：10.16081/j.epae.202411005 |
| 分类号:TM714 |
| 基金项目:国家自然科学基金联合基金资助项目(U2166206)；国网浙江省电力有限公司科技项目(B311HZ22000D) |
|
| Identification of industrial and commercial air conditioning load based on feature recognition and variational auto-encoder network |
|
TAN Weitao1, YAO Bingfeng2, GUO Daqi2, MA Chuang2, MA Lübin3, WANG Chaoliang4, LIN Zhenzhi1
|
|
1.School of Electrical Engineering, Zhejiang University, Hangzhou 310027, China;2.Hangzhou Power Supply Company of State Grid Zhejiang Electric Power Co.,Ltd.,Hangzhou 310000, China;3.Zhejiang Huayun Information Technology Company, Hangzhou 310000, China;4.State Grid Zhejiang Electric Power Co.,Ltd.,Hangzhou 310007, China
|
| Abstract: |
| The accurate calculation of air conditioning load power is crucial for implementing its demand-side management. For this purpose, an air conditioning load identification method for industrial and commercial users based on feature recognition and variational auto-encoder network is proposed. For continuous daily load curves of customers, a similarity measure of load curve shape based on locally weighted linear fitting and fast dynamic time warping is proposed to achieve the measurement of load curve shape features. An ordering points to identify the clustering structure-based algorithm is proposed to achieve the classification of load curves. For the daily load sequence of users under the same classification type, a variational auto-encoder network-based air conditioning load identification method is proposed to achieve accurate calculation of the power consumption for air conditioning loads. The effectiveness of the proposed method is verified by the power consumption data of processing and manufacturing industry and commercial office building users in a city of Zhejiang province. The simulation results show that the proposed method can effectively identify the user’s power consumption of air conditioning loads without the need of high-frequency sampling data of smart meters and obtaining the user’s electrical equipment information and electrical behavior in advance, which provides a basis for quantifying the users’ adjustable potential of air conditioning loads to participate in demand response. |
| Key words: air conditioning loads industrial and commercial users load identification local weighted linear fitting OPTICS algorithm variational auto-encoder network |
