基于三维特征向量的非侵入式电热负荷细分算法

非侵入式负荷监测分解（NILMD）技术是当前居民用能服务深化提升和电力供需互动的重要数据获取手段，然而当前工程上应用广泛的事件驱动型NILMD技术一直无法准确细化分解电热负荷。针对这一问题，文中提出了一种基于三维特征向量的典型电热负荷细化分解算法。首先，基于有功、无功功率和电流谐波等电气负荷特征采用事件检测方法提取电热事件，在有功功率的基础上，引入运行时长、频繁启停次数等非电气负荷特征共同构建三维特征向量电器模型。然后，采用序贯覆盖法设计典型电热负荷细化分解命题学习规则和细化分解算法。最后，基于实证实验数据进行分解验证，发现4种典型电热负荷的细化分解准确率超过85%。实验结果表明，文中所提典型电热负荷细化分解算法有效地提高了4种典型电热负荷分解的准确率。

Non-intrusive load detailed disaggregation algorithm for electrothermal load based on three dimensional characteristics vector

Non-intrusive load monitoring and disaggregation (NILMD) technology is an important data acquisition method for deep improvement of residents'' energy service and the interaction between power supply and demand. However,it is unable to disaggregate accurately the electrothermal load by the NILMD algorithm of edge detection which is widely used in current engineering. In order to solve this problem,a NILMD algorithm for typical electrothermal load based on three dimensional characteristics vector is proposed in this paper. Firstly,the edge detection algorithm is used to extract the electrothermal events through active power,reactive power and current harmonic,and the three dimensional characteristics vector model is constructed together with the non-electric characteristics such as running duration,start and stop times based on active power. Then,the learning rules and algorithm for typical electrothermal load detailed disaggregation are designed by sequential covering. Finally,the disaggregation accuracy of the electrothermal load is more than 85% based on experimental verification. Experimental results show that the NILMD algorithm for typical electrothermal load proposed in this paper effectively improve the disaggregation accuracy of four typical electrothermal load.