基于状态特征聚类的非侵入式负荷事件检测方法

Non-Intrusive Load Event Detection Method Based on State Feature Clustering

针对传统非侵入式暂态事件检测方法局限于单一电气特征量,易出现漏检或误检,难以准确感知负荷事件的问题,该文利用负荷事件发生时特征空间内产生状态域转移的特性,提出状态特征聚类的非侵入式负荷事件检测方法。该方法通过滑动窗差值计算,搜索并确定初始聚类点,然后利用均值漂移(Mean-shift)算法进行状态聚类,并依据稳定状态的时域分布确定负荷事件发生点,实现负荷事件检测。最后,通过真实实验场景进行多种常用电器测试,并与现有一些算法进行对比,结果表明该文算法能够较为可靠地实现负荷事件检测,并为后续准确的负荷辨识奠定基础。

In view of the traditional noninvasive transient event detection methods are limited to a single electrical feature,they are prone to missed or false detection,and it is difficult to accurately perceive load events.This article uses the characteristics of state domain transition in the feature space when load events occur.A non-invasive load event detection method based on state feature clustering is proposed.In this method,the initial clustering points are searched and determined through the calculation of the difference value of sliding window,and then the Mean-shift algorithm is utilized for state clustering,and the load event occurrence points are determined according to the time domain distribution of the stable state,thus realizing load event detection.Finally,through the test of a variety of common electrical appliances in real experimental scenarios,and the comparison with some existing algorithms,the result shows that the proposed method can achieve load event detection more reliably,and lay the foundation for subsequent accurate load identification.