非侵入式负荷监测与分解的事件检测算法研究

Research on Event Detection Algorithm of Non-Intrusive Load Monitoring and Decomposition

摘要: 事件检测算法在非侵入式负荷监测与分解应用中占有很重要的地位，为了提高负荷特征提取工作的有效性，本文提出了一种改进的事件检测方法，先实现一步负荷分离的波形处理工作，再针对留下的负荷波形进行检测计算,判断负荷启动波形的暂态阶段起始点和结束点位置，这方法提高了阈值的筛选效率，确保针对同一负荷所截取到的暂态和稳态波形的一致性。经过实际负荷启动波形数据的分析验证，当前事件检测算法能取得良好的检测效果。

Abstract: Event detection algorithm plays an important role in non-invasive load monitoring and decomposition application. An improved event detection method is proposed in this paper to improve the effectiveness of load feature extraction. Firstly, the waveform processing of load separation is realized, and then the remaining load waveform is detected and calculated to get the starting point and ending point of transient phase of the load starting waveform. This method improves the efficiency of threshold screening and ensures the consistency of transient and steady-state waveforms intercepted for the same load. It is verified with the actual load startup data that the event detection algorithm can achieve good detection effect.