Data for: Multivariate Event Detection Methods for Non Intrusive Load Monitoring in Residential Buildings

The Power_time_series.txt file contains the 34 electrical features related to power that were computed using current and voltage acquisitions made using our own acquisition system based on an Arduino MKR Zero microcontroller with a sampling frequency of 6.25 kHz (see reference "Design of an electricity consumption measurement system for Non Intrusive Load Monitoring", IEEE IREC 2019 from the same authors). In the text file the power time series correspond to the active power P, its harmonic order Pk, where k is ranging between 1 and 15 and the sum of the harmonics PH. There are also the reactive power Q, its harmonic order Qk, where k is ranging between 1 and 15 and the sum of the harmonics QH. All these 34 features are ordered by columns as follows: P, P1, P2, P3, P4, P5, P6, P7, P8, P9, P10, P11, P12, P13, P14, P15, PH, Q, Q1, Q2, Q3, Q4, Q5, Q6, Q7, Q8, Q9, Q10, Q11, Q12, Q13, Q14, Q15, QH These power time series correspond to the same scenario of 12 appliances (microwave, a DVD player, a fan, a screen, a vacuum, a waffle iron, a hair dryer, an iron, a flat iron, a mixer, a CFL and a LED lamp) that are randomly switched on/off every 3 seconds during almost one hour and half. This results in almots 1200 events of diferrent appliances. The text file Events_time_stamping.txt contains the time stamps of each events that were hand-labelled after inspecting the signals. Timestamps of the labeled events were adjusted to match the transitions. The text file Events_labelling corresponds to the labelling of each events (which appliance is responsable of the event). Finally, the file MultivriateAbruptChangeDetectors.pdf is a repport containing the mathematicla derivation of each investigated detector in details.

`Power_time_series.txt` 文件包含34项电力相关电气特征，这些特征由我们基于Arduino MKR Zero微控制器自研的采集系统采集的电流与电压数据计算得到，采样频率为6.25 kHz（详见同作者团队发表于IEEE IREC 2019的论文《Design of an electricity consumption measurement system for Non Intrusive Load Monitoring》）。该文本文件中的电力时间序列涵盖有功功率P、1至15次谐波有功功率Pk以及总谐波有功功率PH；同时还包含无功功率Q、1至15次谐波无功功率Qk以及总谐波无功功率QH。上述34项特征按列顺序排列如下：P, P1, P2, P3, P4, P5, P6, P7, P8, P9, P10, P11, P12, P13, P14, P15, PH, Q, Q1, Q2, Q3, Q4, Q5, Q6, Q7, Q8, Q9, Q10, Q11, Q12, Q13, Q14, Q15, QH。
这些电力时间序列对应12种电器的统一实验场景：微波炉、DVD播放器、风扇、显示器、吸尘器、华夫饼机、吹风机、熨斗、平板熨斗、搅拌机、紧凑型荧光灯（CFL）以及LED灯。所有电器每3秒随机通断一次，实验时长近1.5小时，总计产生约1200个不同电器的通断事件。
`Events_time_stamping.txt` 文件存储了所有事件的时间戳，这些时间戳通过人工检视信号后标注，并经调整以匹配实际的通断时刻。
`Events_labelling` 文件则为每个事件标注了对应的触发电器。
`MultivariateAbruptChangeDetectors.pdf` 为一份研究报告，详细推导了本次研究中所有被考察的突变检测器的数学原理。