米特非侵入式负荷识别算法训练测试数据集

非侵入式负荷识别（NILM）的正常运行数据体系完整记录常见家用电器在无故障状态下的全周期用电行为，涵盖设备启动、稳态运行（如电热设备的恒定功率输出）及待机（低功耗休眠）等典型阶段，采集包含电流电压波形、功率波动、谐波分布等稳态 / 暂态特征的电气指纹。数据覆盖多类型设备（如制冷、厨房、照明器具）的正常工况，记录不同负载的标准用电模式。构建标准化的设备正常用电行为数据库，为负荷分解算法提供纯净训练样本，支撑设备识别（如区分电饭煲加热与保温状态）、能耗统计（稳态功率积分）及基线模型建立（如家庭负荷正常波动范围），确保非侵入式监测中设备状态识别的准确性和异常检测的可靠性。

The normal operation dataset for Non-intrusive Load Monitoring (NILM) comprehensively records the full-cycle electricity consumption behaviors of common household appliances in fault-free states, covering typical stages including equipment startup, steady-state operation (e.g., constant power output of electric heating appliances), and standby (low-power dormant state). It collects electrical fingerprints that encompass steady-state and transient features such as current and voltage waveforms, power fluctuations, and harmonic distributions. The dataset covers normal operating conditions of multiple types of equipment including refrigeration appliances, kitchen appliances, and lighting fixtures, and records the standard electricity consumption patterns of different loads. By constructing a standardized database of normal electricity consumption behaviors of household appliances, it provides pure training samples for load disaggregation algorithms, supporting applications such as equipment identification (e.g., distinguishing between the heating and insulation modes of rice cookers), energy consumption statistics (steady-state power integration), and baseline model establishment (e.g., normal fluctuation ranges of household loads). This ensures the accuracy of equipment state identification and the reliability of anomaly detection in non-intrusive load monitoring.

搜集汇总

背景与挑战

背景概述

该数据集是由南京米特科技股份有限公司登记的非侵入式负荷识别算法训练测试数据集，主要用于家庭能源管理和智能电网调度等领域，包含多种家用电器在无故障状态下的全周期用电行为数据，数据格式为excel。