say up front
This article mainly introduces the current public data sets, tools and others used in the field of non-intrusive load identification. If you need to read the paper and specific code implementation, see my previous article.
In addition, I have not used all the data sets, I have only used UK-DALE, so I don’t know how to deal with other data sets! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !
But if your hands-on ability is not enough, you can consider:
1. Use the nilmtk toolkit directly
I'll mark a few of the most common ones below
Updated: May 9, 2023 22:36:45
Public datasets:
1.REDD(The Reference Energy Disaggregation Data Set) (常用)
id:redd
password:disaggregatetheenergy
2. AMPds (commonly used)
AMPds (The Almanac of Minutely Power Dataset).
The converter that comes with nilmtk corresponds to the AMPds R2013 version
Besides that
AMPds2: The Almanac of Minutely Power dataset (Version 2)
3.CER_Electricity_Data
ISSDA | Commission for Energy Regulation (CER)
4.Umass Smart Data Set
Smart - UMass Trace Repository
5. REFIT (commonly used)
The finest granularity, 8s level
REFIT: Electrical Load Measurements — University of Strathclyde
6. TO BED
7.GREEND
GREEND download | SourceForge.net
8.ElectricityLoadDiagrams
OEDI: Commercial and Residential Hourly Load Profiles for all TMY3 Locations in the United States
no missing point
9. UK-DALE (commonly used)
UK Domestic Appliance-Level Electricity (UK-DALE) dataset | Jack Kelly
https://data.ukedc.rl.ac.uk/browse/edc/efficiency/residential/EnergyConsumption/Domestic
There are three versions
10.ECO data set
DSG - Research Project: ECO data set
11.HES(Household Electricity Study)
12.The tracebase data set
GitHub - areinhardt/tracebase: The tracebase appliance-level power consumption data set
13.ENERTALK
https://www.nature.com/articles/s41597-019-0212-5
Preprocessing and visualization code: GitHub - ch-shin/ENERTALK-dataset: The ENERTALK Dataset, 15 Hz Electricity Consumption Data from 22 Houses in Korea
14.BLUED
BLUED Dataset for Non-Intrusive Load Decomposition_Alex Ching Ho's Blog-CSDN Blog_blued Dataset
15. TWELVE
DEDDIAG, a domestic electricity demand dataset of individual appliances in Germany
16.PLAID (commonly used)
PLAID2018: PLAID 2018
PLAID 2017: PLAID 2017
PLAID 2014: PLAID 2014
17.MORED: A Moroccan Buildings’ Electricity Consumption Dataset
https://github.com/MOREDataset/MORED
paper:https://www.mdpi.com/1996-1073/13/24/6737
18.Residential Power Traces for Five Houses: the iHomeLab RAPT Dataset
paper: Data | Free Full-Text | Residential Power Traces for Five Houses: The iHomeLab RAPT Dataset
pre-process code: https://github.com/ihomelab/RAPT-dataset
dataset: Residential Power Traces for Five Houses: the iHomeLab RAPT Dataset | Zenodo
19.FIRED: A Fully-labeled hIgh-fRequency Electricity Disaggregation Dataset
20.RAE:The Rainforest Automation Energy Dataset for Smart Grid Meter Data Analysis
GitHub - smakonin/RAE.dataset: Scripts of the the Rainforest Automation Energy Dataset (RAE dataset)
21.COOLL:Controlled On/Off Loads Library, a Public Dataset of High-Sampled Electrical Signals for Appliance Identification
Link None.
22.IAWE:Indian Dataset for Ambient Water and Energy
Synthetic dataset:
As the name suggests, the power data here is artificially synthesized, which is different from the data collected by the electric meter above. It is generally used as an enhanced data set.
1.SHED
A Simulated High-frequency Energy Disaggregation dataset for commercial buildings
2.SynD(A Synthetic Energy Consumption Dataset for NILM)
code: GitHub - klemenjak/SynD: A Synthetic Energy Consumption Dataset for Non-Intrusive Load Monitoring
pdf: A synthetic energy dataset for non-intrusive load monitoring in households | Scientific Data
3. SmartSim
A Device Accurate Smart Home Simulator for Energy Analytics
GitHub - sustainablecomputinglab/smartsim
4.Device-Free User Activity Detection using Non-Intrusive Load Monitoring: A Case Study
pdf: https://www.areinhardt.de/publications/2020/Reinhardt_DFHS_2020.pdf
code: GitHub - klemenjak/antgen: The AMBAL-based NILM Trace generator (for NILMTK)
How does Load Disaggregation Performance Depend on Data Characteristics? Insights from a Benchmarking Study. (2020). PDF: https://www.areinhardt.de/publications/2020/Reinhardt_eEnergy_2020.pdf
Tools (frameworks, dataset conversion tools, etc.):
NILM-TK is an open source toolkit for non-intrusive load monitoring, specifically designed to compare energy decomposition algorithms in a reproducible manner. It is a toolkit toolkit made by Jack Kelly and others.
Paper address: NILMTK | Proceedings of the 5th international conference on Future energy systems
NILMTK:
Code: GitHub - nilmtk/nilmtk: Non-Intrusive Load Monitoring Toolkit (nilmtk)
Documentation: NILMTK Documentation
all the state-of-the-art algorithms for the task of energy disaggregation
GitHub - nilmtk/nilmtk-contrib
An evaluation framework for non-intrusive load monitoring algorithms
https://github.com/beckel/nilm-eval
Related papers on NILM evaluation indicators
《On Metrics to Assess the Transferability of Machine Learning Models in Non-Intrusive Load Monitoring》
《Machine learning approaches for non-intrusive load monitoring: from qualitative to quantitative comparation,Artificial Intelligence Review》
Feature Selection Related Papers in NILM
《Comprehensive feature selection for appliance classification in NILM》
DOI:10.1016/j.enbuild.2017.06.042
code:https://github.com/18D070001/Electrical-Devices-Identification-Model
Some extended applications of NILM:
Monitoring the Elderly Living Alone:
《Assessing Human Activity in Elderly People Using Non-Intrusive Load Monitoring》
DOI:10.3390/s17020351
Use NILM to realize the recognition of home activities:
Sampling frequency corresponds to different harmonic characteristics
Some websites about NILM (workshop, association, etc.)
The International Workshop on Non-Intrusive Load Monitoring (NILM)