Data Requirements for Applying Machine Learning to Energy Disaggregation

Energy disaggregation, or nonintrusive load monitoring (NILM), is a technology for separating a household’s aggregate electricity consumption information. Although this technology was developed in 1992, its practical usage and mass deployment have been rather limited, possibly because the commonly used datasets are not adequate for NILM research. In this study, we report the findings from a newly collected dataset that contains 10 Hz sampling data for 58 houses. The dataset not only contains the aggregate measurements, but also individual appliance measurements for three types of appliances. By applying three classification algorithms (vanilla DNN (Deep Neural Network), ML (Machine Learning) with feature engineering, and CNN (Convolutional Neural Network) with hyper-parameter tuning) and a recent regression algorithm (Subtask Gated Network) to the new dataset, we show that NILM performance can be significantly limited when the data sampling rate is too low or when the number of distinct houses in the dataset is too small. The well-known NILM datasets that are popular in the research community do not meet these requirements. Our results indicate that higher quality datasets should be used to expedite the progress of NILM research.

Download Full-text

Energy Disaggregation Using Two-Stage Fusion of Binary Device Detectors

Energies ◽

10.3390/en13092148 ◽

2020 ◽

Vol 13 (9) ◽

pp. 2148 ◽

Cited By ~ 1

Author(s):

Pascal A. Schirmer ◽

Iosif Mporas ◽

Akbar Sheikh-Akbari

Keyword(s):

Power Consumption ◽

Electricity Consumption ◽

Estimation Accuracy ◽

Two Stage ◽

Data Set ◽

Second Stage ◽

Energy Disaggregation ◽

Stage Classification ◽

Load Monitoring ◽

Detection Score

A data-driven methodology to improve the energy disaggregation accuracy during Non-Intrusive Load Monitoring is proposed. In detail, the method uses a two-stage classification scheme, with the first stage consisting of classification models processing the aggregated signal in parallel and each of them producing a binary device detection score, and the second stage consisting of fusion regression models for estimating the power consumption for each of the electrical appliances. The accuracy of the proposed approach was tested on three datasets—ECO (Electricity Consumption & Occupancy), REDD (Reference Energy Disaggregation Data Set), and iAWE (Indian Dataset for Ambient Water and Energy)—which are available online, using four different classifiers. The presented approach improves the estimation accuracy by up to 4.1% with respect to a basic energy disaggregation architecture, while the improvement on device level was up to 10.1%. Analysis on device level showed significant improvement of power consumption estimation accuracy especially for continuous and nonlinear appliances across all evaluated datasets.

Download Full-text

Effect of Sampling Rate on Photovoltaic Self-Consumption in Load Shifting Simulations

Energies ◽

10.3390/en13205393 ◽

2020 ◽

Vol 13 (20) ◽

pp. 5393

Author(s):

Philippe Voinov ◽

Patrick Huber ◽

Alberto Calatroni ◽

Andreas Rumsch ◽

Andrew Paice

Keyword(s):

Sampling Rate ◽

Smart Meter ◽

Data Sampling ◽

Load Shifting ◽

Consumption Data ◽

Data Sampling Rate ◽

Potential Improvement ◽

Using Data ◽

The Uk ◽

The Impact

Grid-connected photovoltaic (PV) capacity is increasing and is currently estimated to account for 3.0% of worldwide energy generation. One strategy to balance fluctuating PV power is to incentivize self-consumption by shifting certain loads. The potential improvement in the amount of self-consumption is usually estimated using smart meter and PV production data. Smart meter data are usually available only at sampling frequences far below the Nyquist limit. In this paper we investigate how this insufficient sampling rate affects the estimated self-consumption potential of shiftable household appliances (washing machines, tumble dryers and dishwashers). We base our analyses on measured consumption data from 16 households in the UK and corresponding PV data. We found that the simulated results have a marked dependence on the data sampling rate. The amount of self-consumed energy estimated with data sampled every 10 min was overestimated by 30–40% compared to estimations using data with 1 min sampling rate. We therefore recommend to take this factor into account when making predictions on the impact of appliance load shifting on the rate of self-consumption.

Download Full-text

Impact Analysis of Guidance Instrument Error Separation Accuracy on Sampling Rate of Inertial Navigation Equipment

Applied Mechanics and Materials ◽

10.4028/www.scientific.net/amm.341-342.880 ◽

2013 ◽

Vol 341-342 ◽

pp. 880-886

Author(s):

Wen Jun Wang ◽

Xiao Jun Duan ◽

Ju Bo Zhu

Keyword(s):

Impact Analysis ◽

Sampling Rate ◽

Inertial Navigation ◽

Optimal Sampling ◽

Data Sampling ◽

Instrument Error ◽

Error Separation ◽

Data Sampling Rate ◽

Simulation Results ◽

Navigation Equipment

Based on the linear model of guidance instrument error separation, study on the separation accuracy affected by data sampling rate of inertial navigation equipment. First, theoretically proved that the higher data sampling rate is, the higher separation accuracy we can get. Second, a method for determining the optimal sampling rate is presented, whose idea is from the model itself. At last, the simulation results can verify the above two conclusions.

Download Full-text

Effects of Data Sampling Rate on Pattern Recognition Analysis of FT-IR Interferograms

Applied Spectroscopy ◽

10.1366/0003702924444443 ◽

1992 ◽

Vol 46 (1) ◽

pp. 49-59 ◽

Cited By ~ 4

Author(s):

Gary W. Small ◽

Scott E. Carpenter

Keyword(s):

Pattern Recognition ◽

Sampling Rate ◽

Data Sampling ◽

Monitoring Applications ◽

Data Sampling Rate ◽

Ft Ir ◽

Pattern Recognition Analysis ◽

Improved Performance ◽

Mixture Samples ◽

The Impact

Fourier transform infrared (FT-IR) data of pure and mixture samples of benzene and nitrobenzene are used to investigate and improve methods for interferogram-based qualitative analyses. For use in dedicated monitoring applications, the methodology employed is based on the application of pattern recognition analysis to short, digitally filtered interferogram segments. In the work described here, the impact of the interferogram data sampling rate on the analysis is studied. The results of this study indicate that optimal pattern recognition prediction performance is achieved by use of linear discriminants developed from faster sampled interferogram data. These findings suggest that improved performance can be obtained in FT-IR monitoring applications through the use of spectrometer designs based on a decreased interferogram scan length, coupled with faster sampling electronics.

Download Full-text

A Low Rate Energy Disaggregation using Non-Intrusive Load Monitoring

International Journal of Innovative Technology and Exploring Engineering - Special Issue ◽

10.35940/ijitee.k2287.0981119 ◽

2019 ◽

Vol 8 (11) ◽

pp. 4204-4209

Keyword(s):

Energy Conservation ◽

Low Cost ◽

Sampling Rate ◽

Energy Usage ◽

Energy Meter ◽

High Sampling Rate ◽

Energy Disaggregation ◽

Load Monitoring ◽

The Cost ◽

Conservation And Management

Nowadays, Energy conservation and management are a must practice due to the exponentially increasing energy usage. One solution for providing for energy conservation is appliance load monitoring. Load monitoring approach should be simple and of low cost in order to be massively deployable. Non-Intrusive load monitoring is a better approach since it can disaggregate energy at the cost of single energy meter. A low sampling rate energy meter incurs low cost compared to a high sampling rate energy meter. In this paper a less complex, low cost energy disaggregation approach has been proposed

Download Full-text

094 Towards objective testing in parkinson’s disease: a systematic review of the literature looking at assessment of postural sway

Journal of Neurology Neurosurgery & Psychiatry ◽

10.1136/jnnp-2019-anzan.82 ◽

2019 ◽

Vol 90 (e7) ◽

pp. A30.2-A30

Author(s):

Wenbo Ge ◽

Deborah Apthorp ◽

Christian J Lueck ◽

Hanna Suominen

Keyword(s):

Systematic Review ◽

Parkinson’S Disease ◽

Parkinson's Disease ◽

Postural Sway ◽

Sampling Rate ◽

Objective Measure ◽

Data Sampling ◽

Data Sampling Rate ◽

Objective Marker

IntroductionParkinson’s Disease (PD) is associated with increased mortality and reduced quality of life. There is currently no accurate objective measure for use in diagnosis or assessment of severity. Analysis of postural sway may help in this regard. This systematic review aimed to assess the effectiveness of the various features currently used to analyse postural sway.MethodsFive databases were searched for articles that examined postural sway in both PD patients and controls. An effect size (ES) was derived for every feature reported in each article. The most effective features and feature-families were determined, along with the influence on these measures of data sampling rate and experimental condition.Results441 papers were initially retrieved, of which 31 met the requirements for analysis. The most commonly-used features were not the most effective (e.g. PathLength had an ES of 0.47 while TotalEnergy had an ES of 1.78). Decreased sampling rate was associated with decreased ES (e.g. ES of PathLength lowered from 1.12 at 100 Hz to 0.40 at 10 Hz). Being off medication was associated with a larger ES (e.g. ES of PathLength was 0.21 on medication and 0.83 off medication).ConclusionsSome measures of postural sway are better able to distinguish PD patients from controls than others. ES is enhanced by using a higher sampling rate and studying patients off medication. These results will inform future studies looking at postural sway in PD and contribute to the aim of finding an objective marker of the disease.

Download Full-text

Subtask Gated Networks for Non-Intrusive Load Monitoring

Proceedings of the AAAI Conference on Artificial Intelligence ◽

10.1609/aaai.v33i01.33011150 ◽

2019 ◽

Vol 33 ◽

pp. 1150-1157 ◽

Cited By ~ 4

Author(s):

Changho Shin ◽

Sunghwan Joo ◽

Jaeryun Yim ◽

Hyoseop Lee ◽

Taesup Moon ◽

...

Keyword(s):

Electricity Consumption ◽

Seminal Paper ◽

Performance Improvements ◽

Energy Disaggregation ◽

Load Monitoring ◽

Significant Performance ◽

Standby Power ◽

Measurement Devices ◽

Classification Probability ◽

The Cost

Non-intrusive load monitoring (NILM), also known as energy disaggregation, is a blind source separation problem where a household’s aggregate electricity consumption is broken down into electricity usages of individual appliances. In this way, the cost and trouble of installing many measurement devices over numerous household appliances can be avoided, and only one device needs to be installed. The problem has been well-known since Hart’s seminal paper in 1992, and recently significant performance improvements have been achieved by adopting deep networks. In this work, we focus on the idea that appliances have on/off states, and develop a deep network for further performance improvements. Specifically, we propose a subtask gated network that combines the main regression network with an on/off classification subtask network. Unlike typical multitask learning algorithms where multiple tasks simply share the network parameters to take advantage of the relevance among tasks, the subtask gated network multiply the main network’s regression output with the subtask’s classification probability. When standby-power is additionally learned, the proposed solution surpasses the state-of-the-art performance for most of the benchmark cases. The subtask gated network can be very effective for any problem that inherently has on/off states.

Download Full-text

The ENERTALK dataset, 15 Hz electricity consumption data from 22 houses in Korea

Scientific Data ◽

10.1038/s41597-019-0212-5 ◽

2019 ◽

Vol 6 (1) ◽

Cited By ~ 5

Author(s):

Changho Shin ◽

Eunjung Lee ◽

Jeongyun Han ◽

Jaeryun Yim ◽

Wonjong Rhee ◽

...

Keyword(s):

Energy Consumption ◽

Sampling Rate ◽

Electricity Consumption ◽

Consumption Pattern ◽

Additional Information ◽

Consumption Data ◽

Energy Disaggregation ◽

Residential Electricity ◽

Residential Electricity Consumption

Abstract AMI has been gradually replacing conventional meters because newer models can acquire more informative energy consumption data. The additional information has enabled significant advances in many fields, including energy disaggregation, energy consumption pattern analysis and prediction, demand response, and user segmentation. However, the quality of AMI data varies significantly across publicly available datasets, and low sampling rates and numbers of houses monitored seriously limit practical analyses. To address these challenges, we herein present the ENERTALK dataset, which contains both aggregate and per-appliance measurements sampled at 15 Hz from 22 houses. Among the publicly available datasets with both aggregate and per-appliance measurements, 15 Hz was the highest sampling rate. The number of houses (22) was the second-largest where the largest one had a sampling rate of 1 Hz. The ENERTALK dataset is also the first Korean open dataset on residential electricity consumption.

Download Full-text