Non-intrusive Load Monitoring Research Articles

Comparing four machine learning algorithms for household non-intrusive load monitoring

The combination of Machine Learning (ML), smart energy meters, and availability of household appliance energy profile data has opened new opportunities for Non-Intrusive Load Monitoring (NILM). However, the number of options makes it challenging in selecting optimal combinations for different energy applications, which requires studies to examine their trade-offs.This paper contributes one such study that investigated four established ML approaches – K Nearest Neighbour (KNN), Support Vector Machine (SVM), eXtreme Gradient Boosting (XGBoost) and Convolutional Neural Network (CNN) – and their performance in classifying appliance events from Alternating Current (AC) and Root Mean Square (RMS) energy data where the sampling frequency and training dataset set size was varied (10 Hz–1 kHz and 50–2000 examples per class, respectively). The computational expense during training, testing and storage was also assessed and evaluated with reference to real-world applications.The CNN classifier trained on AC data at 500 Hz and 11,000 examples gave the best F1-score 0.989 followed by the KNN classifier 0.940. The storage size required by the CNN models was 3̃MB, which is very close to fitting on cost-effective embedded system microcontrollers. This would prevent high-rate data needing to be sent to the cloud as analysis could be performed on edge computing Internet-of-Things (IoT) devices.

Non-intrusive load monitoring based on MoCo_v2, time series self-supervised learning

Traditional non-intrusive load monitoring (NILM) methods rely on massive historical labeled data. However, due to the privacy and high labeling cost of datasets, their generality and feasibility are limited, with poor performance in identifying devices with multiple states or similar features. To address this issue, this paper proposes an SSCL-LM framework based on temporal convolutional network (TCN), integrating contrastive self-supervised learning into NILM. Initially, through contrastive self-supervised pre-training, the framework learns well-represented load temporal characteristics from abundant unlabeled 1D power data. Then a small amount of labeled data is used to fine-tune the classifier to learn load categories represented by different temporal characteristics. Finally, test data is inputted into the model for load identification. Validating on the REDD dataset, the results demonstrate that with only 30% labeled data for fine-tuning, the proposed method achieves a 4.04% higher F1 score compared to traditional supervised methods. Moreover, utilizing only 1D power features, this method exhibits superior identification performance for devices with multiple states or similar features. Furthermore, this method enables performance transfer among loads of the same device with different parameters across different households, verifying its strong generalization and practicality.

Anomaly‐detection‐based learning for real‐time data processing in non‐intrusive load monitoring

AbstractA power system can be regarded as a cyber‐physical system with physical power networks and a cyber system based on increasing engagement with information communication technologies for smart grid functionalities for more efficient operations and control. Non‐intrusive load monitoring (NILM), an emerging smart‐grid technology, can be used to better understand the electricity usage profile and composition of smart meters using advanced data analysis algorithms. Although NILM enables various smart grid services, wider applications of NILM require addressing the challenges regarding cyber security and data privacy risks. Anomaly detection in appliance data is one of the most effective measures against potential cyber intrusions from a data perspective. This study proposes a framework of anomaly detection‐based learning algorithms to identify the anomalous periods of electricity loading data, which may be a subject for potential cyber‐attacks. Comparison studies with the hidden Markov model are performed to validate the proposed approaches. The simulation results show that these anomaly detection‐based learning algorithms work well and can precisely determine anomalous loading periods. Moreover, these trained models perform well on the testing dataset without prior knowledge of the data, providing the possibility of the real‐time assessment of power‐ loading states. The proposed framework can also be used to develop protective measures to ensure secure system operation and user data privacy.

Demand response of large-scale residential load to concentrated variable renewable energy

Large-scale concentrated Variable Renewable Energy (VRE) sources pose a challenge to the balance and security of electricity systems. Studies have shown that the demand side may offer a greater responsiveness based on the shiftability of loads. However, there is currently no known literature on the demand response (DR) of large-scale shiftable loads in the residential sector. Firstly, there is no sufficiently large database of residential appliance-level load. Secondly, scheduling a large number of small residential shiftable loads simultaneously is a major challenge. Ultimately, the DR of large-scale residential buildings encounters the challenge of ensuring sufficient backup shiftable load to effectively cope with real-time VRE fluctuations during the response period. To address these issues, this paper proposes an appliance-level load data generation method based on diffusion neural networks. Meanwhile, a clustering method based on Generalized End to End (GE2E) loss neural networks is proposed to solve the complexity obstacles in simultaneous load swarm scheduling. Thirdly, a two-stage scheduling approach is proposed to address the challenges in the large-scale DR period through constructing day-ahead and real-time response models. The simulation results show that large-scale residential shiftable loads have a strong response capability to VRE fluctuations, and their response cost is lower than the conventional fuel-fired-power-plant-based-response within a certain user preference cost range. In practice, the clustering and scheduling methods proposed in this paper can be applied after the non-intrusive load monitoring (NILM) method is deployed to decompose the load data.

An intelligent non-intrusive load monitoring model based on power encoding and convolutional state modules

Abstract Non-intrusive load monitoring (NILM) identifies device power consumption or on/off states solely based on total power data, which is highly valuable for consumers to understand their appliance usage behavior and take necessary measures to reduce energy consumption, especially for the benefit of energy consumers’ living production. However, a challenge faced by NILM is the tendency to focus excessively on power disaggregation while neglecting the disaggregation of on/off states, leading to lower classification accuracy, particularly owning to imbalanced states. This study proposes a model that integrates the power and on/off states to simultaneously disaggregate the power and device on/off states. The model comprises two main modules: a power encoding module for power disaggregation, and a convolutional state module (CSM) for on/off state disaggregation. The power encoding module utilizes BERT-LSTM and long short-term memory networks for initial energy disaggregation. In contrast, the CSM employs convolutional neural networks for device state disaggregation. The output of the power-encoding module is multiplied by the probability of on/off states to obtain the final power. The proposed model is evaluated using the REDD and UK-DALE datasets. Compared to the baseline models, the results show an improvement in the device state classification average accuracy from 0.948 to 0.957, and a decrease in the average error between the real power and disaggregated power from 26.356 W to 25.108 W. Additionally, real-world experiments conducted using the designed platform for collecting and disaggregating power data achieve an average accuracy of 0.997. The proposed model demonstrates competitiveness in the NILM field and underscores its significance in aiding energy-consumption reduction efforts.

A privacy-preserving distributed energy management framework based on vertical federated learning-based smart data cleaning for smart home electricity data

Nonintrusive load monitoring (NILM) is a part of home energy management systems ((H)EMSs), which can advance the systems to leverage and use gathered electricity data (load data) to achieve cost-effective load monitoring for efficient (residential) demand-side management (DSM). Load data monitored by existing (H)EMSs, whose load monitoring is based on intrusive load monitoring, can be used as a preliminary stage for NILM to be accelerated in energy management for the improvement of its practicality. However, they can be polluted. For example, power-consumption data gathered by appliance-level smart plugs may have an incorrect appliance label (a mislabeled appliance ID). Polluted data must be preprocessed through a data cleaning process before they are leveraged and used. Data-cleaning approaches that improve data quality should be developed considering a decentralized paradigm, because a centralized data-cleaning paradigm cannot be applied to future edge-based IoT applications. Additionally, preventing data leakage is paramount in data management for numerous field applications. This study develops a privacy-preserving distributed energy management framework based on vertical federated learning for smart data cleaning as a demonstrative application of smart home electricity data toward distributed NILM. The framework developed in this study with the advent of AI methodology can achieve smart data cleaning for further distributed NILM to be accelerated for its practical applications; its feasibility and effectiveness have been verified experimentally.

Transient dataset of household appliances with Intensive switching events

With the development of Non-Intrusive Load Monitoring (NILM), it has become feasible to perform device identification, energy consumption decomposition, and load switching detection using Deep Learning (DL) methods. Similar to other machine learning problems, the research and validation of NILM necessitate substantial data support. Moreover, different regions exhibit distinct characteristics in their electricity environments. Therefore, there is a need to provide open datasets tailored to different regions. In this paper, we introduce the Transient Dataset of Household Appliances with Intensive Switching Events (TDHA25). This dataset comprises switch instantaneous data from 10 typical household appliances in China. The TDHA dataset features a high sampling rate, accurate labelling, and realistic representation of actual appliance start-up waveforms. Additionally, appliance switching is achieved through precise control of relay switches, thus mitigating interference caused by mechanical switches. By furnishing such a dataset, we aim not only to enhance the recognition accuracy of existing NILM algorithms but also to facilitate the application of NILM algorithms in regions sharing similar electricity consumption characteristics to those of China.

Learning-Based Non-Intrusive Electric Load Monitoring for Smart Energy Management.

State-of-the-art smart cities have been calling for economic but efficient energy management over a large-scale network, especially for the electric power system. It is a critical issue to monitor, analyze, and control electric loads of all users in the system. In this study, a non-intrusive load monitoring method was designed for smart power management using computer vision techniques popular in artificial intelligence. First of all, one-dimensional current signals are mapped onto two-dimensional color feature images using signal transforms (including the wavelet transform and discrete Fourier transform) and Gramian Angular Field (GAF) methods. Second, a deep neural network with multi-scale feature extraction and attention mechanism is proposed to recognize all electrical loads from the color feature images. Third, a cloud-based approach was designed for the non-intrusive monitoring of all users, thereby saving energy costs during power system control. Experimental results on both public and private datasets demonstrate that the method achieves superior performances compared to its peers, and thus supports efficient energy management over a large-scale Internet of Things network.

Resilient data-driven non-intrusive load monitoring for efficient energy management using machine learning techniques

The integration of smart homes into smart grids presents numerous challenges, particularly in managing energy consumption efficiently. Non-intrusive load management (NILM) has emerged as a viable solution for optimizing energy usage. However, as smart grids incorporate more distributed energy resources, the complexity of demand-side management and energy optimization escalates. Various techniques have been proposed to address these challenges, but the evolving grid necessitates intelligent optimization strategies. This article explores the potential of data-driven NILM (DNILM) by leveraging multiple machine learning algorithms and neural network architectures for appliance state monitoring and predicting future energy consumption. It underscores the significance of intelligent optimization techniques in enhancing prediction accuracy. The article compares several data-driven mechanisms, including decision trees, sequence-to-point models, denoising autoencoders, recurrent neural networks, long short-term memory, and gated recurrent unit models. Furthermore, the article categorizes different forms of NILM and discusses the impact of calibration and load division. A detailed comparative analysis is conducted using evaluation metrics such as root-mean-square error, mean absolute error, and accuracy for each method. The proposed DNILM approach is implemented using Python 3.10.5 on the REDD dataset, demonstrating its effectiveness in addressing the complexities of energy optimization in smart grid environments.

Enhancing NILM classification via robust principal component analysis dimension reduction

Non-intrusive load monitoring (NILM) techniques estimate the consumption of individual appliances in a household or facility, based on aggregated reading from a centralized meter. Usually, NILM techniques are shown to be improved when various power features and additional power quality parameters are included. However, adding power features leads to increased time complexity which is a disadvantage to real-time operation. Previous attempt to operate a principal component analysis (PCA) method to reduce the dimension of the problem managed to improve the run time but with considerably low accuracy. To this end, we utilize a robust PCA approach, to mitigate the influence of outliers in the data as a measure for improved performance. The proposed procedure achieves extraordinary results with accuracy over 96% for 600 hours long record of power quality measurements of the consumption of seven appliances from the standard AMPds dataset.

Unknown appliances detection for non-intrusive load monitoring based on vision transformer with an additional detection head

Non-intrusive load monitoring (NILM) offers precise insights into equipment-level energy consumption by analyzing current and voltage data from residential smart meters, thus emerging as a potential strategy for demand-side management in power systems. However, a prevalent limitation in current NILM techniques is the presupposition of a known inventory of household appliances, an assumption that often becomes impractical due to the regular introduction of new appliances by consumers. To address this challenge, our approach integrates a vision transformer network with an additional detection head (ViTD), utilizing V-I trajectory images. Initially, the ViT model is trained to classify known appliances. Subsequently, an additional detection head is incorporated to manipulate the embedded features, encouraging the formation of distinct, compact class centers for the known appliance categories. During testing, samples are identified as either known or unknown appliances based on their proximity to these class centers. We utilize two public datasets, PLAID and WHITED, to demonstrate the effectiveness and superiority of our proposed method.

A Multi-Task Deep Learning Approach for Non-Intrusive Load Monitoring of Multiple Appliances

A Multi-Task Deep Learning Approach for Non-Intrusive Load Monitoring of Multiple Appliances

Evaluation of Deep Learning-Based Non-Intrusive Thermal Load Monitoring

Non-Intrusive Load Monitoring (NILM), which provides sufficient load for the energy consumption of an entire building, has become crucial in improving the operation of energy systems. Although NILM can decompose overall energy consumption into individual electrical sub-loads, it struggles to estimate thermal-driven sub-loads such as occupants. Previous studies proposed Non-Intrusive Thermal Load Monitoring (NITLM), which disaggregates the overall thermal load into sub-loads; however, these studies evaluated only a single building. The results change for other buildings due to individual building factors, such as floor area, location, and occupancy patterns; thus, it is necessary to analyze how these factors affect the accuracy of disaggregation for accurate monitoring. In this paper, we conduct a fundamental evaluation of NITLM in various realistic office buildings to accurately disaggregate the overall thermal load into sub-loads, focusing on occupant thermal load. Through experiments, we introduce NITLM with deep learning models and evaluate these models using thermal load datasets. These thermal load datasets are generated by a building energy simulation, and its inputs for the simulation were derived from realistic data like HVAC on/off data. Such fundamental evaluation has not been done before, but insights obtained from the comparison of learning models are necessary and useful for improving learning models. Our experimental results shed light on the deep learning-based NITLM models for building-level efficient energy management systems.

Study on real-time estimation of carbon emissions based on the non-intrusive power load monitoring

Monitoring carbon emissions, as well as evaluating and controlling the level of carbon emissions, are important prerequisites for combating climate change and promoting sustainable development. In this paper, real-time estimation of carbon emissions based on non-intrusive electricity load monitoring is investigated, existing machine learning approaches and related issues of load identification and marginal carbon emission factors are elucidated, and existing carbon emission monitoring approaches are analyzed and summarized. The methods of real-time carbon emission measurement are summarized, and their development is anticipated. Most of the existing methods for measuring and analyzing carbon emissions in power systems are unable to meet the development needs of real-time carbon emission calculation and have the disadvantages of higher installation and maintenance costs and poorer economy of monitoring equipment. Therefore, in this paper, a real-time carbon emission monitoring scheme based on deep learning is used to estimate carbon emissions by dividing them into direct carbon emissions and indirect carbon emissions.

MMNet-NILM: Multi-Target MobileNets for non-intrusive load monitoring

Energy is a critical resource for daily activities and lifestyles with direct impacts on the economy, health and environment. Therefore, monitoring its efficient use is essential to reduce energy waste and lessen related concerns such as global warming and climate change. One of the prominent and evolving solutions is Non-Intrusive Load Monitoring (NILM) smart meters, which enables consumers to track their per-appliance energy consumption more effectively. Some recent approaches have proposed deep learning as a powerful tool for energy disaggregation. However, it is difficult to employ these models in resource-constrained end devices for effective energy monitoring. In this paper, we explore and evaluate a lightweight improved model for multi-target non-intrusive load monitoring based on MobileNet architectures. With extensive experiments using the ENERTALK dataset, the results show that MobileNetV3-large is the most appealing for energy disaggregation as it requires about 55% less storage for trained model and about 6% less training time than MobileNetV2 with almost the same performance. On average, version 3 large has a 17.63% reduction in SAE and requires 54.21% and 8.93% less space and less training time than version 2, respectively. Moreover, the average performance is boosted using an ensemble multi-target MobileNet model across all houses, leading to significant reduction of MAE, SAE, and RMSE errors of about 6%, 48%, and 4%, respectively. In comparison to other work, the proposed MMNet-NILM shows superior performance for the majority of appliances in terms of all considered evaluation metrics.

Beyond privacy and security: Exploring ethical issues of smart metering and non-intrusive load monitoring

Artificial intelligence is believed to facilitate cost-effective and clean energy by optimizing consumption, reducing emissions, and enhancing grid reliability. Approaches such as non-intrusive load monitoring (NILM) offer energy efficiency insights but raise ethical concerns. In this paper, we identify most prominent ethical and societal issues by surveying relevant literature on smart metering and NILM. We combine these findings with empirical insights gained from qualitative workshops conducted with an electricity supplier piloting the use of AI for power load disaggregation. Utilizing the requirements for trustworthy AI, we show that while issues related to privacy and security are the most widely discussed, there are many other equally important ethical and societal issues that need to be addressed, such as algorithmic bias, uneven access to infrastructure, or loss of human control and autonomy. In total, we identify 19 such overarching themes and explore how they align with practitioners' perspectives and how they embody the seven core requirements for trustworthy AI systems defined by the Ethics Guidelines for Trustworthy AI.

How Can Non-Intrusive Load Monitoring Contribute to the Assessment of the Smart Readiness Indicator?

The Smart Readiness Indicator (SRI) is a framework introduced by the EU in 2018 to assess smart buildings in various aspects. However, the SRI has been criticized for several limitations, including its ambiguous service definitions. This paper proposes the application of Non-Intrusive-Load Monitoring (NILM) technology to enhance SRI evaluation on the example of SRI service E-12. NILM can be used to disaggregate energy consumption data to end use levels and allows for granular non-intrusive energy consumption measurement. The study involves a rigorous methodology using open sensor data and NILM algorithms to evaluate device-specific energy consumption We evaluate the IDEAL dataset and three different frequencies (5s, 15min, 1h), three different algorithms (CO, RNN, Seq2Point) and one data imputation strategies (forward filling). The results show that with a higher frequency, the performance metrics (F-score, normalized absolute error) increase. Regarding further considerations, we identify a trade-off between resource and energy efficiency, as well as privacy considerations with increasing measurement frequency. To achieve its aims for awareness, the SRI needs to consider interoperability and appropriate aggregations (frequency and spatial).

Non-Intrusive Load Identification Based on Retrainable Siamese Network.

Non-intrusive load monitoring (NILM) can identify each electrical load and its operating state in a household by using the voltage and current data measured at a single point on the bus, thereby behaving as a key technology for smart grid construction and effective energy consumption. The existing NILM methods mainly focus on the identification of pre-trained loads, which can achieve high identification accuracy and satisfying outcomes. However, unknown load identification is rarely involved among those methods and the scalability of NILM is still a crucial problem at the current stage. In light of this, we have proposed a non-intrusive load identification method based on a Siamese network, which can be retrained after the detection of an unknown load to increase the identification accuracy for unknown loads. The proposed Siamese network comprises a fixed convolutional neural network (CNN) and two retrainable back propagation (BP) networks. When an unknown load is detected, the low-dimensional features of its voltage-current (V-I) trajectory are extracted by using the fixed CNN model, and the BP networks are retrained online. The finetuning of BP network parameters through retraining can improve the representation ability of the network model; thus, a high accuracy of unknown load identification can be achieved by updating the Siamese network in real time. The public WHITED and PLAID datasets are used for the validation of the proposed method. Finally, the practicality and scalability of the method are demonstrated using a real-house environment test to prove the ability of online retraining on an embedded Linux system with STM32MP1 as the core.

Perfednilm: a practical personalized federated learning-based non-intrusive load monitoring

Non-Intrusive Load Monitoring (NILM) is a valuable technique for breaking down overall power consumption into the energy usage of individual appliances. Understanding power usage patterns through NILM plays an important role in reducing energy costs and achieving carbon reduction goals. However, privacy concerns often deter consumers from sharing their electricity consumption data. To address these privacy concerns, Federated Learning (FL) has been introduced in NILM, which enables the training of NILM models while keeping power consumers’ data locally. However, FL’s reliance on a single global model leads to poor performance on clients with unique power consumption patterns. In response to this challenge, we present a Personalized Federated Learning NILM algorithm (PerFedNILM), a practical personalized FL approach for NILM. PerFedNILM limits the local update bias across clients and trains personalized models for individual clients to improve load-monitoring performance. In addition, it mitigates the negative impact of client dropout, which is a common issue in practice. Our experiments on using real-world energy data demonstrate that PerFedNILM outperforms previous FL-based NILM methods, especially in client dropout scenarios.

Enhancing non-intrusive load monitoring with weather and calendar feature integration in DAE

The construction of modern power system is key to achieving dual carbon goals, where non-intrusive load monitoring (NILM) plays a vital role in enhancing energy utilization efficiency and energy management. For example, to enable prosumers to better understand the extent of their flexible loads for demand response and peer-to-peer trading, it is essential to be aware of the types and states of loads using the method of NILM. To improve the predictive accuracy and implementation effectiveness of NILM technology, this paper proposes a novel NILM method integrating meteorological and calendar features. It delves deeply into the close connection between external factors such as temperature, precipitation, wind speed, and holidays, and the energy consumption of electrical appliances, constructing additional associative mappings in the training of the Denoising Autoencoder (DAE) model. Test results on the UK-DALE public dataset show that the NILM method proposed in this paper has significant advantages over traditional NILM methods that consider only single-dimensional electrical data features, in terms of load pattern recognition and accuracy in load energy consumption monitoring. This confirms the potential of multi-dimensional feature fusion technology in the application of NILM.