Article
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

We asset about the analysis of electrical appliance consumption signatures for the identification task. We apply Hidden Markov Models to appliance signatures for the identification of their category and of the most probable sequence of states. The electrical signatures are measured at low frequency (10-1 Hz) and are sourced from a specific database. We follow two predefined protocols for providing comparable results. Recovering information on the actual appliance state permits to potentially adopt energy saving measures, as switching off stand-by appliances or, generally speaking, changing their state. Moreover, in most of the cases appliance states are related to user activities: the user interaction usually involves a transition of the appliance state. Information about the state transition could be useful in Smart Home / Building Systems to reduce energy consumption and increase human comfort.We report the results of the classification tasks in terms of confusion matrices and accuracy rates. Finally, we present our application for a real-time data visualization and the recognition of the appliance category with its actual state.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... Initially, NILM systems were predominantly developed using signal processing, optimization, and traditional machine learning (ML) methods. Hidden Markov models (HMM) stand out as one of the commonly utilized techniques for NILM [11], with several studies This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4795206 ...
... Eqs. (10), (11), and (12) outline the criteria used to calculate the overlapping energy ( ), missed energy ( ), and extra energy ( ), respectively while computing JEOI, which slightly differs from the original implementation proposed in [16]. JEOI is expected to serve as a robust metric for evaluating the performance of smart meter disaggregation, regardless of the appliance load patterns. ...
... For number of files (f) do 10 Extract the timestamp from the file name and convert it to standard format for matching 11 Match appliance and smart meter timestamp 12 ...
Preprint
Full-text available
Non-intrusive load monitoring (NILM), achieved through the disaggregation of smart meter load consumption data, offers numerous applications in managing the demand side of smart households. These applications include optimizing device usage, non-intrusive monitoring and control of devices, cost savings, and inexpensively detecting line and device faults or transients. While existing state-based NILM techniques are developed to estimate the continuous load profiles of individual devices, current state-of-the-art load disaggregation methods encounter challenges in simultaneously generating real-time load states of multiple devices. Furthermore, the metrics currently used to evaluate state-based NILM methods often lack precision in quantifying their load disaggregation capabilities across diverse appliances. In this study, we introduce SM-DAGG, a deep-learning-based framework designed to reconstruct the load consumption patterns of individual devices from aggregated smart meter data. The complete load profiles are generated based on min-max normalization factors saved per household during data preprocessing. We present DAGG-Net, a novel 1D-reconstruction network designed for efficient disaggregation. Additionally, we propose the Jaccard of Energies of Interest (JEOI) metric to effectively evaluate SM-DAGG's load disaggregation performance. The ECO dataset, comprising aggregated smart meter and corresponding appliance load consumption data from multiple households, is utilized for training and evaluation. Through extensive experimentation on the ECO dataset, we significantly surpass state-of-the-art results for most devices. SM-DAGG demonstrates the potential for advancing NILM applications by facilitating robust and precise real-time non-intrusive monitoring of household devices.
... In this framework, several automated appliance classification techniques have been reported [9,10,11]. In [9], authors have used zero-normalization consumption time series with delta and delta-delta coefficients as attributes of intended appliances. ...
... The classification is performed by using the Gaussian mixture model (GMM) classifier. In [10], authors have used entropy of consumption time series with delta and delta-delta coefficients as attributes of intended appliances. The classification is performed with hidden Markov models (HMM). ...
... Conventionally, the smart meters data is collected in a time-invariant fixed-rate manner [9,10,11], resulting in the collection, storage, processing, and classification of a significant amount of unnecessary data [12]. In the suggested metering framework: AQ1 • Event-driven sensing is incorporated to attain significant real-time data compression. ...
Chapter
The utilization of smart meters is increasing with the technical developments. A proper collection and analysis of metering data is important to better serve various smart grid stakeholders. The classical sensing mechanism is known to be time-variant, which results in vast amounts of excessive data to be collected, distributed, processed and stored. It causes an unnecessary increase of processing activity and consumption. In this context, this research uses the event-driven sensing and processing methods for an effective processing and collection of smart meter data. New adaptive rate techniques for data acquisition, processing, segmentation, and extraction of features are suggested. The extracted features, related to consumption patterns of appliances, are then used to classify these appliances. The classification is carried out by using robust algorithms namely k-Nearest Neighbor (KNN), Artificial Neural Network (ANN) and Naïve Bias. The findings showed more than a 3-fold benefit in compression and computational efficiency while attaining 94.4% highest classification accuracy for a 6-class dataset.
... To accomplish this task, we add a module dedicated to the detection of the interactions to our existing system architecture based on Hidden Markov Models (HMM). The existing system is able to recognise the appliance class and the sequence of hidden states [7]. In Section 2 we provide details about related work dealing with activity recognition when using NILM and ILM. ...
... In this Section we clarify the context of our work by presenting our system for the appliance and state identification based on HMM [7] and the ACS-F2 database used for its evaluation [15]. The system takes as input features coming from a Smart Plug and communicates with a PC that performs the feature extraction and the machine learning task. ...
... We performed a feature selection based on the computation of the entropy and the information available from the theory of electricity. For the intersession protocol we were able to reduce the feature space by a half, while for the unseen appliance protocol we did not apply any modification [7]. As machine learning algorithm we used HMM. ...
... We finally perform feature selection as the six raw features contains redundant information [15]. For the Intersession protocol we reduced the feature space by a half, removing the RMS voltage, frequency of the network and the phase of voltage relative to current and their dynamic coefficients; for the Unseen Appliance protocol no meaningful conclusions could be established because of the large variability of the features that is intrinsic in this protocol so we decided not to perform feature selection for this protocol. ...
... We obtained respectively 83.1% and 89.8% for the Intersession protocol and respectively 53.6% and 73.3% for the Unseen Appliance protocol. In our last work we applied HMMs on the ACS-F2 database using a quite different approach [15]. We labeled the data by recovering the state sequences and we associated the states of the models with real (physical) states of the appliances. ...
... A total of 60 accuracy rates are computed. We apply the feature selection for both the protocols as explained in [15]. Figures 4A and 4B show the trend of the accuracy rates for the Dynamic Intersession and the Dynamic Unseen appliance protocol. ...
Article
The automatic identification of appliances through the analysis of their electricity consumption has several purposes in Smart Buildings including better understanding of the energy consumption, appliance maintenance and indirect observation of human activities. Electric signatures are typically acquired with IoT smart plugs integrated or added to wall sockets. We observe an increasing number of research teams working on this topic under the umbrella Intrusive Load Monitoring. This term is used as opposition to Non-Intrusive Load Monitoring that refers to the use of global smart meters. We first present the latest evolutions of the ACS-F database, a collections of signatures that we made available for the scientific community. The database contains different brands and/or models of appliances with up to 450 signatures. Two evaluation protocols are provided with the database to benchmark systems able to recognise appliances from their electric signature. We present in this paper two additional evaluation protocols intended to measure the impact of the analysis window length. Finally, we present our current best results using machine learning approaches on the 4 evaluation protocols.
... stand-by). In this work, we demonstrate the recognition of the appliance category and state using an implementation inspired from [4]. For training the models we use the ACS-F2 database [3], containing 450 electrical signatures recorded from 225 appliances of different brands and / or models. ...
... The signatures in the ACS-F2 database are represented in a six dimensional space as described above. Given the small quantity of memory available on the OpenPicus, we decide to reduce as much as possible the feature space, removing the non relevant frequency and voltage features, as well as the phase angle which is redundant with the active and reactive power information [4]. We selected Hidden Markov Models (HMMs) to represent the state-based nature of the signals. ...
... However, in this approach some of the appliances were skipped. Another study that uses Hidden Markov Models, which is another machine learning algorithm [11], has achieved relatively better results. A study using moving average for data preparation [12] has reached to 99 % by random forest classification and multilayer perceptron. ...
Article
Full-text available
Nowadays, everything is getting smarter such as mobile phones, cars, watches and home appliances. Our powerlines are also getting smarter. There are many smart grid and smart home applications. Designing of recognition devices to identify appliances for these smart networks is a new task to do it. There are many different approaches on recognition and identification these power consumer devices and appliance. This study aims to develop an effective method that does not require any additional hardware. This method has been developed by using powerline parameters such as current, phase angle, voltage, active and reactive power. These data have been classified and normalized by using a validation method and grey relational analysis to train an artificial neural network. This neural network was trained by using power parameters of many different common appliances like heater, coffee machine, television, radio, lamp, computer, fan, refrigerator etc. This identification algorithm can be used within a low-cost embedded system for collecting appliance information over a powerline to provide info for smart homes and smart grids.
... The former approaches use Bayesian networks [244,245] and hierarchical clustering models [246]. The latter use two different machine-learning-based algorithms: HMMs [239,[247][248][249] and NNs [250][251][252]. To model occupants' indoor behavior and activities in interaction with appliances, diverse algorithms were employed in the studies, such as pedestrian dead reckoning [253], Bayesian network mode and linear regression [254], k-means and Gaussian mixture [69], random forest [255], and SVMs [256]. ...
Article
In the last four decades several methods have been used to model occupants’ presence and actions (OPA) in buildings according to different purposes, available computational power, and technical solutions. This study reviews approaches, methods and key findings related to OPA modeling in buildings. An extensive database of related research documents is systematically constructed, and, using bibliometric analysis techniques, the scientific production and landscape are described. The initial literature screening identified more than 750 studies, out of which 278 publications were selected. They provide an overarching view of the development of OPA modeling methods. The research field has evolved from longitudinal collaborative efforts since the late 1970s and, so far, covers diverse building typologies mostly concentrated in a few climate zones. The modeling approaches in the selected literature are grouped into three categories (rule-based models, stochastic OPA modeling, and data-driven methods) for modeling occupancy-related target functions and a set of occupants’ actions (window, solar shading, electric lighting, thermostat adjustment, clothing adjustment and appliance use). The explanatory modeling is conventionally based on the model-based paradigm where occupant behavior is assumed to be stochastic, while the data-driven paradigm has found wide applications for the predictive modeling of OPA, applicable to control systems. The lack of established standard evaluation protocols was identified as a scientifically important yet rarely addressed research question. In addition, machine learning and deep learning are emerging in recent years as promising methods to address OPA modeling in real-world applications.
... A post-processing stage is refining the power estimates form the regression model by mapping them to apriori known device states using a Look-Up- Table (LUT), i.e., if the distance of the regression output to any state in the device model is larger than 25W the regression output is mapped to the closest device state. In order to define the number of states per device the K-Means algorithm was used for initialisation followed by Expectation-Maximization (EM) clustering to calculate the power consumption for each state of each device and form the LUT for the post-processing stage [48]. ...
Article
Full-text available
In this paper we evaluate several well-known and widely used machine learning algorithms for regression in the energy disaggregation task. Specifically, the Non-Intrusive Load Monitoring approach was considered and the K-Nearest-Neighbours, Support Vector Machines, Deep Neural Networks and Random Forest algorithms were evaluated across five datasets using seven different sets of statistical and electrical features. The experimental results demonstrated the importance of selecting both appropriate features and regression algorithms. Analysis on device level showed that linear devices can be disaggregated using statistical features, while for non-linear devices the use of electrical features significantly improves the disaggregation accuracy, as non-linear appliances have non-sinusoidal current draw and thus cannot be well parametrized only by their active power consumption. The best performance in terms of energy disaggregation accuracy was achieved by the Random Forest regression algorithm.
... The electric signals then observed are corresponding to one or few appliances, making the detection and identification task easier. Different machine learning modelling strategies are used to capture the specificities of electric signature with ILM, such as the use of Gaussian Mixture Models or Hidden Markov Models that reported good identification accuracy up to 95% [8,9]. ...
Article
Electricity load monitoring in residential buildings has become an important task allowing for energy consumption understanding, indirect human activity recognition and occupancy modelling. In this context, Non Intrusive Load Monitoring (NILM) is an approach based on the analysis of the global electricity consumption signal of the habitation. Current NILM solutions are reaching good precision for the identification of electrical devices but at the cost of difficult setups with expensive equipments typically working at high frequency. In this work we propose to use a low-cost and easy to install low frequency sensor for which we improve the performances with an active machine learning strategy. At setup, the system is able to identify some appliances with typical signatures such as a fridge. During usage, the system detects unknown signatures and provides a user-friendly procedure to include new appliances and to improve the identification precision over time.
... Six features are recorded by the smart plug: active power, reactive power, phase angle, RMS current, RMS voltage and frequency of the electrical network. We excluded the phase angle, RMS Voltage and frequency of the network, demonstrated to be not relevant or redundant [14]. We use Hidden Markov Models (HMMs) as machine learn-ing algorithm for the appliance recognition. ...
Article
Full-text available
Internet-of-Things (IoT) devices, especially sensors are producing large quantities of data that can be used for gathering knowledge. In this field, machine learning technologies are increasingly used to build versatile data-driven models. In this paper, we present a novel architecture able to execute machine learning algorithms within the sensor network, presenting advantages in terms of privacy and data transfer efficiency. We first argument that some classes of machine learning algorithms are compatible with this approach, namely based on the use of generative models that allow a distribution of the computation on a set of nodes. We then detail our architecture proposal, leveraging on the use of Web-of-Things technologies to ease integration into networks. The convergence of machine learning generative models and Web-of-Things paradigms leads us to the concept of virtual things exposing higher level knowledge by exploiting sensor data in the network. Finally, we demonstrate with a real scenario the feasibility and performances of our proposal.
Chapter
The thermal SPA sector is currently experiencing a stable growth trend, which according to the World Tourism Organization (WTO) is expected to continue over the upcoming years. In Portugal, the sector has a very significant profile, with the existence of almost a hundred SPAs and thermal SPAs that generate a business volume (direct and indirect) of over 30 M€ per year. Although the beginning of the process of digital transformation of the sector is already visible, there is no holistic view of the sector which means that the currently existing information systems (IS) do not present a useful response to the needs faced by the sector. Therefore, an architecture proposal was conceived and described for an IS that provides a useful, efficient, and agile response to the needs of the entire thermalism sector and its stakeholders.KeywordsThermalismThermal SPAInformation systems architectureThermalism observatoryDigital transformation
Chapter
This research studies the effects of development architectures on the quality of APIs by conducting a computational laboratory experiment comparing the performance efficiency of a GraphQL API, a REST API, and a GraphQL API that wraps a REST API. Open data from the Electronic Chamber of Commerce of Ecuador, part of a national e-commerce research project, was used. To characterize quality, we used ISO/IEC 25,023 metrics in different use cases of e-commerce data consumption and insertion. Finally, we statistically analyzed the experiment results, which indicate a difference in quality between the REST API, the GraphQL API, and the GraphQL API (wrapper); this being the case, the GraphQL API performs more efficiently.KeywordsREST APIGraphQL APIWrapperComputational laboratory experimentISO/IEC 25,023e-commerce
Chapter
Motivation—Industrial internet of things (IIoT) refers to interconnected sensors, instruments, and other devices networked together with computers’ industrial applications, including manufacturing and energy management. This connectivity allows for data collection, exchange, and analysis, potentially facilitating improvements in productivity and efficiency, as well as other economic benefits. IIoT provides more automation by using cloud computing to refine and optimize process controls. Problem—Detection and classification of events inside industrial settings for process monitoring often rely on input channels of various types (e.g. energy consumption, occupation data or noise) that are typically imprecise. However, the proper identification of events is fundamental for automatic monitoring processes in the industrial setting, allowing simulation and forecast for decision support. Methods—We have built a framework where process events are being collected in a classic cars restoration shop to detect the usage of equipment such as paint booths, sanders and polishers, using energy monitoring, temperature, humidity and vibration IoT sensors connected to a Wifi network. For that purpose, BLE beacons are used to locate cars being repaired within the shop floor plan. The InfluxDB is used for monitoring sensor data, and a server is used to perform operations on it, as well as run machine learning algorithms. Results—By combining location data and equipment being used, we are able to infer, using ML algorithms, some steps of the restoration process each classic car is going through. This detection contributes to the ability of car owners to remotely follow the restore process, thus reducing the carbon footprint and making the whole process more transparent.KeywordsProcess activity recognitionIIoTIoT sensorsIntrusive load monitoringMachine learningIndoor locationClassic cars restorationCharter of Turin
Book
This book evaluates the role of innovative machine learning and deep learning methods in dealing with power system issues, concentrating on recent developments and advances that improve planning, operation, and control of power systems. Cutting-edge case studies from around the world consider prediction, classification, clustering, and fault/event detection in power systems, providing effective and promising solutions for many novel challenges faced by power system operators. Written by leading experts, the book will be an ideal resource for researchers and engineers working in the electrical power engineering and power system planning communities, as well as students in advanced graduate-level courses. • Offers innovative machine learning and deep learning methods for dealing with power system issues; • Provides promising solution methodologies; • Covers theoretical background and experimental analysis.
Chapter
Machine learning (ML) methods and their applications are among the most innovative and attractive engineering topics. The advent of artificial intelligence has given birth to various tools widely used in science and engineering. ML is utilized in solving various problems in the power system engineering community, such as power system planning and operation. In this chapter, the authors will investigate different machine learning methods, and we will discuss their applications in solving power system planning problems, including load forecasting. The authors will discuss different ML methods used in the power engineering field, and other ML applications in planning problems such as optimization problems will be studied. This chapter’s main objective is to serve as an introduction to ML for power system planning and the basic concepts of the ML methods commonly used in this field.
Article
The installation of smart meters is fast growing to effectively support various smart grid stack holders. Collection and processing of fine-grained metering data is important for proper analysis and decision support. The traditional smart meters are based on standardized and time-invariant tactics to acquire and process the data. This results in the collection, storage, and processing of a huge amount of unneeded data. The focus of this paper is to enhance the contemporary smart meters data acquisition and processing chains. The objective is to attain real-time compression and computational effectiveness to enhance the system performance in terms of data analysis, storage and transmission and to diminish its consumption overhead. In this framework, the signal-piloted event-driven sampling and processing tactics are exploited. The novel adaptive rate techniques are used for data segmentation and extraction of features. Household appliances consumption patterns related features are being classified subsequently. It is realized by employing the mature k-Nearest Neighbor and the Artificial Neural Network classifiers. Results demonstrate a 3.8-fold compression gain and computational effectiveness of the designed solution over traditional counterpart while securing the highest classification accuracy of 94.4% for the 6-class appliances dataset.
Conference Paper
Recognizing the appliance according to the flowed electric current through it is quite a meaningful work which can help the electric management system to make effective policy of energy conservation. We designed an algorithm based on an improved k-nearest neighbor which can classify the unlabelled appliances’ running power data into its most similar data clusters. In other words, this algorithm is able to recognize the appliance only according to its running power data series. The classification is based upon the multifarious features extracted from the time series data sensed from the running appliance with the power metering sensors. Appliance recognition is performed with a mean accuracy over 90% in five-class classification problem.
Article
Demand-side management technology is a key element of the proposed smart grid, which will help utilities make more efficient use of their generation assets by reducing consumers' energy demand during peak load periods. However, although some modern appliances can respond to price signals from the utility companies, there is a vast stock of older appliances that cannot. For such appliances, utilities must infer what appliances are operating in a home, given only the power signals on the main feeder to the home (i.e., the home's power consumption must be disaggregated into individual appliances). We report on an in-depth investigation of multi-label classification algorithms for disaggregating appliances in a power signal. A systematic review of this research topic shows that this class of algorithms has received little attention in the literature, even though it is arguably a more natural fit to the disaggregation problem than the traditional single-label classifiers used to date. We examine a multi-label meta-classification framework (RAkEL), and a bespoke multi-label classification algorithm (MLkNN), employing both time-domain and wavelet-domain feature sets. We test these classifiers on two real houses from the Reference Energy Disaggregation Dataset. We found that the multilabel algorithms are effective and competitive with published results on the datasets.
Conference Paper
Full-text available
The continuous growth of energy needs and the fact that unpredictable energy demand is mostly served by unsustainable (i.e. fossil-fuel) power generators have given rise to the development of Demand Response (DR) mechanisms for flattening energy demand. Building effective DR mechanisms and user awareness on power consumption can significantly benefit from fine-grained monitoring of user consumption at the appliance level. However, installing and maintaining such a monitoring infrastructure in residential settings can be quite expensive. In this paper, we study the problem of fine-grained appliance power-consumption monitoring based on one house-level meter and few plug-level meters. We explore the trade-off between monitoring accuracy and cost, and exhaustively find the minimum subset of plug-level meters that maximize accuracy. As exhaustive search is time- and resource-consuming, we define a heuristic approach that finds the optimal set of plug-level meters without utilizing any other sets of plug-level meters. Based on experiments with real data, we found that few plug-level meters - when appropriately placed - can very accurately disaggregate the total real power consumption of a residential setting and verified the effectiveness of our heuristic approach.
Conference Paper
Full-text available
To optimize the energy utilization, intelligent energy management solutions require appliance-specific consumption statistics. One can obtain such information by deploying smart power outlets on every device of interest, however it incurs extra hardware cost and installation complexity. Alternatively, a single sensor can be used to measure total electricity consumption and thereafter disaggregation algorithms can be applied to obtain appliance specific usage information. In such a case, it is quite challenging to discern low-power appliances in the presence of high-power loads. To improve the recognition of lowpower appliance states, we propose a solution that makes use of circuit-level power measurements. We examine the use of a specialized variant of Hidden Markov Model (HMM) known as Factorial HMM (FHMM) to recognize appliance specific load patterns from the aggregated power measurements. Further, we demonstrate that feature concatenation can improve the disaggregation performance of the model allowing it to identify device states with an accuracy of 90% for binary and 80% for multi-state appliances. Through experimental evaluations, we show that our solution performs better than the traditional event based approach. In addition, we develop a prototype system that allows real-time monitoring of appliance states
Article
Full-text available
In automated energy management systems, to make instantaneous decisions based on the appliance status information, continuous data access is a key requirement. With the advances in sensor and communication technologies, it is now possible to remotely monitor the power consumption data. However, before an appliance is actively monitored, it must be identified using the obtained power consumption data. Appropriate methods are required to analyse power consumption patterns for proper appliance recognition. The focus of this work is to provide the model structure for storing and distinguishing the recurring footprints of the household appliances. Hidden Markov model based method is proposed to recognize the individual appliances from combined load. It is found that the proposed method can efficiently differentiate the power consumption patterns of appliances from their combined profiles.
Article
Full-text available
This paper presents the current sensor based non-intrusive appliance recognition method for intelligent outlet. Our system has two main functions; one is the remote control function of power supply through the Internet. The other is monitoring function observe the state of appliance. In this pa-per, the monitor function is especially focused. To recognize the state of appliance, we extract nine features based on mea-sured current signal. In the experiment, we gathered a number of signals with various appliances, and found that three fea-tures I peak , I avg , and I rms yield valid recognition results of 81.3%, 84.0%, and 87.4% for classifying the state of appli-ance into three categories.
Conference Paper
Full-text available
This paper presents a new nonintrusive appliance load monitoring technique based on the integer programming. Nonintrusive appliance load monitoring is a problem to identify the operating conditions of the electric appliances in a house only by observing the overall load current and voltage. Since the overall load current is expressed as a superposition of each current of the operating appliance, the monitoring problem can be formulated as an integer quadratic programming problem by expressing the operating conditions as integer variables. This problem is solvable with sufficiently small computational burden thanks to the recent development of the commercial software. The proposed method does not require the relearning even when a new appliance is installed in the house. Furthermore, the proposed formulation is applicable to cases that some appliance has multiple modes, and cases that some same-type appliances operate simultaneously. Usefulness of the proposed technique is verified through some experimental results.
Conference Paper
Full-text available
Microgrids are well-suited for electrification of remote off-grid areas. This paper sketches the concept of a plug-and-play microgrid with a minimum of configuration effort needed for setup. When the load of such an off-grid microgrid grows over the generation capacity and energy storage is not sufficient, demand has to be reduced to prevent a blackout. In order to decide which loads are inessential and can be shedded, automated load recognition on the basis of measured power consumption profiles is needed. Two promising approaches from the area of speech recognition, Dynamic Time Warping and Hidden Markov Models, are compared for this application. It is found that a key feature to achieve good recognition efficiency is a careful selection of the features extracted from the measured power data.
Conference Paper
Full-text available
Feature selection, as a preprocessing step to machine learning, has been eective in reduc- ing dimensionality, removing irrelevant data, increasing learning accuracy, and improving comprehensibility. However, the recent in- crease of dimensionality of data poses a se- vere challenge to many existing feature se- lection methods with respect to eciency and eectiveness. In this work, we intro- duce a novel concept, predominant correla- tion, and propose a fast filter method which can identify relevant features as well as re- dundancy among relevant features without pairwise correlation analysis. The eciency and eectiveness of our method is demon- strated through extensive comparisons with other methods using real-world data of high dimensionality.
Conference Paper
Full-text available
Fear of increasing prices and concern about climate change are motivating residential power conservation efforts. We investigate the effectiveness of several unsupervised disaggregation methods on low frequency power measurements collected in real homes. Specifically, we consider variants of the factorial hidden Markov model. Our results indicate that a conditional factorial hidden semi-Markov model, which integrates additional features related to when and how appliances are used in the home and more accurately represents the power use of individual appliances, outperforms the other unsupervised disaggregation methods. Our results show that unsupervised techniques can provide perappliance power usage information in a non-invasive manner, which is ideal for enabling power conservation efforts. 1
Article
Full-text available
Nous présentons une méthodologie basée sur les chaînes de Markov cachées pour la modélisation et l'analyse statistique de courbes de consommation électrique. Suite à une analyse de variance qui permet d'estimer l'effet sur la log-consommation de facteurs contrôlés (mois, jour, heure, type de contrat et puissance maximale souscrite), nous modélisons les résidus par une chaîne de Markov cachée. Ensuite, les états cachés sont restaurés puis interprétés grâce à un tableau de contingence les mettant en relation avec la consommation de différents appareils électriques (ou usages) lorsque celle-ci est disponible. Nous montrons comment cette approche, par l'usage conjoint du tableau de contingence et des états cachés restaurés, permet une estimation de la consommation de chaque usage dans le cas où elle est inconnue. La prise en compte d'informations a priori sur la consommation rend cette estimation plus réaliste. Enfin, nous abordons le problème de la sélection de modèles. Les critères classiques conduisent à des modèles complexes qui rendent très difficile l'interprétation des états cachés par leur mise en correspondance avec les usages. Nous présentons et commentons des méthodes de sélection alternatives qui tiennent compte de l'objectif de mise en relation des états cachés et des classes de consommation.
Article
We present ACS-F2, a new electric consumption signature database acquired from domestic appliances. The scenario of use is appliance identification with emerging applications such as domestic electricity consumption understanding, load shedding management and indirect human activity moni-toring. The novelty of our work is to use low-end electricity consumption sensors typically located at the plug. Our approach consists in acquiring signatures at a low frequency, which contrast with high frequency transient analysis approaches that are costlier and have been well studied in former research works. Electrical consumption signatures comprise real power, reactive power, RMS current, RMS voltage, frequency and phase of voltage relative to current. A total of 225 appliances were recorded over two sessions of one hour. The database is balanced with 15 different brands/models spread into 15 categories. Two realistic appliance recognition protocols are proposed and the database is made freely available to the scientific community for the experiment reproducibility. We also report on recognition results following these protocols and using baseline recognition algorithms like k-NN and GMM.
Conference Paper
The aim of appliance identification methods is to get electrical energy consumption at appliance level based on aggregate measurements from a single energy meter. Disaggregation of measurements from a single meter allow to reduce the costs of the hardware part of the energy management systems. The article presents results of home appliances identification based on active power measurements. The Factorial Hidden Markov Model is applied to identify different appliances in the same time. Independent changes in active power of every appliance is described by each Markov chain. Having measurements of active power from single meter it is necessary to compute hidden variables defining states of appliance. The Additive Factorial Approximate MAP algorithm allows to designate states of each appliance. Moreover, an analysis of available solutions in terms of measurement frequency and appliance mathematical modeling is presented. We present the flowchart of the prototype appliance identification system with low measurement frequency energy meter. In the experimental part of the article, results of the selected home appliances identification are presented. Based on the results we conclude that probabilistic models of appliances allow to identify appliances working simultaneously.
Article
Energy and sustainability issues raise a large number of problems that can be tackled using approaches from data mining and machine learning, but traction of such problems has been slow due to the lack of publicly available data. In this paper we present the Reference Energy Disaggregation Data Set (REDD), a freely available data set containing de-tailed power usage information from several homes, which is aimed at furthering research on energy disaggregation (the task of determining the component appliance contributions from an aggregated electricity signal). We discuss past ap-proaches to disaggregation and how they have influenced our design choices in collecting data, we describe the hardware and software setups for the data collection, and we present initial benchmark disaggregation results using a well-known Factorial Hidden Markov Model (FHMM) technique.
Conference Paper
We report on the evaluation of signal processing and classification algorithms to automatically recognize electric appliances. The system is based on low-cost smart-plugs measuring periodically the electricity values and producing time series of measurements that are specific to the appliance consumptions. In a similar way as for biometric applications, such electric signatures can be used to identify the type of appliance in use. In this paper, we propose to use dynamic features based on time derivative and time second derivative features and we compare different classification algorithms including K-Nearest Neighbor and Gaussian Mixture Models. We use the recently recorded electric signature database ACS-Fl and its intersession protocol to evaluate our algorithm propositions. The best combination of features and classifiers shows 93.6% accuracy.
Article
Electrical load disaggregation for end-use recognition in the smart home has become an area of study of its own right. The most well-known examples are energy monitoring, health care applications, in-home activity modeling, and home automation. Real-time energy-use analysis for whole-home approaches needs to understand where and when the electrical loads are spent. Studies have shown that individual loads can be detected (and disaggregated) from sampling the power at one single point (e.g. the electric service entrance for the house) using a non-intrusive load monitoring (NILM) approach. In this paper, we focus on the feature extraction and pattern recognition tasks for non-intrusive residential electrical consumption traces. In particular, we develop an algorithm capable of determining the step-changes in signals that occur whenever a device is turned on or off, and which allows for the definition of a unique signature (ID) for each device. This algorithm makes use of features extracted from active and reactive powers and power factor. The classification task is carried out by Support Vector Machines and 5-Nearest Neighbors methods. The results illustrate the effectiveness of the proposed signature for distinguishing the different loads.
Article
A nonintrusive appliance load monitor that determines the energy consumption of individual appliances turning on and off in an electric load, based on detailed analysis of the current and voltage of the total load, as measured at the interface to the power source is described. The theory and current practice of nonintrusive appliance load monitoring are discussed, including goals, applications, load models, appliance signatures, algorithms, prototypes field-test results, current research directions, and the advantages and disadvantages of this approach relative to intrusive monitoring
ACS-F2 -A New Database of Appliance Consumption Analysis
  • A Ridi
  • C Gisler
  • J Hennebert
A. Ridi, C. Gisler, and J. Hennebert. ACS-F2 -A New Database of Appliance Consumption Analysis. In Proceedings of the International Conference on Soft Computing and Pattern Recognition (SocPar 2014), 2014.
Current Sensor based Non-intrusive Appliance Recognition for Intelligent Outlet
  • T Saitoh
  • Y Aota
  • T Osaki
  • R Konishi
  • K Sugahara
T. Saitoh, Y. Aota, T. Osaki, R. Konishi, and K. Sugahara. Current Sensor based Non-intrusive Appliance Recognition for Intelligent Outlet. In Proceedings of the 23rd International Technical Conference on Circuits/Systems, Computers and Communications (ITC-CSCC), pages 349-352, 2008.