`through a Single Electricity Sensor
`
`A.G. Ruzzelli, C. Nicolas†, A. Schoofs, and G.M.P. O’Hare
`CLARITY: Centre for Sensor Web Technologies
`School of Computer Science and Informatics, University College Dublin, Ireland
`†Institute of Higher Education in Computer Science and Communication,
`University of Rennes, France
`{ruzzelli, anthony.schoofs, gregory.ohare}@ucd.ie
`clement.nicolas@etudiant.univ-rennes1.fr
`
`Abstract—Sensing, monitoring and actuating systems are ex-
`pected to play a key role in reducing buildings overall energy
`consumption. Leveraging sensor systems to support energy effi-
`ciency in buildings poses novel research challenges in monitoring
`space usage, controlling devices, interfacing with smart energy
`meters and communicating with the energy grid. In the attempt
`of reducing electricity consumption in buildings,
`identifying
`individual sources of energy consumption is key to generate
`energy awareness and improve efficiency of available energy
`resources usage. Previous work studied several non-intrusive
`load monitoring techniques to classify appliances; however, the
`literature lacks of an comprehensive system that can be eas-
`ily installed in existing buildings to empower users profiling,
`benchmarking and recognizing loads in real-time. This has
`been a major reason holding back the practice adoption of
`load monitoring techniques. In this paper we present RECAP:
`RECognition of electrical Appliances and Profiling in real-time.
`RECAP uses a single wireless energy monitoring sensor easily
`clipped to the main electrical unit. The energy monitoring unit
`transmits energy data wirelessly to a local machine for data
`processing and storage. The RECAP system consists of three
`parts: (1) Guiding the user for profiling electrical appliances
`within premises and generating a database of unique appliance
`signatures;
`(2) Using those signatures to train an artificial
`neural network that is then employed to recognize appliance
`activities (3) Providing a Load descriptor to allow peer appliance
`benchmarking. RECAP addresses the need of an integrated and
`intuitive tool to empower building owners with energy awareness.
`Enabling real-time appliance recognition is a stepping-stone
`towards reducing energy consumption and allowing a number
`of major applications including load-shifting techniques, energy
`expenditure breakdown per appliance, detection of power hungry
`and faulty appliances, and recognition of occupant activity. This
`paper describes the system design and performance evaluation
`in domestic environment.
`
`I. INTRODUCTION
`Electricity represents 41% of the total energy used in
`American homes [1]. The delivered energy use per household
`declines at an average annual rate of 0.6 percent, mostly due
`to technological progress in power efficiency [2]. To further
`increase that trend, smart energy grids are being promoted to
`address optimal management and improved control of energy,
`by introducing intelligence into the electricity grid. The recent
`momentum for Smart Grid meters, visible in a number of
`government driven large-scale pilot deployments such as in
`
`Italy [11] and in the US [12], intends to accelerate their
`introduction into households.
`
`Fig. 1. Typical energy consumption in domestic premises
`
`Within this context, embedded sensor networks and actu-
`ating systems are expected to play a key role in monitoring
`and reducing building’s overall energy consumption. Recent
`standardization efforts have generated a push towards the
`integration of sensor systems in building automation systems
`and home environments. IEEE 802.15.4, ZigBee [16], and
`IETF 6LoWPAN/ROLL [5], [6] are enabling technologies that
`facilitates the connections of low-cost sensing and monitoring
`units and gather energy consumption information in real-time.
`Low-power wireless networking has enabled easy access to
`households meter readings, making them available to energy
`utilities for monitoring and control, and to building owners
`for direct feedback on their energy consumption e.g. Ted
`energy detective [14]. Fine-grained energy decomposition is
`nevertheless not available. Key is to process the energy data
`to finally provide meaningful information to empower building
`owners with hints for reducing their energy cost. To this end,
`providing a breakdown of the energy expenditure per appliance
`is of particular interest to identify energy hungry devices and
`provide other interesting services to the homeowner (e.g. home
`activity patterns).
`Figure 1 shows typical domestic energy consumption over
`a time period with some appliance activity annotations. The
`
`APPLE 1018
`
`1
`
`
`
`main aim of this work is to develop a low-cost system that
`can attribute names to appliances contributing to each of the
`energy spikes in real time with a single energy monitor. In this
`paper, we present Real-time Electrical Appliance Recognition
`(RECAP), a system that provides fine-grained recognition of
`appliances in real-time, based on one single Zigbee-based
`building energy monitor attached to the main electrical unit.
`RECAP system components are appliance signature profiling,
`real-time signature recognition, and intuitive user feedback. Up
`to now, several works studied non-intrusive load monitoring
`techniques to classify appliances. However, the market shows
`a lack of practical adoption of such techniques due to practical
`issues when deploying the systems into into existing buildings.
`In contrast, this paper focuses on devising a testing a compre-
`hensive system to allow system plug-and play capability and
`empower users profiling, benchmarking and recognizing loads
`in real-time, which was holding back the deployment of such
`systems into practice.
`The rest of the paper is organized as follows: Section II
`refers to existing appliance load monitoring techniques. In
`Section III, we describe the system challenges and motivate
`our design choice. In Section IV, we detail the design of the
`RECAP system including appliance profiling, recognition, the
`database of signatures and user interface. Section V presents
`the experimental results from a real deployment. Finally,
`Section VI discusses the system and future work before
`concluding the paper in Section VII.
`II. RELATED WORK
`Many approaches to appliance load monitoring have been
`investigated. Hart paved the way with the Nonintrusive Ap-
`pliance Load Monitoring (NALM) [4]. NALM segments nor-
`malized power values, to characterize the power signal in
`successive steps or events, and match them to appliance
`signatures. The technique has achieved an average error of
`6.3% for total household energy consumption. Remaining
`challenges to NALM, which are addressed by RECAP, were
`the ability to decompose a power signal made of overlapping
`on/off events on multiple appliances, and to recognize complex
`appliance patterns.
`The load disaggregation algorithm [9] takes a very similar
`approach of comparing each change in the total power signal to
`each appliance operating range. In order to differentiate tricky
`cases where observed patterns may fit multiple appliances, a
`classification of appliances according to their frequency of use
`balances the decision making to the frequently used device.
`With ViridiScope, Kim et Al. [15] use indirect sensing to
`evaluate the power consumption of home appliances. Ambient
`signals placed near appliances estimate power consumption
`by measuring sound and magnetic field variations when ap-
`pliances are on or off. Even though sound sensors may be
`cheaper than a home energy monitor, one sensor and one
`transmitter per appliance are needed. Furthermore, more than
`the unaesthetic aspect, inaccessible or outdoor appliances as
`well as the addition of new appliances make the installation
`and correct operation of sound sensors difficult. In contrast,
`
`RECAP aims at achieving appliance recognition by deploying
`a single energy monitor clipped around the live wire of the
`main electrical unit.
`Patel et Al. [10] detect the electrical noise on residential
`power lines created by the abrupt switching of electrical
`devices and the noise created by certain devices while in op-
`eration. The approach relies on the fact that abruptly switched
`electrical loads produce broadband electrical noise either in
`the form of a transient or continuous noise. The deployment
`phase consists in collecting and recording noise signatures
`from appliances in the on, off and idle states. Aforementioned
`problems of variable power drawn by some appliances as well
`as concurrent on/off events affect similarly this approach.
`Quantum Consulting Inc. developed an algorithm with rules
`based on pattern recognition. The input is the premise level
`load data, information about standard appliances and assump-
`tions about the customer’s behavior [7]. Forty houses were
`evaluated during four summer months. Disaggregated load
`profiles have differed by less than 10%. Unfortunately, this
`system requires at least one sensor per appliance deployed for
`several days for the setting of initial operating characteristics
`. This is not a cost-effective solution and makes the system
`hardly applicable in real scenarios.
`Farinaccio et Al. [8] use a pattern recognition approach
`to disaggregate the total electricity consumption in a house
`into the major end-uses. However,
`this work does not
`address appliance profiling and assumes a constant appliance
`signature, which in reality varies with the house/room load
`and the way the appliance is set. Other techniques use wired
`solutions or employ smart sockets. This requires retrofitting
`the whole building, which is not cost-effective and may
`apply only to new structures. In contrast RECAP is based
`on a single wireless and low-cost low-power solution that
`integrates profiling of appliances, namely Unique Appliance
`Signatures (UAS), storing of signatures for
`further use,
`autonomous recognition through machine learning technique
`and a simple user interface.
`
`Overall, although the literature shows some existing re-
`search activities on this domain, existing systems address
`requirements in a disconnected manner, target specific cases
`and fail to meet system usability requirements. Up to now there
`is a lack of a low-cost tool that addresses system usability
`to empower the user with a system that integrates appliance
`profiling, generation of unique signatures, relational signature
`storage, and a basic user interface for appliance activity
`recognition, which are the main focus of this contribution to
`extend the current literature.
`
`III. CHALLENGES
`The main challenges in recognizing appliance activity are
`mainly due to the following:
`• Appliances with similar current draw: The system
`should be able to discriminate between two appliances
`with similar or same energy consumption;
`
`2
`
`
`
`• Appliances with multiple settings: Some appliances can
`be either tuned according to user needs or have different
`phases with different associated consumption, e.g. stand-
`by mode or washing cycles. The system should either
`understand the various appliance settings or recognize
`appliances based on additional data independent from the
`chosen setting;
`• Parallel appliances activity: The system should disag-
`gregate appliances activity identifying each constituent
`accounting for the total power consumption;
`• Environment noise: The system should be resilient to
`external factors such as not-profiled appliances that can
`be turned on unexpectedly;
`• Load variation: The energy provider can deploy devices
`at substation level for power factor correction, which can
`destabilize the matching with the appliance profile;
`• Long appliance cycles: The system should be able to
`cope with appliance with long working cycle, which may
`result in long profiling periods.
`In order to address these challenges, system adaptivity
`and resilience to dynamic and unpredictable environment
`are needed. To this end, the properties of existing machine
`learning techniques represent a suitable solution to reach the
`goal. In attempting identify an appropriate machine learning
`technique for recognizing appliances, we initially considered
`the following classifiers:
`• Markov Chain classifier: Although Markov Chains
`(MC) are employed in many classification and pattern
`recognition algorithms a negative aspect is that simple
`Markov Chains can merely handle one state at time. This
`means that the number of states could grow greatly if we
`map each state with a possible combination of appliances
`active at the same time. Although MC can be a suitable
`solution for monitoring a limited number of appliances,
`the system may not scale well to handle appliances in
`the order of tens via a single energy meter, which is a
`major objective of RECAP. Multistate Markov chains are
`a possible solution to address this issue but they may
`greatly increase the complexity of the system when a
`large number of appliances is profiled. Another limitation
`of this solution is its flexibility. If the user wants to add
`a new appliance, the Markov chain requires a number of
`parameters to be set, which can obstruct system usability
`in light of the fact that the system may be used by non-
`IT experts. In view of such drawbacks, we opted for the
`investigation of a more scalable classifier.
`• Bayesian classifier: A main advantage of this solution
`is the simplicity of the algorithm. Despite its apparent
`minimalism, Bayesian classifiers can give appropriate
`results with only limited data. A limitation of this type of
`classifiers is the resistance to parameter variations such as
`power variation and duration. Since parameter variations
`are a key elements due to power factor correction by the
`energy providers and signature aging control existing cir-
`cuit breakersThis factor is key for the scope of appliance
`
`recognition
`In contrast to previous techniques, Artificial Neural Net-
`work (ANN) to perform appliance recognition are manifold
`including: (1) the ability to handle any type of data (2) the
`unnecessary prior understanding of appliance behaviour; (3)
`the easy extensibility to higher number of inputs, many types
`of values or dissimilar kind of data; (4) the learning process
`that can be automated for example through additional profiling
`sensors that can turn on/off appliances remotely; (5) the ability
`to learn while running through mechanisms of error feedback
`from the user; (6) the ability to handle multiple simultaneous
`appliance states. In contrast, a drawback of the ANN solution
`is the lengthy training process that may take few minutes, e.g.
`in the presence of more than 15 appliances to profile or if
`some appliances have long signatures e.g. a washing machine
`with a multi-state signature.
`
`IV. SYSTEM DESIGN
`
`A. Data Acquisition System
`Recent standardization efforts have generated an increasing
`trend towards the integration of sensor systems in building au-
`tomation systems, allowing the connection of low-cost sensing
`and monitoring units and the gathering of energy consumption
`information in real-time.
`
`Fig. 2. Energy Monitoring Data Acquisition System
`
`Although the RECAP system is independent from the com-
`munication protocol used by the energy monitor, the unit used
`for testing transfers data via a ZigBee-based acquisition system
`to a gateway connected to a local machine, which connects
`to either a local or remote relational database for storage,
`as shown in Figure 2. The RECAP system resides on the
`local machine and processes energy data as they arrive from
`the network. In particular RECAP is able to firstly generate
`appliance signatures and then train an ANN to recognize
`appliance activities on the spot. This starts with the appliance
`profiling phase, a one-off procedure that allows RECAP to
`characterize appliances that the user wants to recognise. The
`profiling will create a set of unique appliance signatures that
`will then be used for the real-time activity recognition. To
`keep record of appliance activity times, once an appliance is
`turned on/off, the system records this into a dedicated table in
`a remote database.
`
`3
`
`
`
`B. Appliance Profiling
`A crucial aspect to consider is what parameters will con-
`tribute to the generation of a given signature. For example
`the real power consumption can discriminate between appli-
`ances with dissimilar power consumption but may fail when
`appliance consumption is similar.
`In order to identify the important constituents for a unique
`appliance signature, we now highlight the main electrical
`parameters for an appliance working on alternate current (AC).
`According to its internal circuit, an appliance can be of
`resistive, inductive, or capacitive, predominance. For example
`a kettle is almost purely resistive while a fan can be predom-
`inantly inductive. Inductors and capacitors affect the power
`consumption by shifting the alternate current with respect to
`the alternate voltage. In particular, capacitors delay the current
`with respect to the voltage while the opposite happens for
`inductors. Considering that the power is the multiplication
`of voltage and current, if voltage and current are shifted,
`the power transferred to the appliance is less. This effect is
`captured by the active and reactive power components, which,
`in mathematical terms, correspond to real and imaginary part
`respectively, as shown in Figure 3. In general, appliances
`work through the real power (active), while the reactive power
`(passive) is due to the presence of storage elements in the
`appliance circuit (inductors or capacitors), does not work at
`the load and heats wires. Pure resistive appliances show no
`shift of current and voltage, the reactive part is null and all
`the power is transferred to the load. In contrast, the larger
`the current/voltage shift the greater the imaginary component.
`Reactive and active powers are key parameters to calculate
`the power factor, which is captured by the energy meter.
`Equation 1 reports the relation between the active, reactive
`and power factor.
`
`S = P + jQP f = P/|S|
`
`(1)
`
`where S = Apparent/Complex power, Q = Reactive Power,
`P = Active Power, P f = Power Factor, and |S|= real part of
`the apparent power.
`
`C. Unique Appliance Signature
`Based on the relations between the power components and
`how they map to appliance types, this section introduces the
`constituents of a unique appliance signature. The real power is
`the first important constituent that can discriminate appliances
`of dissimilar consumption. To address appliance with similar
`consumption, the power factor can discriminate between ap-
`pliances of resistive, capacitive and inductive types. Following,
`the peak current relates to the appliance circuit specifics, as
`it represents the maximum amount of energy the appliance
`allows before reacting. RECAP collects also RMS current
`that provides consumption information independently from the
`voltage given by the energy provider. Finally, peak voltage
`and RMS voltage relate to the specific voltage provided
`when the signature is made. Overall, the system identifies 6
`constituents to generate a unique appliance signature, which is
`the base to discriminate between multiple appliances activity.
`Additional factors captured when profiling appliances are the
`signature length and the meter sampling frequency. These
`parameters are key to translate signatures from dissimilar
`types of energy meters into a standard signature. In fact,
`when profiling appliances, users may generate signatures of
`dissimilar duration in order to capture diverse appliance power
`modes. For example, an electric oven presents an initial
`period of almost constant current draw followed by periodic
`deactivations when the set temperature is reached a shown in
`Figure 4. Finally, to avoid inconsistencies between signatures
`generated with meters at dissimilar sampling frequencies, RE-
`CAP implements a simple function that translates signatures
`into a standard frequency before storing it in the relational
`database. Figure 4 shows power signatures for 4 appliances of
`different lengths and standard sampling frequency of 1 value
`per minute.
`
`Fig. 3. Relation between reactive and active power
`
`Fig. 4. Active Power Signatures for 4 appliances
`
`D. Signature Database
`Once a new appliance is profiled, the signature together
`with some metadata relative to the appliance model and
`location are stored locally and duplicated in a remote database.
`
`4
`
`
`
`The duplication allows the creation of a common repository
`of signatures used to train the ANN and share signatures
`with other users, namely Unique Signature State Informa-
`tion (USSI) database. In fact, providing a common signature
`repository for multiple homes can progressively reduce the
`initial training phase required by RECAP. Figure 5 shows
`the USSI database associated to RECAP . USSI consists of
`6 main relational tables. 3 main tables, namely Captured
`Parameters, Physical and Environmental relate directly to the
`signature. Since certain types of appliance type with either
`dissimilar models or from various manufacturers are likely to
`have dissimilar signatures, the table ”Physical” captures the
`specifics of the appliance.
`As the USSI database grows, it is necessary to provide tech-
`niques to present to the user an initial standard set of relevant
`signatures. Through the Environmental table the system can
`provide a common list of appliance signatures based on user
`location (password protected). It is in fact common to have
`same appliance models concentrated within the same area or
`region (e.g. electric showers are very common in Ireland and
`UK while a certain HVAC model are more common in warmer
`countries). By using the RECAP interface, the user can then
`browse the list of appliance models in the area or search
`for other signatures should the appliance be not in the list.
`Currently, the USSI system in RECAP is implemented in an
`SQL-based relational database.
`Furthermore, the Environmental table provides information
`of surrounding conditions during measurements as this may
`affect the signature accuracy. USSI was designed with a broad
`use in mind such as a large number of signatures generated
`by contributors. To address multiple contributors for the same
`signature, the database implements a Contributor table that
`includes a confidence rate, which increases according to the
`reputation of the contributor. We envision that a reputation
`would increase based on collection of opinions from other
`users. The USSI system can handle multiple signatures of the
`same appliance ID according to the Signature Property table.
`Similar to contributor reputation, the Energy Meter table stores
`the accuracy of the energy meter, which can be used to tune
`the appliance recognition algorithm. For example, in RECAP
`this would enable testing the meter accuracy and associate
`accuracy levels to different activation functions.
`E. Training and Recognition
`Following the profiling phase, the generated signatures are
`used to train an ANN for the recognition of appliances.
`The basic element of an ANN is a neuron, which can be
`represented as a simple succession of mathematical operations,
`such as weight balancing, sum and an activation function as
`shown in Figure 6.
`Each input of a neuron is balanced by a different weight
`and is then aggregated into an activation function that can be
`as simple as a step function, or a more complex function such
`as hyperbolic tangent. An ANN consists of several neurons
`interconnected. Figure 7 shows a common type of ANN, 3-
`layer ANN, which is in fact the type adopted for RECAP, as
`
`Fig. 5.
`
`Integration of unique signature state information
`
`Fig. 6. Single neuron showing input weigths, weighted sum and activation
`function
`
`provides a judicious balance between complexity and response
`time. The first layer consists of Inputs Neurons i.e., neurons
`with one or more inputs connected to external or internal data.
`The second layer consists of Hidden Neurons that have inputs
`connected to the outputs of the first layer and are not in direct
`relation with inputs and outputs of the ANN. The Third layer
`consists of Output Neurons that have inputs connected to the
`outputs of the hidden neurons. Output neurons represent the
`direct outputs of the ANN. The connections between the layers
`and neurons can vary. For example, the input layer can be
`connected to the output of the ANN in order to provide a
`feedback informing the new input state about the previous
`output. RECAP uses a similar feedback mechanism to improve
`
`5
`
`
`
`the accuracy by allowing the user to notify the system of an
`incorrect guess.
`
`Fig. 7.
`neurons
`
`An example of 3-layer ANN showing input, hidden and output
`
`We now describe the ANN learning process. At the begin-
`ning of the learning phase, all weights are random. Weight co-
`efficients of neurons are modified using the data from a train-
`ing data set, based on combination of appliance signatures.
`The data and corresponding weight modification propagated
`through the hidden layer and then the output layer. During
`training phase, the output signal generated by the ANN is
`then compared against the desired value, namely the target,
`as given in the training data set. The difference between the
`target and the output layer of the network is called error δ. This
`error is then back propagated to the hidden neurons and the
`input neurons and the weight of each neuron may be modified
`accordingly. Equation 2 provides the calculation used to tune
`the weights.
`
`W n = W o − (δ ∗ Lr)
`
`(2)
`where W n is the new Weight, W o is the old weight and
`Lr is the learning rate sets to a constant value to avoid
`inconsistencies due to rapid weight changes.
`The back propagation ends when all the layer weights are
`adjusted and the ANN is ready for another ”wave” of data.
`Naturally, the more sophisticated the training data set, the finer
`the weights will be tuned to recognize any combination of
`appliance activation. To improve system usability, RECAP im-
`plements an automatic training program (ALP) that allows
`autonomous training of the system. ALP uses the generated
`signatures and creates a training data set with all possible
`combinations of appliance activity, which is then used to tune
`the neuron weights autonomously.
`It is important to determine the correct number of neurons
`and the configuration of the network. In general a large number
`of hidden neurons may result in long training times and a
`
`system that may perform extremely well on the training data
`set but cannot handle unseen data. In contrast, an inadequate
`number of neurons cause the inability to handle complex
`combination of appliances and poor results. Although the
`literature offers several programs to find the optimal number of
`neurons, we opted for tuning the ANN empirically. A number
`of trials allowed the identification a saturation point after
`which the network showed little further improvement. The
`current version of RECAP adopts 6 input neurons matched
`with 6 hidden neurons, which performed adequately for the 6
`inputed parameters used to generate the signatures.
`Finally, an important aspect of the ANN is the Activation
`Function, which shapes the output of the neurons. Considering
`that the output of the network should correspond to one of the
`profiled appliances, after several empirical trials we adopted a
`general sigmoid function, which is close to a threshold function
`with smooth angles to allow some neuron uncertainty in case
`of errors within the order of 5% or more e.g., to account for
`small variation of power delivered by the energy provider.
`F. Graphical User Interface
`In order to facilitate user interaction, RECAP provides a ba-
`sic user interface, which integrates the profiling of appliances,
`the network training and the final displaying of appliance
`activity. These functionalities are divided in 4 separate panels
`as follows:
`• Add An Appliance: As shown in Figure 8, this panel
`allows the user entering a JPEG image, name/model and
`type of appliance to profile. The panel provides also a
`predetermined list of appliance types already available in
`the database for the convenience of the user. Once the
`information is entered, the user can press on ”save and
`go to next step”, which is the appliance profiling phase.
`
`Fig. 8. The ”Add Appliance” panel used for inserting the appliances to be
`profiled
`
`• Generate A Signature: This panel, shown in Figure 9,
`is used to record appliances profile in order to generate
`
`6
`
`
`
`signatures. The panel is provided with a list of added
`appliances; by clicking on the button ”Start Recording”,
`the user will be prompted with ”please, turn on the
`appliance to be profiled”. The system will automatically
`recognize the delta variation relative to the appliance
`being turned on and will start recording the parameters.
`For the duration of the profiling period, the button will
`stay red and display ”recording in progress”. During
`this period, the system requires that no other appliances
`be turned on or off. The button will turn green after
`receiving a sufficient number of values per each energy
`parameter. Based on some initial experiments, the mini-
`mum number of values to build an accurate signature has
`been set to 4, which correspond to 4 packets from the
`energy meter. Obviously, the time needed to generate a
`signature depends on the meter transmission rate. In our
`experiments, the meter was transmitting at a rate of 1
`packet/min as preset by the manufacturer. Should system
`detect some anomalies during this recording period, the
`appliance profiling is aborted and the user may restart the
`process.
`
`and steady signatures increase the system confidence and
`provide a better performance forecast.
`
`Fig. 10. The ”Train system” panel used for training the ANN with profiled
`appliances
`
`• Recognition: As shown in Picture 11, this panel displays
`the real-time activity of appliances. Active appliances
`are displayed in colour while inactive appliances are
`displayed in black and white. At the bottom of the panel,
`a statistics window allows following the historical data
`relative to appliance activity and energy cost associated
`to each appliance.
`
`Fig. 9. The ”Record Values” panel used for generating appliance signatures
`
`• Training: This panel, as shown in Figure 10, consists of
`a list of profiled appliances, on the left hand side of the
`figure, which will be used to train the network. In case of
`some unutilized appliances in the list, their removal from
`the training list is done through the Remove button on the
`right-hand side of the panel. The training of the ANN
`is performed by pressing the ”Start learning” button,
`while the expected recognition performance and training
`progress bar are on the lower part of the panel. Over-
`lapping appliance activities during profiling may cause
`jagged signatures and poor performing training periods.
`In case the recognition performance does not meet the
`requirements dictated by the application, the user can
`go back to the ”Record Values” panel and re-record
`the uncertain signatures. We also identified that longer
`
`Fig. 11. The ”Watt’s On” paned used for displaying appliances activity and
`statistics
`
`V. EXPERIMENTATIONS
`The experimentations of the RECAP system have been con-
`ducted in the kitchen area of the CLARITY centre at Univer-
`sity College Dublin. We connected the energy monitor to the
`
`7
`
`
`
`miniature circuit breaker in the electrical unit, which controls
`the whole kitchen area and the corridor including sockets and
`lights. For the experiments, we used the off-the shelf ZEM-30
`ZigBee-based Energy Monitor from Episensor [3], as shown
`in Figure 12. This energy monitor is equipped with a current
`transformer (CT) clipped around the live wire of the consumer
`electrical unit.
`
`Fig. 12. Episensor ZEM-30 ZigBee Energy Monitor
`
`The first objective of the experiment was to recognize
`3 main appliances present in the kitchen area: a kettle, a
`microwave and a fridge. Initially, 3 appliances were profiled
`and signatures produced for them. We tested the system for
`one week and instructed people using appliances and sockets
`in the area to annotate time of usage. During the week, teh
`annotation reported that the 3 main appliances were activated
`several times per day together with other lower consuming
`devices, such as lights and a small fan. Occasionally people
`used the sockets to plug in their laptops and phone rechargers.
`This initial trial showed accuracy over 95% due to a much
`higher power consumption of the 3 appliances than the lights
`and other low consuming appliances.
`Given this initial encouraging performance, we decided
`to test the system in a more demanding scenario in which
`a user profiles an appliance that shows both similar power
`consumption and power factor characteristics as other existing
`signatures. We identified an electric fire as a resistive appliance
`with similar consumption of microwave and kettle. The electric
`fire was introduced in the kitchen and its signature generated.
`The objective o