GB2502062A

GB2502062A - Analysing power consumption of electrical appliances to determine a person's state of health

Info

Publication number: GB2502062A
Application number: GB1208398.6A
Authority: GB
Inventors: Georgios Kalogridis
Original assignee: Toshiba Research Europe Ltd
Current assignee: Toshiba Europe Ltd
Priority date: 2012-05-14
Filing date: 2012-05-14
Publication date: 2013-11-20
Also published as: GB201208398D0

Abstract

A data analysing unit for analysing electrical power usage of one or more electrical appliances in a dwelling comprises a means S21 for monitoring electrical power, a means S22 for defining an electrical signature of energy usage in the dwelling, and a means S23 for comparing the electrical signature with an expected electrical signature. If an anomaly S24 exists between the electrical signature and the expected electrical signature, a change S25 in the physical well-being of a person resident in the dwelling may be identified. For example, if an appliance such as an oven is left switched on for a prolonged period, this may indicate that the person experiences memory loss. The anomaly in the electrical usage may be associated with a particular symptom e.g. heating use may indicate fever. An alert may be issued and/or an appliance may be controlled remotely in response to anomaly detection.

Description

I

Data analysing unit for analysing electrical power usage data

FIELD

Embodiments described herein relate generally to methods and systems for perc&ving changes in the physical well being of a person in a dwelling, based on electrical energy usage withEn the dwelling.

BACKGROUND

The development of personal biosensors has made it easier to monitor an indMdua]rs health risks, without that individual having to leave their own home. As one example, heart sensors can be fitted to a patient and used to record cardiac events, develop a heart sound energy signature and identify both normal and abnormal heart sound recordings. The data collected by these sensors can then be sent to e.g. relatives, or medical practitioners, at locations remote from the person's dweilFng. If the data indicates that the patient is at risk, these third parties can take appropriate steps to mitigate the risk.

However, such sensors are costly, and in addition, place an extra power burden an the household. Where an older member of a family is living by themselves, for example, relatives of that family member may wish to be alerted to changes in the person's health without the cost of investing in such biosensors.

It follows that that there is a need to provide assisted living and medical services to people without the use of special eqLripment such as home medical appliances or biosensors.

BRIEF DESCRIPTION OF FIGURES

Embodiments of the invention will now be described by way of example with reference to the accompanying drawings in which: Figure 1 shows a schematic of a system according to an embodiment; Figure 2 shows a series of steps carried out by a system according to an embodiment; Figures 3a and 3b show examples of electrical signatures obtained in accordance with an embodiment; Figure 4 shows an example of a method for sampling electrical power usage according to an embodiment; Figure 5 shows a series of steps carried out by a system according to another embodiment; Figure 6 shows a series of steps carried out by a system according to another embodiment; Figure 7 shows a series of steps carried out by a system according to another embodiment; and Figure 8 shows a schematic of a system according to another embodiment.

DETAILED DESCRIPTION

According to a first embodiment, there is provided a data analysing unit for analysing electrical power usage data, the data analysing unit comprising: means for receiving data from an electrical power monitoring unit, said data indicating the electrical power usage of one or more electrical appliances in a dwelling, means for defining a first electrical signature of energy usage in the dwelling, based on the received data; means for comparing the first electrical signature with an expected electrical signature for the dwelling, means for determining whether an anomaly exists between the first electrical signature and the expected electrical signature, and for identifying whether a change in the physical well-being of a person resident in the dwelling has occurred, based on the detected anomaly.

In some embodiments, the means for determining whether an anomaly exists between the first electrical signature and the expected electrical signature is configured to determine whether the person resident in the dwelling is unwell.

In some embodiments, the received data comprises a series of data values indicating the power supplied to each appliance at predetermined intervals in time.

In some embodiments, the expected electrical signature comprises a predetermined sequence of values. The data analysing unit may be configured to determine whether any anomalies exist between the first electrical signature and the expected electrical signature by calculating a deviation between the received data and the predetermined sequence of values.

In some embodiments, the data analysing unit is configured to determine that an anomaly is present when the deviation is above a predetermined threshold.

In some embodiments, the data analysing unit is configured to identify a physical symptom associated with the anomaly by identifying a plurality of possible symptoms and determining a probability that the identified anomaly relates to each one of those symptoms.

In some embodiments, for each symptom, the data analysing unit is configured to access a predetermined sequence of data values representing an electricar power usage profile associated with that symptom, and to compare the predetermined sequence of data values with the data received from the power monitoring unit.

In some embodiments, the data analysing unit is configured to periodically update the expected electrical signature using the data received from the power monitoring unit.

In some embodiments, the data analysing unit is configured to check whether data received from the power monitoring unit comprises an anomaly before the data is used to update the expected electrical signature.

According to a second embodiment, there is provided a system comprising: a data analysing unit according to the first embodiment, and a response determination unit for determining a response to the anomaly.

In some embodiments, the response determination unit is configured to determine a response by accessing a look up table that associates different anomalies with appropriate responses.

In some embodiments, the system comprises a response implementation unit, the response implementation unit being configured to implement a determined response by: sending an alert to a person resident in the dwelling or to a third party not resident in the dwelling; and / or: adjusting the power supplied to one or more of the electrical appliances.

In same embodiments, the system includes a power monitoring unit for monitoring the electrical power usage of the one or more electrical appliances. The power monitoring unit may be configured to transmit data representing the monitored electrical power usage to the data analysing unit.

In some embodiments, the power monitoring unit is configured to sample the electrical power usage of the one or more electrical appliances at predetermined intervals.

In some embodiments, the power monitoring unit is configured to transmit the data representing the electrical power usage of the one or more electrical appliances to the data analysing unit at intervals larger than those used to sample the electrical power usage of the appliances.

According to a third embodiment, there is provided a method of detecting changes in the physical well-being of a person resident in a dwelling, the method comprising: receiving data from an electrical power monitoring unit, said data indicating the electrical power usage of one or more electrical appliances in a dwelling, defining a first electrical signature of energy usage in the dwelling, based on the received data; comparing the first electrical signature with an expected electrical signature for the dwelling, determining whether an anomaly exists between the first electrical signature and the expected electrical signature, and if so, identifying whether a change in the physical well-being of a person resident in the dwelling has occurred, based on the detected anomaly.

In some embodiments, the step of determining whether an anomaly exists between the first electrical signature and the expected electrical signature comprises identifying whether the person resident in the dwelling is unwell.

In some embodiments, the expected electrical signature comprises a predetermined sequence of values, and the step of determining whether any anomalies exist between the first electrical signature and the expected electrical signature comprises calculating a deviation between the received data and the predetermined sequence of values.

In some embodiments, the method comprises determining whether or not the calculated deviation is above a predetermined threshold.

In some embodiments, the method comprises identilying a physical symptom associated with the detected anomaly by identifying a plurality of possible symptoms and determining a probability that the identified anomaly relates to each one of those symptoms.

In some embodiments, for each symptom, the data received from the power monitoring unit is compared with a predetermined sequence of data values representing an electrical power usage profile associated with that symptom.

In some embodiments, the expected electrical signature is periodically updated using the data received from the power monitoring unit.

In some embodiments, the data received from the power monitoring unit is checked to see if the data comprises an anomaly before the data is used to update the expected electrical signature.

In some embodiments, the method comprises determining a response to the anomaly.

In some embodiments, the response is determined by accessing a look up table that associates different anomalies with appropriate responses.

In some embodiments, the response comprises: sending an alert to a person resident in the dwelling or to a third party not resident in the dwelling; and I or: adjusting the power supplied to one or more of the electrical appliances.

According to another embodiment, there is provided a computer readable storage medium comprising computer executable code that when executed by a computer will cause the computer to carry out a method according to the third embodiment.

Some embodiments detect anomalies by analysing data obtained from smart meters that record the energy consumption of a dwelling.

Some embodiments include a system for protecting the privacy of (detected) medical symptoms by means of energy management.

Figure 1 shows a schematic of a system I including a data analysing unit 3 according to a first embodiment. The system 1 also includes a power monitoring unit 6 that is in communication with the data analysing unit.

Although shown as part of a single block in Figure 1, there is no requirement for the power monitoring unit and data analysing unit to be co-located with one another; for example, the two may be located in different buildings, with communication between the two taking place over a network connection.

The power monitoring unit monitors the electrical power supplied to different appliances 2, 4, 6 within a person's dwelling. Such appliances may be powered by a mains source inside the dwelling. The electrical appliances may include, for example, heaters, television sets, electric ovens, cookers, and kettles.

In some embodiments, data is supplied to the power monitoring unit by one or more power sensors, which sample the flow of electricity to different power outlets inside the dwelling. For example, the sensors may sample the power supplied to each outlet from a mains power source. Such sensors may include smart meters that have been pre-installed inside the dwelling, in order to gather data for remote reporting of electricity usage.

In other embodiments, the power monitoring unit may comprise a hardware module that itself directly monitors or samples the flow of electricity to different power outlets.

The power monitoring unit may be specially designed to disaggregate power supplied to different home appliances without the need for additional sensors or appliance-based meters.

Data obtained by the power monitoring unit is communicated to the data analysing unit, which in turn uses this data to infer information about the usage of electrical power within the dwelling and to compare this with expected patterns of usage. In the event that there is a mis-match or anomaly between the data obtained by the power monitoring unit and the expected pattern, the data analysing unit may determine that the person resident in the dwelling has experienced a change in their physical well-being.

Figure 2 shows the steps carried out by the data analysing unit in more detail. In step 21, electrical usage data is received from the power monitoring unit. The data is then used to define a first electrical signature (step 22).

In step 23, the data analysing unit compares the first electrical signature with an expected electrical signature. In step 24, any anomalies between the two signatures are detected. Depending on the type of appliance in question, and the duration and/or frequency of the anomaly, the data analysing unit may identify various physical symptoms associated with the person resident in the dwelling (step 25).

The concept of an anomaly can be understood by reference to Figures 3a and 3b.

Figure 3a shows an example of a first electrical power signature generated by a particular electrical appliance. In the example of Figure 3a, the appliance has two states -an on state and an off state (XON and X0, respectively). The state of the appliance can, therefore, be characterised during the time intervals (t1, t2, ... t) by a sequence of data symbols (x1, x2, ...xw) where x1 = X0, 0rXQFF. In practice, each symbol will be obtained by sampling the electrical power usage of the appliance at a particular point in time. The symbols may themselves be represented as different numerical values. As shown in Figure 3a, the appliance varies between the on and off states over time.

Figure 3a also shows the expected electrical signature for the appliance in question.

The expected electrical signature represents the normal (expected) use profile of an appliance. In other words, the expected electrical signature represents the pattern of usage of the appliance by a healthy individual during their normal daily course of activities.

The expected electrical signature comprises a sequence of normal (expected) symbols, for example, XON, or XOFF, that represents the expected state of the appliance at a particular point in the day. The sequence can be divided into blocks of symbols, where each block reflects a particular period. For example, the expected electrical signature may comprise 24 blocks of symbols, with each block representing a single hour in a day.

It is possible that a persons pattern of power usage may vary from day to day, without necessarily reflecting a change in the health profUe of that person. In order to accommodate small changes in the user's power usage, each block of symbols in the expected electrical signature is also accorded an expected deviation. The expected deviation reflects the extent to which the symbols in a block of data may vary from the expected values without signifying an anomaly.

The expected deviation can be calculated, for example, by assembling blocks of non-anomalous symbols that are recorded for the same time period on different days, and determining an average block of symbols for that time period. The deviation between each block and the average block can be recorded. The normal distribution of the obtained deviation results can then be used to obtain the expected deviation for that time period.

In the example shown in Figure 3a, the expected electrical signature and the measured signature coincide with one another. More specifically, the variation between the monitored electrical signature and the expected signature falls within the expected deviation. In this instance, therefore, the person in the dwelling is using the appliance in accordance with a predicted trend.

Figure 3b shows an example in which the electrical signature obtained by monitoring the electrical power usage of the appliance differs from the expected electrical signature. Specifically, during the period t1 to t2, the appliance is determined to be in an "on" state (X0,4, whereas it would normally be expected for the appliance to be in the "ocr state (Xo11) during that period, as shown by (he profile of the expected electrical signature. The difference between the two signatures constitutes an anomaly in the electrical signature of the appliance.

In order to obtain the first electrical signature, it is necessary to sample the electrical power usage of the appliance at different time intervals. Figure 4 shows an example of a sampling method used in some embodiments. As in Figure 3, the system monitors the electrical power usage of an appliance having two states, SOn" and "Qff' represented by the symbols X0N and XQFF, respectively.

The time periods denoted a, b, c denote intervals during which the electrical power usage of the appliance is sampled. The frequency with which the electricar power is sampled is chosen such that in any one interval, the likelihood is that the appliance will not switch state more than once. In some embodiments, the sampling interval may vary from milliseconds up to seconds.

In practice, the power monitoring unit may gather symbols representing the power usage of the appliance and send these as separate blocks to the data analysing unit.

In Figure 4, the interval d, which is longer than the time intervals a, b, c, represents the interval between sending successive blocks of symbols to the data analysing unit.

The interval d should be long enough to allow accurate anomaly detection results, but short enough to ensure that events are analysed and any response initiated soon after a change in the physical well-being of the person is identified. Typically, the interval between sending successive blocks of symbols to the data analysing unit will range from a few minutes up to a few hours.

The interval d may, in practice, also depend on other factors. For example, where the power monitoring unit and the data analysing unit communicate with one another via a network connection, the interval d may be determined in part by how quickly information can be uploaded and sent over that network connection.

In addition to the examples shown in Figures 3 and 4 (which relate to a single electrical appliance), some embodiments may involve monitoring the electrical power usage of multiple electrical appliances. In such cases, there will be a set of appliances A = (A1, A2... A. Each appliance may have S possible states of operation (the states "on" or "off" illustrated in Figure 3 are just two examples of such states).

When all of the appliances are considered together, the total number of distinct electrical states will be given by Equation 1:

M

-s. (Equation 1)

-I

Where there are multiple appliances present, the process of generating the electrical signature can be simplified by only registering actual changes in the states of the appliances. In other words, if the state of an electrical appliance does not change from one sampling interval to the next, the state of the electrical appliance need not be recorded for that second interval; instead, the absence of a symbol for an appliance in any interval will serve as an indication that the state of the appliance has not changed since the last time a symbol was recorded for that appliance.

Similar to Figure 4, the frequency with which the electrical power is sampled can be chosen such that in any one sampling interval, the likelihood is that no more than one appliance will switch state. Since only one of the appliances will change state in any sampling interval, it is possible to represent the eFecirical power usage of the plurality of appliances by a single sequence of symbols, with each symbol signifying a change in state of operation of one or other of the appliances.

The process of using a detected anomaly to identify a symptom associated with the person in the dwelling will now be described by reference to Figure 5. As previously explained, the process begins when a block of symbols is sent from the power monitoring unit to the data analysing unit. At this point, the deviation between the expected electrical signature and the monitored electrical signature is determined, and may be compared with a threshold value. If the computed deviation is larger than the expected (normal) deviation, then the data analysing unit will determine an anomaly to be present.

In practice, the step of detecting anomalies between the monitored electrical signature and the expected electrical signature can be performed according to a host of different techniques. For example the anomary detection may include a technique derived from Kolmogorov complexity measures, as described in the paper by E. Keogh, Slonardi, and Caratanamahatana "Towards Parameter Free Data Mining" in proceedings of the tenth ACM SIGKKD International Conference on Knowledge Discovery and Data Mining ACM Press New York Ny, USA 206-215. In this paper, the authors propose a technique derived from Kolmogorov complexity measures to compute an abstract difference metric between two sequences of symbols, called the Compression Based Dissimilarity Measure (CDM). This metric is based on the information content of the sequences and can be simply approximated using standard compression algorithms such as DEFLATE or BZIP2, which have freely available implementations.

An alternative to CDM is the measure of K-Divergence as proposed by T.M. Cover and IA. Thomas in the publication "Elements of Information Theory", .John Wiley & Sons, Inc. New York, Ny, USA, 2006. This is an information theoretic metric that computes the distance between two data systems in terms of the probability of obtaining one data sequence, given a second data sequence. The metric has been used to calculate the privacy of a certain transformation of data1 but it can also be applied to detecting unusual events, as disclosed for example in G. Kalogridis and S. Denic, "Data Mining and Privacy of Personal Behaviour Types in Home Energy Signals", published in the third IEEE International Workshop on Privacy Aspects of Data Mining, December 2011, Vancouver, Canada.

Embodiments may use any method that is capable of identifying deviations between the sequence of symbols in the monitored electrical signature, and those in the expected electrical signature.

In the event that an anomaly has been identified (step S 51) the data analysing unit proceeds to identify a list of possible symptoms associated with the anomaly (step S 52). Different symptoms may be associated with different changes in the person's wellbeing. Examples of symptoms are that the person is cold, that the person is hot, that the person is experiencing memory loss or that the person is tired.

The data analysing unit may access one or more of a plurality of predetermined blocks of symbols, where each predetermined block of symbols reflects a particular electrical usage pattern associated with a particular event. The predetermined symbols may themselves be encoded by compiling a history of the person's electrical power usage under different circumstances (for example, during both periods of illness and good health).

For example, a block of symbols that reflects a heating appliance being switched on for an extended period may serve to indicate that the person in the dwelling is cold.

Feeling cold could be a symptom of influenza, for example. Alternatively, a heightened use of a fan or air-conditioning unit could suggest the person has a high temperature.

The person may, therefore, have sustained a fever. Further, an unusually prolonged operation of the cooker/oven could suggest that the person has experienced memory loss, which could be a symptom of dementia, for example.

Thus, each one of the predetermined blocks of symbols will reflect a particular pattern of electrical power usage, indicative of an underlying physical symptom of the person in the dwelling. In this way, each of the predetermined blocks of symbols can be considered as representing a particular physical symptom of the person in the dwelling.

For each predetermined block of symbols, the data analysing unit calculates a probability that the anomaly reflects a symptom represented by that respective block (step S 53). To do so, the data analysing unit may compute the deviation between the received block of signals (i.e. the sequence corresponding to the detected anomaly) and each of the predetermined blocks of symbols. The probability that the anomaly reflects a particular symptom can be inferred from the deviation between the received block of symbols and the block representing that symptom.

At step S 54, the data analysing unit identifies the symptom for which the probability is greatest, and so which is most likely to explain the anomaly. In some embodiments, the identification of one or more symptoms may serve as a bridgehead to identifying a specific ailment that the person resident in the dwelling is suffering from.

Figure 6 shows an example embodiment in which the expected electrical signature is periodically updated with information received from the power monitoring unit. In this example, steps S 61 -65 correspond to steps S 21 -25 of Figure 2, respectively. If it is established that a newly arrived block of symbols does not constitute an anomaly, then those symbols are used to update the expected electrical signature (step S 66).

For example, those data symbols may be used to refine the expected deviation.

In addition, an optional feedback mechanism may be included, whereby blocks of symbols that are initially determined to be anomalous, but which are later identified to be false alarms are also used to recalibrate the expected electrical signature. Such a recalibration may again also include re-computing the expected deviation for the block in question.

In some cases, recalibration of the normal medical profile may be performed infrequently (i.e. not every time a non anomalous block of symbols arrives). For example, the expected electrical signature may be updated once a day.

Figure 7 shows an embodiment in which the system is configured to formulate a response based on a change in physical wellbeing indicated by an anomaly in the monitored electrical signature. In this example, steps S 71 -75 correspond to steps S 21 -25 of Figure 2, respectively. If the change in wellbeing is detected with a high enough probability, the system may follow one or more of the following predetermined procedures; 1. Send information, statistics and alerts to a unit in the dwelling.

2. Send an alert over some public network (cellular or Internet) such as SMS or email, to a user (either a resident of the dwelling) or relatives, if appropriate and if pre-authorised by the user.

3. Create an automated call for emergency services (e.g. fire, ambulance or police).

4. Send a command to a unit in the dwelling to neutralize the data in order to protect the privacy of associating medical events.

A further embodiment of the system will now be described with reference to Figure 8.

In this embodiment, a power monitoring unit 9 is situated within a dwelling 11, with the data analysing unit 13 being situated in a medical centre 15 remote from the dwelling.

The power monitoring unit may comprise appliance detection software that analyses smart metering data from the dwelling, or a hardware module specially designed to disaggregate home appliances (i.e. without the need for additional sensors or appliance-based meters).

The power monitoring unit communicates with the data analysing unit via a network interface 1/that supports private and secure communications between the dwelling and the medical centre.

The dwelling and the medical centre also include respective coms modules 19, 21 (such as a home hub) that interconnect components of the system housed En the dwelling and the medical centre, ]especlively.

The comms module located in the dwelling may also be used to communicate additional data to other destinations. For example, the comms module may communicate smart metering data obtained from the dwelling to data centres and I or utilities companies. In order to maintain the privacy of data sent between the dwelling and medical centre, any additional data sent from the hub is communicated via a different network interface to that used to communicate information between the dwelling and the medical centre.

The network interfaces and the entities that handle data support standard cryptographic functions to corroborate confidentiality and integrity of communicated and stored data, authorisation, and mutual authentication among all entities concerned.

The power monitoring unit detects and assembles data representing the usage profile of electrical appliances within the dwelling and communicates this information via the network to the data analysing unit. The data analysing unit in turn determines whether any anomalies are present in the data and if so, identifies any likely changes in the person's physical wellbeing.

In addition to the power monitoring unit and data analysing unit, the system shown in Figure 8 also includes a response determination unit 23 that communicates with the data analysing unit. Where the data analysing unit identifies a particular symptom associated with the person in the dwelling, the response determination unit identifies an appropriate means of response. For example, the response examining unit may access a look-up table of responses associated with different types of symptoms.

Such responses may include any one of those discussed above in relation to Figure 7.

The response determining unit in turn may communicate with a response implementation unit 25 that puts the response into effect. For example, where the response determining unit determines that an appropriate response is to contact an ambulance, the response implementation unit may route a telephone call to a hospital or the emergency services accordingly.

Although the response implementation unit is shown as being located in the dwelling in Figure 8, it is possible that the response implementation could be located in the medical centre or any other location remote from the dwelling.

In some embodiments, the response implementation unit is used to further protect the privacy of data linked with medical events described above. In such embodiments, the response determining unit sends commands to the response implementation unit to actuate the operation of appliances and energy storage devices, for example batteries.

The objective is to mask electrical events and the corresponding electrical energy signatures in order to help prevent other parties detecting the electrical signature from detecting anomalies as described above.

In more detail, the data analysing unit or response determination unit may be pre-programmed with instructions to mask an anomaly in the electrical signature of the dwelling. For example, the response determination unit may instruct the response implementation unit to change or discharge a battery when a certain block of symbols are detected in the monitored electrical signature. Alternatively or additionally, the response implementation unit may be instructed to switch on different electrical appliances, in order to emulate the occurrence of a particular electrical event or to mask the energy profile of a particular appliance. Such electrica] appliances may include, for example, smart devices such as a home smart water boiler, a smart washing machine or a smart vacuum cleaner.

Instructing the response implementation unit to switch on or switch off appliances at certain times will impact the erectrical signature generated by the power monitoring unit. More specifically, the symbols sent from the power monitoring unit to the data analysing unit will now reflect a superposition of the true underlying energy profile of the user with additional symbols caused by the activation I deactivation of different appliances by the response implementation unit.

The response implementation unit may also maintain a secure log, comprising a sequence of symbols S' that would have been received by the power monitoring unit, in the event that the response implementation unit had not been instructed to actuate any of the appliances within the dwelling. For example, the response implementation unit lb may be instructed to switch on an appliance for a 2 hour period in the afternoon. The symbols obtained by the power monitoring unit will then reflect the fact that the appliance was switched on during that period. The response implementation unit will meanwhile record a log a series of symbols indicating that the appliance in question was switched off (as would be the case if the response implementation unit had not been instructed to actuate the device). The response implementation unit will then transmit those symbols to the data analysing unit.

On receipt of the symbols from the response implementation unit, the data analysing unit will use these to determine the correct series of symbols (i.e. the series that correctly characterises the persons pattern of energy usage in the absence of interference by the response implementation unit). Each time a block of symbols arrives at the data analysing unit from the power monitoring unit, the data analysing unit will check if it has also received a corresponding block of symbols from the response implementation unit for the time period in question. If so, it will use the symbols received from the response implementation unit to recover the electrical signature that reflects the true pattern of energy usage of the person in the dwelling.

The above privacy mechanism can help to protect the privacy of medical data because only the parties that know the sequence cf symbols S' will be able to "decrypt" the block of symbols that correspond to typical user activities (without energy masking).

Examples of how the system may function in practice are provided in the following scenarios: Scenario 1 Suppose that an individual uses an average of 2 hours per day of electricity for heating, in the afternoon after work, and during the cold wFnter days. If this individual returns home from work on a cold day and does not use electricity for heating, this may pose a health risk. By comparing the block of symbols being received on an hourly basis with the corresponding average block of symbols, the data analysing unit may determine that: i) The individual has left home to go (presumably) to work. This can be assumed if no anomaly is detected during the morning, noon, and afternoon hours.

ii) The individual has come back home and stayed all evening at homer without there being a special occasion. This can be assumed if no anomaly is detected in the symbols that characterise home activities (for example, lights, TV, computer, cocker) without taking intc account symbols corresponding to heating appliances. This can be done by replacing the symbols that correspond to the different states of the heating appliance(s) with a common neutral symbol.

iii) The individual has not turned the heating on, regardless of the fact that it is a cold evening. This can be assumed if an anomaly is detected in the symbols that relate to the status of the heating appliance(s).

In the above scenario, the response determination unit may instruct the response implementation unit to phone the person in the dwelling, or to make an automated query (for example, to request that the person should press a number to signify that they are well and everything is OK in the dwelling).

The response implementation unit may also act to protect the person against third party medical mining. This protection, for example, may simply take the form of a warning sent to the person in the dwelling. In response to this warning, the user could a) turn on the heating, b) turn off the lights or the TV (remotely) in order to confirm that they are absent, or c) provide further information that invalidates the initial hypothesis.

Scenario 2 Suppose that an elderly individual gets cold during the winter and has a fever. The activities of this person within the home will typically change. For example, the individual might use the kettle more often to prepare more hot drinks, or the individual might get up more often during the night to take some medicine and change clothing.

In this scenario, symbols of electrical power usage are sampled by the power monitoring unit once a minute. The symbols are transmitted in blocks of 60 symbols once an hour. The symbols will reflect the kettle being operated at different times during the day and will also capture the lights being turned on at different times during the night.

During the day, the data analysing unit will detect an anomaly since the individual does not normally operate the kettle that often. However, based on this anomaly alone, the data analysing unit may not be able to determine with high enough probability that the person is feverish. In this instance, the response determination unit may simply send information to the response implementation unit indicating the fact that the kettEe has been used more often than usual.

During the day, the data analysis unit will detect a further anomaly, as the individual does not normally turn on the light so frequently. Given the fact that the kettle is being operated more frequently than usual, and the lights are being turned on many times during the night, the data analysis unit may determine with sufficiently high probability that the individual is feverish. In this instance] the response determination unit may instruct the response implementation unit to carry out instructions as follows: * Send a polite and discrete SMS to a close relative's mobile number indicating the potential medical risk.

* Determine, if possible, that there is sufficient heating and if so create further alerts.

The response determination unit may also send instructions to the response implementation unit to help protect privacy of the medical data. For example, the response implementation unit may be instructed to discharge a battery when the lights are turned on during the night, or to cause a water boiler to heat up at scheduled times during the following week, in order to obfuscate the energy signature created by the kettle. At the same time, as described above, the response implementation unit may record and send back to the data analysing unit the sequence of symbols S' facilitating the disambiguation of electrical events in the dwelling.

Scenario 3.

In this scenario, the oven is switched on for several hours. The anomaly is detected as described above and the user is notified. If, for example, a similar anomaly is detected on a regular basis, then this may suggest that the user is experiencing memory loss. In this case, family or medical services may be contacted, upon appropriate authorisation.

If necessary, the over is turned off remotely.

While certain embodiments have been described, these embodiments have been presented by way of example only, and are not intended to limit the scope of the invention. Indeed, the novel methods, devices and systems described herein may be embodied in a variety of forms; furthermore, various omissions, substitutions and changes in the form of the methods and systems described herein may be made without departing from the spirit of the invention. The accompanying claims and their equivalents are intended to cover such forms or modifications as would fall within the scope and spirit of the inventions.

Claims

Claims: 1. A data analysing unit for analysing electrical power usage data, the unit comprising: means for receiving data from an electrical power monitoring unit said data indicating the electrical power usage of one or more electrical appliances in a dwelhng, means for defining a first electrical signature of energy usage in the dwelling, based on the received data; means for comparing the first electrical signature with an expected electrical signature for the dwelling, means for determining whether an anomaly exists between the first electrical signature and the expected electrical signature, arid for identifying whether a change in the physical well-being of a person resident in the dwelling has occurred, based on the detected anomaly.
2. A data analysing unit according to claim 1, wherein the means for determining whether an anomaly exists between the first electrical signature and the expected electrical signature is configured to identify whether the person resident in the dwelling is unwell.
3. A data analysing unit according to claim 1 or 2, wherein the received data comprises a series of data values indicating the power supplied to each appliance at predetermined intervals in time.
4. A data analysing unit according to claim 3, wherein the expected electrical signature comprises a predetermined sequence of values, the data analysing unit being configured to determine whether any anomalies exist between the first electrical signature and the expected electrical signature by calculating a deviation between the received data and the predetermined sequence of values.
5. A data analysing unit according to claim 4, wherein the data analysing unit is configured to determine that an anomaly is present when the deviation is above a predetermined threshold.
6. A data analysing unit according to any one of the preceding claims, wherein the data analysing unit is configured to identify a physical symptom associated with the anomaly by identifying a plurality of possible symptoms and determining a probability that the identified anomaly relates to each one of those symptoms.
7 A data analysing unit according to claim 6, wherein for each symptom, the data analysing unit is configured to access a predetermined sequence of data values representing an electrical power usage profile associated with that symptom, and to compare the predetermined sequence of data values with the data received from the power monitoring unit.
8. A data analysing unit according to any one of the preceding claims, wherein the data analysing unit is configured to periodically update the expected electrical signature using the data received from the power monitoring unit.
9. A data analysing unit according to claim 8, wherein the data analysing unit is configured to check whether data received from the power monitoring unit comprises an anomaly before the data is used to update the expected electrical signature.
10. A system comprising: a data analysing unit according to any one of the preceding claims, and a response determination unit for determining a response to the anomaly.
11. A system according to claim 10, wherein the response determination unit is configured to determine a response by accessing a look up table that associates different anomalies with appropriate responses.
12. A system according to claim 10 or 11, comprising a response implementation unit, the response implementation unit being configured to implement a determined response by: sending an alert to a person resident in the dwelling or to a third party not SO resident in the dwelling; and I or: adjusting the power supplied to one or more of the electrical appliances.
13. A system comprising: a data analysing unit according to any one of claims Ito 9, and a power monitoring unit for monitoring the electrical power usage of the one or more electrical appliances, the power monitoring unit being configured to transmit data representing the monitored electrical power usage to the data analysing unit.
14. A system according to claim 13, wherein the power monitoring unit is configured to sample the electrical power usage of the one or more electrical appliances at predetermined intervals.
15. A system according to claim 14, wherein the power monitoring unit is configured to transmit the data representing the electrical power usage of the one or more electrical appliances to the data analysing unit at Fntervals larger than those used to sample the electrical power usage of the appliances.
16. A method of detecting changes in the physical well-being of a person resident in a dwelling, the method comprising: receiving data from an electrical power monitoring unit, said data indicating the electrical power usage of one or more electrical appliances in a dwelling, defining a first electrical signature of energy usage in the dwelling, based on the received data; comparing the first electrical signature with an expected electrical signature for the dwelling, determining whether an anomaly exists between the first electrical signature and the expected electrical signature1 and if so, identifying whether a change in the physical well-being of a person resident in the dwelling has occurred, based on the detected anomaly.
17. A method according to claim 16, wherein the step of determining whether an anomaly exists between the first electrical signature and the expected electrical signature comprises identifying whether the person resident in the dwelling is unwell.
18. A method according to claim 16 or 17, wherein the received data comprises a series of data values indicating the power supplied to each appliance at predetermined intervals in time.
19. A method according to claim 18, wherein the expected electrical signature comprises a predetermined sequence of values, and the step of determining whether any anomalies exist between the first electrical signature and the expected electrical signature comprises calculating a deviation between the received data and the predetermined sequence of values.
20. A method according to claim 49, comprising determining whether or not the calculated deviation is above a predetermined threshold.
21. A method according to any one of claims l6to 20, wherein the method comprises idenUfying a physical symptom associated with the detected anomaly by identifying a plurality of possible symptoms and determining a probability that the identified anomaly relates to each one of those symptoms.
22. A method according to claim 211 wherein for each symptom, the data received from the power monitoring unit is compared with a predetermined sequence of data values representing an electrical power usage profile associated with that symptom.
23. A method according to any one claims 16 to 22, wherein the expected electrical signature is periodically updated using the data received from the power monitoring unit.
24. A method according to claim 23, wherein the data received from the power monitoring unit is checked to see if the data comprises an anomaly before the data is used to update the expected electrical signature.
25. A method according to any one of claim 16 to 24 comprising determining a response to the anomaly.
26. A method according to claim 25, wherein the response is determined by accessing a look up table that associates different anomalies with appropriate responses.
27. A method according to claim 26, wherein the response comprises: sending an alert to a person resident in the dwelling or to a third party not resident in the dwelling; and / or adjusting the power supplied to one or more of the electrical appliances.
28. A computer readable storage medium comprising computer executable code that when executed by a computer will cause the computer to carry cut a method according to any one of claims 16 to 27.