GB2465367A

GB2465367A - Non-intrusive electricity metering identifying individual loads from current waveform analysis

Info

Publication number: GB2465367A
Application number: GB0820812A
Authority: GB
Inventors: James Donaldson; Malcolm Mcculloch
Original assignee: Oxford University Innovation Ltd
Current assignee: Oxford University Innovation Ltd
Priority date: 2008-11-13
Filing date: 2008-11-13
Publication date: 2010-05-19
Anticipated expiration: 2028-11-13
Also published as: GB0820812D0; GB2465367B

Abstract

Non-intrusive electricity metering, whereby individual loads on a supply may be identified from analysis of a delta waveform which is indicative of change in the supply current waveform over time. The apparatus has an input section arranged to receive values representative of the total instantaneous supply of electrical current as a function of time from an alternating voltage supply. Current waveforms comprising sets of values representative of the cyclic waveform of the electric current supply are obtained. A delta waveform generator calculates the difference between a current waveform and an earlier current waveform, by subtracting the respective sets of values to obtain a delta waveform. An edge detector is arranged to detect an edge or edges in the delta waveform, and an analysis section is arranged to identify at least one appliance load based at least on information on the edge or edges detected by the edge detector, and to determine the electrical energy consumed by said appliance load.

Description

I

VARIABLE POWER LOAD DETECTOR APPARATUS AND METHOD

The present invention concerns an apparatus and method for detecting a variable power load for use in metering the use of electricity supplied to a plurality of appliances, and in particular determining the electrical power consumed by one or more individual appliances among the plurality of appliances.

There is an increasing concern to reduce the consumption of resources, both at a domestic level in residential buildings, and at a commercial level in offices, shops, factories and so forth. The reasons for this are both to save costs and also because of concerns for the environment, such as to reduce CO2 emissions, and to conserve finite resources such as coal, gas and oil.

Conventionally, consumers receive bills from utility companies that may indicate the quantity of the utility used since the last bill, for example monthly or quarterly, based on periodic meter readings or even based on estimates of consumption since the last meter reading. For example, in the case of electricity supply, the information is presented to the consumer in terms of the number of kilowatt hours of electrical energy that has been used, which is meaningless to many people, and gives very little idea about how they are actually using the energy and where they can cut back. Studies have shown that the effect of providing consumers with real-time detailed information about the energy they are using is that their consumption reduces by up to 20%. Tn order to provide this information, it is necessary to identify where the energy drawn from this supply is ending up, i.e. which appliances are being used, how much and when. It is a problem to provide this information.

Devices are known which can be plugged into a conventional electricity outlet socket that can monitor the energy consumption by a particular appliance (an appliance will also be referred to herein more generally as an electrical load or simply a load') plugged into that socket. However, this information is inconvenient to obtain, and for fully monitoring the consumption at a particular site, such as a house, a separate metering device would have to be plugged into every socket to monitor every appliance, and it is generally not possible to coimect such metering devices to permanently-wired appliances, such as cookers, which are typically some of the largest consumers of energy. Lighting accounts for a significant amount of energy usage in domestic residences, for example on average 20% of the typical electricity bill in the UK is spent on lighting. Much lighting is provided in permanently wired light fittings, so a non-intrusive monitoring system is desired.

Non-intrusive appliance load monitoring systems are known which attempt to detect signatures in the electricity supply that are characteristic of particular appliances, including, for example, monitoring to detect events when appliances are switched on or off. For example, US 4,858,141 (Hart et al.) discloses monitoring the voltage and current of the electricity supply to a residence to try to determine which appliances are running at any particular time and to determine the energy consumed by each.

However, dimming devices (also called dimmer switches) are often fitted to lighting systems to allow variable control of the lighting level. These dimmer switches present a significant challenge to electricity usage monitoring systems because they transform a load that is nominally resistive and of fixed power, to a continuously variable power load, which additionally has a variable reactive power dependent on the level of dimming. There is a problem in providing a reliable way of distinguishing such loads and of measuring the power consumed by this class of device.

The present invention aims to alleviate, at least partially, one or more of the above problems.

Accordingly, the present invention provides a variable power load detector apparatus, for use in a non-intrusive electrical load meter for metering the use of electricity supplied to a plurality of loads, the apparatus comprising: an input section arranged to receive values representative of the total instantaneous supply of electrical current as a function of time from an alternating voltage supply; a monitor section arranged to determine current waveforms comprising sets of values representative of the cyclic waveform of the electric current supply; a delta waveform generator arranged to calculate the difference between a current waveform and an earlier current waveform, by subtracting the respective sets of values determined by the monitor section, to obtain a delta waveform; an edge detector arranged to detect an edge or edges in the delta waveform; and an analysis section arranged to identify at least one load based at least on information on the edge or edges detected by the edge detector.

Another aspect of the present invention provides a method for detecting a variable power load, for use in non-intrusive electrical load metering, for metering the use of electricity supplied to a plurality of loads, the method comprising: receiving values representative of the total instantaneous supply of electrical current as a function of time from an alternating voltage supply; determining current waveforms comprising sets of values representative of the cyclic waveform of the electric current supply; generating a delta waveform by calculating the difference between a current waveform and an earlier current waveform, by subtracting the respective sets of waveform values; detecting an edge or edges in the delta waveform; and identifying at least one load based at least on information on the detected edge or edges.

The present invention further provides a computer program comprising computer-executable code that when executed on a computer system, causes the computer system to perform a method according to the above aspect of the invention.

The invention also provides a computer-readable medium storing a computer program according to the invention above, and a computer program product comprising a signal comprising a computer program according to the invention above.

Embodiments of the invention will now be described, by way of example only, with reference to the accompanying drawings in which: Figure 1 depicts schematically a system using an apparatus for metering the use of electricity according to an embodiment of the invention; Figure 2 is a graph of voltage and current waveforms over one cycle of an alternating supply for a TRIAC-controlled load; Figures 3(a), (b) and (c) show graphs of voltage and current wavefonns for a TRIAC-controlled device, Fig. 3(a) is when the device is turned off, Fig. 3(b) is when the device is turned on at a power level below full power, and Fig. 3(c) shows the change in current waveform between Fig. 3(b) and Fig. 3(a); Figure 4 shows graphs of voltage arid current waveforms for a TRIAC-controlled device, Fig. 4(a) is when the device is at an initial power setting, Fig. 4(b) is when the device is turned up to an increased power level but below full power, and Fig. 4(c) shows the change in current waveform between Fig. 4(b) and Fig. 4(a); Figure 5 shows graphs of voltage and current waveforms for a TRIAC-controlled device, Fig. 5(a) is when the device is at an initial power setting, Fig. 5(b) is when the device is turned up to full power, and Fig. 5(c) shows the change in current wavefonn between Fig. 5(b) and Fig. 5(a); and Figure 6 is a schematic flow chart of a method embodying the invention.

An apparatus according to a first embodiment of the invention will now be described. Figure 1 shows the hardware components of a system incorporating the apparatus for metering the use of electricity, or more correctly for metering electrical energy. The apparatus will be referred to simply as the meter.

In Figure 1, the electricity supply to the site, for example a house, apartment, office, shop, school and so forth is denoted 10. The electricity is supplied to a plurality of appliances 12A, 12B, 12C, 12... by means of conventional wiring 14.

The appliances and wiring are simply shown schematically in Figure 1, but may, of course, be configured in any appropriate way, such as via a consumer unit with circuit breakers or fuses, and with one or more ring main circuits with branches or spurs. A sensor 16 is provided to measure the total instantaneous current being provided to all of the appliances 12 from the supply 10, and also to measure the instantaneous voltage of the electricity supply 10. The current is measured by any suitable sensor, for example a current clamp placed around one of the conductors of the electricity supply wiring 14. The current clamp typically comprises a magnetizable material, such as ferrite, which forms a magnetic circuit around the conductor, and acts as a transformer to induce a voltage in a secondary winding around the magnetizable material, from which the current flowing in the supply wiring 14 can be obtained. As an alternative to this current-transformer, a Hall-effect sensor can be used to measure the magnetic field in the ioop of magnetizable material around the wire which is related to the current flowing through the wire.

Other suitable ways may, of course, be used for sensing the current.

The voltage of the electricity supply can also be measured by any suitable volt meter. This, of course, typically requires access to two of the conductors in the wiring 14. This can be achieved, for example, by probes which strap around the respective cables and have spikes which penetrate the insulation to make contact with the conductor. Alternatively, connections could be made to terminals in the consumer unit, or, for example, at a location where fuses or circuit breakers are insertable. Non-invasive capacitive voltage detectors could also be used.

As shown in Figure 1, the sensor 16 is connected to the meter 20. It is, of course, possible that some or all of the sensor 16 is incorporated within the meter 20, for example that wires connect the supply wiring 14 to the meter 20, and the voltage is measured within the meter 20. Alternatively, in a different embodiment, the sensor 16 may be self-contained and may communicate with the meter wirelessly, sending analogue or digital values of the instantaneous current and instantaneous voltage. In one option, the meter 20 can derive its own power supply by virtue of being connected to the portion of the sensor 16 for measuring voltage. In one particular form of this, the meter 20 is simply plugged into an electrical outlet in the same way as an appliance 12 to obtain its power supply and also to measure the supply voltage. However, in the preferred embodiment, the meter 20 and sensor 16 are conveniently located near where the utility supply 10 enters the building, such as near where the conventional electricity meter is or would be located.

The meter 20 comprises a number of different units. It is possible to implement each of the various units as dedicated hard-wired electronic circuits; however the various units do not have to be separate from each other, and could all be integrated onto a single electronic chip. Furthermore, the units can be embodied as a combination of hardware and software, and the software can be executed by any

S

suitable general-purpose microprocessor, such that in one embodiment the meter 20 could be a conventional personal computer (PC).

The meter 20 comprises an input section 22 that receives current and voltage values from the sensor 16. The values are input or measured preferably multiple times per cycle of the alternating electricity supply to a level of accuracy as required by the application. If the values are supplied as analogue voltages, then the input section 22 may comprise, for example, an analogue to digital converter, such that the rest of the meter 20 can be implemented using digital electronics. The input section 22 also receives time data from a clock 24 which provides the actual present time.

The clock 24 could, of course, be integral with other components of the meter, or could be part of the sensor 16, or could receive a clock signal from an external source such as a transmitter broadcasting time data. In one preferred embodiment the clock 24 comprises a quartz oscillator together with other timer circuitry that is an integral part of a processor 26 (described below). In this case, the input section 22 for receiving the time data is also an integral part of the processor 26. The processor performs a number of different functions, as described below that may be referred to by names of items, such as an edge detector and so forth; in the preferred embodiment of the invention, these items are implemented as software modules.

The voltage and current values together with the time data are received by a processor 26. From the raw data, the processor calculates a number of signature values to characterise the present usage. Examples of suitable signature values include, but are not limited to: (a) the total real power consumption; (b) the phase difference (angle) between the current and voltage which depends on the load applied by the various appliances 12 and whether it is purely resistive or also reactive, i.e. containing capacitive or inductive loads such as motors and transformers; (c) the root-mean-squared (RMS) current.

Clearly some of the signature values mentioned above are averages, typically over a minimum of one cycle of the electricity supply, typically supplied at 50 or 60 hertz so one cycle is approximately 0.02 seconds. However, mean values of all of the various signature values can be calculated over a longer predetermined time interval. The present values of the signature values are compared with the running mean value of each signature value over the previous cycle or cycles to obtain a change or delta' in each signature value.

The processor 26 then uses inference techniques to assign a probability to the state of all of the appliances 12 connected to the supply 10, in terms of whether each appliance is on or off, and the present power consumption by each appliance 12. The inference can assign a probability to the ensemble of appliances being in any particular state based on the calculated probability that the appliances were in any particular state during the previous cycle or at the previous calculation, together with the new evidence from the changes in the various signature values calculated as described above, together with appliance data obtained from a store 28 of the meter 20. The meter 20 is not limited to knowing in advance which appliances 12 are connected to the supply. If a new appliance is added, inferences can be made regarding what that appliance is based on stored characteristics of various classes of appliance.

The appliance data stored in the store 28 can include information such as, but not limited to, statistical information on the probability of a specific appliance consuming a particular amount of power, information on the time of day, duration of use and interval between use of electrical energy by particular appliances, information on likely groupings of devices with increased probability of simultaneous operation, and information on the likelihood of usage and variation in energy consumption of appliances as a function of ambient temperature (where ambient temperature is included as another parameter fed to the processor).

Referring again to Fig. 1, the store 28 in this embodiment may be any suitable computer-readable storage medium, such as a solid-state computer memory, a hard drive, or a removable disc-shaped medium in which information is stored magnetically, optically or magneto-optically. The store 28, may even be remote from the meter and accessible, for example, via a telephone line or over the internet.

The store 28 may be dynamically updateable, for example by downloading new appliance data. This could be done via the supply wiring 14 itself or, in one optional version, the store 28 is provided as an IC-card insertable by the user into a slot in the meter 20. Manufacturers of electrical appliances provide the necessary appliance data either directly to the consumer, or to the utility company. New IC-cards can be mailed to the user to update their meter 26. The software that the processor 26 runs to perform the analysis may also be stored in the store 28 and updated as desired in the same ways as the appliance data (e.g. by downloading, by inserting a new medium such as a disc or IC-card, and so on) Suitable inference techniques to perform the analysis include, for example, probabilistic methods such as Bayesian inference, classifiers such as neural networks, and possibilistic methods such as fuzzy logic. Other suitable methods niay of course be used.

Naturally, the state of the appliances with the highest probability is assunied to be the correct present state of all of the appliances 12. A confidence-limit can also be assigned to the present state. If a new appliance 12 is connected about which the store 28 does not have information, then this will be picked up as a low confidence, in which case the meter can enter a learning mode to obtain information about the power characteristics of the new appliance, either autonomously, or by prompting the user to input new appliance information.

Using the basic signature value information from the electricity supply signals together with inference techniques can successftully discriminate between a large number of different appliances 12. Embodiments of the present invention are particularly concerned with detecting variable power predominantly resistive loads, such as TRIAC-controlled lighting, determining the energy consumption by such loads, and tracking separately each such load when more than one is present. The following description uses the particular example of a dinimer switch controlling an incandescent light. It is, of course, understood that in this context "resistive" refers to the voltage and current flowing through the load being substantially in phase with each other; the load need not necessarily be ohmic nor linear. Similarly, the invention preferably applies to detecting devices employing intra-cycle switching to variably control the power supplied to a load. The TRIAC is just one specific example of a controlled switch for such devices; other examples include: SCRs (silicon-controlled rectifiers), thyristors and transistors.

Background to the operation of the invention.

A modern dinimer switch uses a TRIAC semiconductor device. This is a non-linear device that is oniy turned on for a portion of the electrical cycle. Figure 2 shows the voltage and current waveforms for an idealised dimmer switch controlled incandescent light running at just over half power. The voltage waveform in Figure 2 and all subsequent Figures is the sinusoidal waveform and is shown for one cycle of the alternating electricity supply. For the early part of the cycle, no current is drawn, then at a particular point the TRIAC is triggered and starts conducting such that there is a step change in current flow. The current then flows (approximately proportional to the voltage) for the remainder of the half cycle until the voltage changes polarity (at a zero-crossing of the voltage waveform) at which the TRIAC stops conducting. The second half cycle is then the same as the first half cycle, just with the opposite polarity. When the TRIAC is not conducting, no voltage is applied across the load itself; the voltage waveform shown is that from the supply which is applied across the TRIAC circuit driving the load.

The point at which the TRIAC turns on can be continuously varied, typically by adjusting a variable resistor associated with the TRIAC circuit, generally from anywhere from the beginning to the end of the half cycle. The point at which the conduction begins will be referred to as a phase angle in radians in terms of the cycle of the alternating supply, and is also called the "firing angle". The firing angle can be anything from 0 to ir and in Figure 2 it is somewhere between ir/4 and ir/2. By varying the firing angle, the power consumption can be varied from substantially zero to substantially 100 % of the nominal power rating of the lighting load.

Broadly, there are six scenarios that are of interest with regards the state changes of dimmer controlled lighting systems. These are:- 1. From off to dimmer setting 2. Increase in power to higher brightness 3. From dimmer setting to fully on 4. From fully on to dimmer setting 5. Decrease in power to lower brightness 6. From dimmer setting to off The trivial case of off to fully on (and fully on to off) is omitted because this is already covered by methods concerned with a purely resistive load.

Waveforms for the first three scenarios are discussed below with reference to Figs. 3, 4 and 5. The second three scenarios are identical to the first three except that, the delta waveforms are inverted. In each of Figures 3, 4 and 5, the first and second Figures (a) and (b) show the waveforms before and after, respectively, the change with which that particular scenario is concerned. The third graph (c) in each figure is the delta waveform of the current obtained by subtracting the current waveform (a) from the current waveform in (b); the sinusoidal voltage waveform is shown superimposed for reference. Of course, the waveforms shown in (a) and (b) for each figure are idealised, and represent the current for a single TRIAC-controlled load, In practice, many other appliances will be operating with a significant baseload, so the waveforms will be much more complicated, however, by subtracting to obtain the delta waveform, the baseload is removed, and the current change due solely to the TRIAC-controlled device is obtained.

The following description also makes use of the gradient of the delta waveform and denotes this simply as "dldt".

In embodiments of the invention it is necessary to detect the sharp turn on' and turn off' edges of the waveform and the deltas. The methods of detecting these will be covered in a later section. Note that in the following text, when we refer to edges, these are the edges in the first half of the cycle. For every edge in the first cycle, there will be a corresponding edge in the second half of the cycle which is of opposite polarity.

1) A TRIAC type device turns on -Fig. 3.

In this case, the delta waveform Fig. 3(c) shows a single edge with dldt> 0.

This is at firing angle cz The change in real power is positive.

2) A TRIAC type device increases in power consumption -Fig 4.

In this case, the delta waveform Fig. 4(c) shows two edges. The first has dldt > 0 and is at. The second has dldt < 0. Note that the position of this edge is identical to the %ri' from section one.

The change in real power is positive.

3) A TRIAC type device increases in power consumption to fully on -Fig. 5.

In this case, the delta waveform shows a single edge with dldt < 0. The change in real power is positive.

S

Table 1: Summary of the Scenarios ___________ ___________ ___________ Scenario Waveform Delta First Edge Second Edge Change in _________________ _____________________ d/dt d/dt Real Power 1. Off to dimmer >0 N/A +ve 2. Dimmer to off <0 N/A -ye \\ 3. Dimmer increase >0 <0 +ve o o4 oa o eol 04 0014 ooie 40, 4. Dimmer decrease <0 >0 -ye 5. Dimmer to fully <0 N/A +ve °n / L N / OW. 0 OW. 001 0042 0010 0410 4049 442' 6. Fully on to >0 N/A -ye dimmer Thus it can be seen that by considering the number of edges in the delta waveform, the polarity of those edges and the change in real power associated with the delta, that we can fully detect each scenario.

Identification of the device which has changed power consumption In most domestic residences, there are multiple loads controlled by dimmer switches and each load can have a different full power. For an advanced and accurate NIALM system, it is necessary to not only apportion the change in power to a class of loads lighting' but in actual fact to track the power consumption of individual loads.

Following the detection of a TRIAC event, the first stage is to calculate the full power load of that device. By doing this, we can identify the difference between say a 100W load @ 30% power compared to a 60W load @ 50% power.

To do this is non-trivial. One method is to calculate the power consumption of the device by the integral of the power and hence relate the total power consumption to the firing angle of the TRIAC.

P = _i-JV.I.d(wt) = sin(cot).i0 sin(wt)d(at) where a is the turn on point and f is the turn off point.

However, this may be wildly inaccurate in some circumstances on account of the non-linearity of the load with firing angle.

Instead, a preferred way is to calculate the effective voltage' that the load sees.

The effective RMS voltage as seen by the TRIAC-controlled load cart be calculated as follows (VsE)2 =ijvo2sin2(t)d(wt) 2Jta Which leads to a solution...

(vsE)2 =-{r-a+0.5sin(2a) } (1) when 3 ir this corresponds to the case when we turn a TRIAC on. (i.e. it was previously off.) v0 is the peak of the voltage supply (equal to the actual RMS line voltage x Th.

The power consumed by a perfectly resistive load is proportional to the square of the nns voltage -the constant of proportionality being 1/resistance. In actual fact, in the case of an incandescent bulb, the power is proportional to the rmsVoltage to the power approximately 1.5.

Knowing this relationship, it is thus possible, given the power of a light and the effective RMS voltage applied to calculate the power that would be consumed at the nominal line voltage (240V RMS in the UK) according to the following formula... TI 7

p rRMS.LJNE 2 Norm -Observed >< TI r r RMS-E where y is approximately equal to 1.5 and VRMS.UNE is the RMS line voltage of the supply (nominally 240V in the UK.) Thus, when a TRIAC turns on (scenario 1 above), we know the peak voltage, the firing angle (a) and PObserverd (which is the power delta) and thus can work out the nominal power of the load.

This is useful in the two cases where we turn the TRIAC on and off, but we can generalise further....

Consider that we turn on a TRIAC controlled load, with firing angle a and power change i.Pi. Using this information, we can calculate the effective RMS voltage and thus the nominal power of the load.

Rearranging equation (2) above leads to A1 = VaiT X Norm (3) where V is the effective RMS voltage delivered when the firing angle is al. We now increase the power to the load by decreasing the TRTAC's firing angle. The observed power change is AP2, the firing angle is a2 and the total power being delivered to the load is (AP2 + P1) We can calculate the effective RMS voltage as seen by the load and this is denoted V. Substituting into equation (2) above gives Noym = (Af + LF) (4) Substituting in equation (4) above and rearranging gives the expression Norm AF x VRMSLIWE (5) Thus for any change in state of the TRIAC, we can calculate the nominal power of the load and thus identify the load being controlled. Corresponding expressions can be derived for loads controlled by SCRs or other types of controlled switches.

(Note, when turning a TRIAC on, AP2 is the observed power change, and Vai is zero.) In addition to calculating a value for PN0, equation 5 can also be used to derive the value of y. Following the change in state of a triac-controlled load the values of al and a2 and AP2 are known. These can then be stored awaiting a further change in state of this load. Following a further change in state of this load, one can solve for y since PNorni will be the same in both cases. Thus y is the only unknown and can be solved by conventional mathematical techniques. It is of course possible to calculate y from multiple data points to further increase accuracy. One can simultaneously solve for PNo using equation (5) and thus use PNo to match up state changes to the same appliance. In the event that PN0 is unknown, one can instead match up unknown events to an appliance based on prior knowledge of the state of that appliance, for example from the firing angle.

Additionally, in certain cases where there are multiple loads under TRIAC control, one can further aid identification by considering the positions of the edges. A summary of methods for identifying the specific load following a TRIAC event are shown below in Table 2.

Table 2

Scenario Waveform Delta Methods for identification of device 1. Off to Calculate the nominal power of the load based on dimmer firing angle. The load must have been previously off.

2. Dimmer to Calculate the nominal power of the load based on off i/ \\ J firing angle. In the case that there are multiple loads ,4 of this power on, can identify specific load by L / matching the position of the edge with the last known * firing angle.

3. Dimmer Calculate the nominal power of the load based on increase / firing angle. In the case that there are multiple loads 1' of this power on, can identify specific load by matching the position of the *-ve edge with the last -. known firing angles. The new firing angle is given by ____________ ________________ the position of the +ve edge.

4. Dimmer Calculate the nominal power of the load based on decrease / \ firing angle. In the case that there are multiple loads 1 I of this power on, can identify specific load by : L-' \\// matching the position of the -ye edge with the last known firing angles. The new firing angle is given by ____________ ________________ the position of the +ve edge.

5. Dimmer to T7 1 Calculate the nominal power of the load based fully on r on firing angle, In the case that there are multiple \ TA loads of this power on, can identify specific load by _____________ matching the position of the -ye edge with the last known firing angles. The new firing angle is given by ___________ ________________ the position of the +ve edge.

6. Fully on to, Calculate the nominal power of the load based on the dimmer V \/ firing angle. In the case that there are multiple loads \ of this power, one can restrict the search to devices J\7 which are known to be fully on.

Edge detection As is apparent, it is necessary to accurately detect the edges in the signal.

There is a vast body of literature available with regards edge detection algorithms, though accuracy can be improved in this system through three mechanisms...

1: By employing a sensor / pre-processing stage that removes DC offset from the current signal, we can be assured that prior to a positive edge, the average value of the current delta must be zero, or close to zero such that the magnitude of the difference from zero is below a threshold value (i.e. approximately zero). Similarly, following a negative edge, the average value of the current delta must be approximately zero. These statements apply when either: (i) the edge is in the first half of the waveform and the change in real power is positive; or (ii) the edge is in the second half of the waveform and the change in real power is negative. For the cases of (iii) an edge in the first half of the waveform and a negative change in real power or (iv) the edge in the second half of the waveform and a positive change in real power, then a positive edge should be followed by an approximately zero value of the current delta, and a negative edge should be preceded by an approximately zero value of the current delta. Thus, by measuring the current delta value either side of the edge, one can improve the accuracy of detected edges and thus reject noise and other signal artefacts that would otherwise be mis-identified as a TRIAC turning on.

2: One can assume that as long as the total power measured by the monitor is stable, then the position of the edges are constant. Edges due to noise etc. will not be constant assuming random noise patterns and hence once can average the results over multiple cycles to improve accuracy, either by taking an average of the current delta waveform, or alternatively by running the edge detection algorithm and averaging the results.

3: Assuming that there is no DC offset, the system should exhibit symmetry -for every edge al measured in the first half of the cycle (0 <al < 7r), there will be another edge aQ in the second half of the cycle a2, where ir < c2 <2ir and c2 -ir = al. In reality, it is possible that c2 -r does not exactly equal c1 due to imperfections in the system and the devices, hence, the algorithm should be tolerant to small deviations.

Operation of an Embodiment of the Invention The following processes, described with reference to the flow chart of Fig. 6, are carried out by one example of an apparatus embodying the invention. The processes may be performed by the general processor 26 for example as software modules, or may be implemented in hard-wired dedicated hardware.

Measure signature values of interest (based on instantaneous current and voltage values received at an input section, step Si 0) at a pre-determined rate (at a rate of every cycle, or slower. One could also average over multiple cycles. To date, Real Power has proved to be the most reliable signature, but there are others) to monitor whether the background load is stable' (i.e. inter-cycle variation in the measured signatures is below some pre-determined power). If the signature is deemed to be stable, then it can be assumed that there has been no change in the power signatures as drawn by all appliances on the supply and this stable signature is recorded along with the current waveform.

If there has been a change in signature(s), then we assume that an appliance has changed the amount of power that it is drawing. A change in signature (such as the amount of power) is detected by an event detector in step S20. We may then run multiple analyses designed to detect specific appliance classes and compare the results from each classifier to identify which appliance has changed state. The following describes, by way of example, a classifier to detect variable power loads, such as lighting circuits.

A monitor section determines current waveforms in step S30. The stable' current waveform can be a single waveform preceding the detected event or can be a weighted mean of preceding waveforms. The weighted mean can be a simple average (i.e. all weights equal 1) or can place greater weight for example on more recent samples. The waveform after the detected event can also either be a single waveform or a weighted average. In step S40 a delta waveform generator then calculates the current delta' waveform by subtracting each sample of the stable' current waveform from the present current waveform (after the detected event).

An edge detector then analyses the delta waveform to look for edges in step S50. A simple method would be to threshold on dldt. A more advanced method looks for areas of local maximum in dldt (i.e. the differential at a sample is greater than the samples either side) or by looking at zero crossings in the second derivative. For more details, reference Edge Detection Techniques -An Overview.' By Ziou and Tabbone. To improve the detection of edges, the waveform may first be filtered to remove noise.

If one or more edges are detected, two further checks can be made.

1: If dldt is positive, then the delta current level between the zero crossing preceding the edge of the waveform and the edge should be approximately zero. This can be calculated by numerous methods:-e.g. one could check that the magnitude of each sample is below the maximum noise level of the system. Alternatively, one could average or integrate the current prior to the edge and check that this is below the expected noise level.

Similarly, if d/dt is negative, then the current following the edge should be approximately zero.

As explained previously, and as is apparent from Table 1, tl1ese statements apply to edges in the first half of the waveform for situations in which the change in real power is positive (scenarios 1, 3 and 5). The polarity of the edges should be reversed for scenarios 2, 4, 6, and reversed (again) for edges in the second half of the waveform.

2: To improve accuracy, one can look over multiple cycles. As long as the signal is stable, then the edges should remain in the same position from cycle to cycle. Thus, one can remove false edge detections by looking over multiple cycles.

Once it has been confirmed that one or more edges has been detected, an analysis section then consults table 1 to work out which one of the six scenarios is occuring, based on the number of edges, the order of the edges and the real power delta (change in real power).

Finally, in step S60, the analysis section can now identify the specific load that has changed state.

One can calculate the effective RMS voltage based on the two firing angles, using equation (1) above. Alternatively, one could use a look up table if so desired to ease computation at the expense of memory. Finally, by substituting the effective RMS into equation (5), one can calculate the nominal full power of the load.

Knowing the full power of the load allows us to identify the specific appliance I class of appliance that has changed state -e.g. 100W light bulb on dimmer. Secondly, one can iterate through each appliance of that class that is currently known (in the data store 28) to establish which of those could have changed to the new measured state, based on it's current state. For example, suppose that we have identified that we are in scenario 3 -increase in power.' That means that the only appliances that can have changed state are those that are currently in a dimmer mode. Finally, one can match up the firing angles to identify the specific appliance -e.g. if in scenario 3, then the -ye dldt edge on the new waveform must match up with the current +ve d/dt edge of the appliance which has changed state. Further information on disambiguating identified appliances is given in Table 2.

It is likely that having iterated through the algorithm, there may be more than one contender, each with a measured likelihood,' which may be a probability, or may be a possibility measure. These may be then combined with the scores from other classifiers using a master classifier, which may be (but not exclusively) for example a Bayesian engine, or a Neural Net. If no match with a known appliance is found, then a new appliance entry can be made in the data store 28 for future use.

Following the analysis, in this example, the processor produces a log of the electrical energy utilisation for each appliance (step S70 of Fig. 6), comprising total energy consumption, time of day and duration of each usage. This information is output by an output section 40 to a user terminal 42 (such as a PC or a dedicated device for utility-use feedback) so that the information can be conveniently presented to the user. The output section 40 in the preferred embodiment communicates wirelessly, for example by radio frequencies (RF) link, or optically, or by infrared, or acoustically. However, it is also possible that the communication with the user terminal 42 is done through the supply wiring 14 if the user terminal 42 is plugged into one of the supply outlets as an appliance. In a further embodiment, the output section 40 can also act as a receiver, such that communication between the meter 20 and user terminal 42 is two-way. This enables the user terminal 42 to be used as a further means for updating the appliance data in the store 28.

The user terminal 42 can be a standard desktop or laptop computer with an attached monitor and/or printer, or can be a dedicated device. Although the meter 20 and the user terminal 42 are shown as separate devices in Figure 1, they could, of course, be part of the same device.

The first stage in using the meter is the analysis stage as already described to identify which appliances are being used at any particular time and how much of the or each particular utility they are consuming. The second stage is to provide the user with short-term feedback via the user terminal 42. For example, if the user terminal is a dedicated device in a prominent place in the house, it could give immediate feedback, for example that a particular appliance was left on overnight when that is not usual. It could also highlight changes in the behaviour of appliances, for example if an electric water heater were running more frequently than usual, then the thermostat might be faulty, or if the energy consumption by a refrigerator or any other appliance showed an increase above an expected level, then the user terminal could suggest that the appliance needs servicing.

A further use of the apparatus is to change the way billing is done, by acting as a "smart meter". The data from the meter 20 can be transmitted automatically to a central unit via radio frequency/mobile links which would eliminate the necessity for manual reading of a meter and would also eliminate estimation of meter readings.

Billings and hence feedback can be carried out more frequently which also has a positive impact on reducing the quantity of energy being consumed.

A third stage in the use of the apparatus is long-term feedback. For example, the user can perform trend analysis with the user terminal 42, particularly if it is a personal computer. The user can assess what behavioural changes have made the greatest impact on reduced consumption; the user can compare his energy usage profile with other users of similar sized properties, and communities of users can engage in interactive activities, such as exchanging tips on reducing usage and also in introducing a competitive element to achieve the greatest reductions.

In the embodiments of the invention described above, only electrical energy is measured and discussed. However, the meter could be concerned with two or more utilities, for example additionally measuring water and/or gas consumption to improve inference of which appliances are in use at a particular time; in general the meter may aggregate information about multiple utilities to improve confidence in the inferred usage (for example by particular appliances) of each one of the utilities.

Claims

CLAIMS1. A variable power load detector apparatus, for use in a non-intrusive electrical load meter for metering the use of electricity supplied to a plurality of loads, the apparatus comprising: an input section arranged to receive values representative of the total instantaneous supply of electrical current as a function of time from an alternating voltage supply; a monitor section arranged to determine current waveforms comprising sets of values representative of the cyclic waveform of the electric current supply; a delta waveform generator arranged to calculate the difference between a current waveform and an earlier current waveform, by subtracting the respective sets of values determined by the monitor section, to obtain a delta waveform; an edge detector arranged to detect an edge or edges in the delta waveform; and an analysis section arranged to identify at least one load based at least on information on the edge or edges detected by the edge detector.
2. Apparatus according to claim 1, further comprising an event detector arranged to detect an event representing a change in the total electrical energy being supplied per cycle; and wherein the delta waveform generator is arranged to calculate the difference between the current waveforms before and after the detected event.
3. Apparatus according to claim 1 or 2, wherein the edge detector is further arranged to determine information on the gradient of at least one of the edges.
4. Apparatus according to claim 1, 2 or 3, wherein the edge detector is further arranged to determine information on the position of any edges in the delta waveform.
5. Apparatus according to claim 4, wherein the analysis section is further arranged to determine the nominal full power of the at least one load based on the position of at least one detected edge.
6. Apparatus according to any one of the preceding claims, wherein the edge detector is further arranged to determine information on the number of edges in the delta waveform.
7. Apparatus according to any one of the preceding claims, wherein the analysis section is further arranged to identify at least one load based on the presently known powers of loads known to the apparatus.
8. Apparatus according to any one of the preceding claims, wherein the monitor section is arranged to determine current waveforms for the whole or half of a cycle of the alternating electricity supply.
9. Apparatus according to any one of the preceding claims, wherein at least one current waveform determined by the monitor section is a weighted mean over a plurality of cycles of the alternating electricity supply.
10. Apparatus according to any one of the preceding claims, wherein at least one load has its power varied by a controlled switch; preferably said controlled switch comprises a TRIAC, an SCR or a thyristor.
11. Apparatus according to any one of the preceding claims, wherein the analysis section is arranged to determine the electrical energy consumed individually by each load.
12. Method for detecting a variable power load, for use in non-intrusive electrical load metering, for metering the use of electricity supplied to a plurality of loads, the method comprising: receiving values representative of the total instantaneous supply of electrical current as a function of time from an alternating voltage supply; determining current waveforms comprising sets of values representative of the cyclic waveform of the electric current supply; generating a delta waveform by calculating the difference between a current waveform and an earlier current waveform, by subtracting the respective sets of waveform values; detecting an edge or edges in the delta waveform; and identifying at least one load based at least on information on the detected edge or edges.
13. Method according to claim 12, further comprising detecting an event representing a change in the total electrical energy being supplied per cycle; and wherein the delta waveform is calculated as the difference between the current waveforms before and after the detected event.
14. Method according to claim 12 or 13, further comprising determining information on the gradient of at least one of the edges.
15. Method according to claim 12, 13 or 14, further comprising determining information on the position of any edges in the delta waveform.
16. Method according to claim 15, further comprising determining the nominal full power of the at least one load based on the position of at least one detected edge.
17. Method according to any one of claims 12 to 16, further comprising determining information on the number of edges in the delta waveform.
18. Method according to any one of claims 12 to 17, further comprising identifying at least one load based on the presently known powers of loads being supplied with electricity.
19. Method according to any one of claims 12 to 18, wherein the current waveforms are determined for the whole or half of a cycle of the alternating electricity supply.
20. Method according to any one of claims 12 to 19, wherein at least one current waveform is an average over a plurality of cycles of the alternating electricity supply.
21. Method according to any one of claims 12 to 20, wherein at least one load has its power varied by a controlled switch; preferably said controlled switch comprises a TRIAC, an SCR or a thyristor.
22. Method according to any one of claims 12 to 21, further comprising determining the electrical energy individually consumed by each load.
23. A computer program comprising computer-executable code that when executed on a computer system, causes the computer system to perform a method according to any one of claims 12 to 22.
24. A computer-readable medium storing a computer program according to claim 23.
25. A computer program product comprising a signal comprising a computer program according to claim 23.AMENDMENTS TO CLAIMS HAVE BEEN FILED AS FOLLOWSCLAIMS1. A variable power load detector apparatus, for use in a non-intrusive electrical load meter for metering the use of electricity supplied to a plurality of loads, the electricity supply providing an alternating voltage supply and an electrical current supply, the apparatus comprising: an input section arranged to receive values representative of the total instantaneous supply of electrical current as a function of time from the alternating voltage supply; a monitor section arranged to receive the values from the input section and to determine a current waveform comprising a set of values representative of a cyclic waveform of the electrical current supply; a delta waveform generator arranged to calculate a difference between a current waveform and an earlier current waveform, by subtracting the respective sets *: *.: 15 of values determined by the monitor section, to obtain a delta waveform; an edge detector arranged to detect an edge or edges in the delta waveform; *. .: and : an analysis section arranged to identify at least one load based at least on * * ** information on the edge or edges detected by the edge detector. ***. * 20*..*.* * 2. Apparatus according to claim 1, further comprising an event detector arranged to detect an event representing a change in the total electrical energy being supplied per cycle; and wherein the delta waveform generator is arranged to calculate the difference between the current waveforms before and after the detected event.3. Apparatus according to claim 1 or 2, wherein the edge detector is further arranged to determine information on the gradient of at least one of the edges.4. Apparatus according to claim 1, 2 or 3, wherein the edge detector is further arranged to determine information on the position of any edges in the delta waveform.5. Apparatus according to claim 4, wherein the analysis section is further arranged to determine the nominal full power of the at least one load based on the position of at least one detected edge.6. Apparatus according to any one of the preceding claims, wherein the edge detector is further arranged to determine information on the number of edges in the delta waveform.7. Apparatus according to any one of the preceding claims, wherein the analysis section is further arranged to identifSl at least one load based on the presently known powers of loads known to the apparatus.8. Apparatus according to any one of the preceding claims, wherein the *a*.monitor section is arranged to determine current waveforms for the whole or half of a * .* * S. * cycle of the alternating electricity supply.9. Apparatus according to any one of the preceding claims, wherein at least one current waveform determined by the monitor section is a weighted mean . 20 over a plurality of cycles of the alternating electricity supply.*..*.S * . 10. Apparatus according to any one of the preceding claims, wherein at least one load has its power varied by a controlled switch; preferably said controlled switch comprises a TRIAC, an SCR or a thyristor.11. Apparatus according to any one of the preceding claims, wherein the analysis section is arranged to determine the electrical energy consumed individually by each load.12. Method for detecting a variable power load, for use in non-intrusive electrical load metering, for metering the use of electricity supplied to a plurality of loads, the electricity supply providing an alternating voltage supply and an electrical current supply, the method comprising: receiving values representative of the total instantaneous supply of electrical current as a function of time from the alternating voltage supply; determining from the received values a current waveform comprising a set of waveform values representative of a cyclic waveform of the electrical current supply; generating a delta waveform by calculating a difference between a current waveform and an earlier current waveform, by subtracting the respective sets of waveform values; detecting an edge or edges in the delta waveform; and identifying at least one load based at least on information on the detected edge or edges.13. Method according to claim 12, further comprising detecting an event *.: 15 representing a change in the total electrical energy being supplied per cycle; and wherein the delta waveform is calculated as the difference between the current ** * waveforms before and after the detected event.S14. Method according to claim 12 or 13, further comprising determining information on the gradient of at least one of the edges.S.. S** * . 15. Method according to claim 12, 13 or 14, further comprising determining information on the position of any edges in the delta waveform.16. Method according to claim 15, further comprising determining the nominal full power of the at least one load based on the position of at least one detected edge.17. Method according to any one of claims 12 to 16, further comprising determining information on the number of edges in the delta waveform.loads, the electricity supply providing an alternating voltage supply and an electrical current supply, the apparatus comprising: an input section arranged to receive values representative of the total instantaneous supply of electrical current andlor power as a function of time from the alternating voltage supply, the received values comprising a set of waveform values representative of a cyclic waveform of the electrical current and/or power supply; a delta waveform generator arranged to calculate a difference between a cyclic waveform and an earlier cyclic waveform, by subtracting the respective sets of waveform values, to obtain a delta waveform; an edge detector arranged to detect an edge or edges in the delta waveform; and an analysis section arranged to identify at least one load based at least on information on the edge or edges detected by the edge detector. S... * .* ..: 15 27. Method for detecting a variable power load, for use in non-intrusive electrical load metering, for metering the use of electricity supplied to a plurality of * loads, the electricity supply providing an alternating voltage supply and an electrical current supply, the method comprising: * * *. receiving values representative of the total instantaneous supply of electrical *..current and/or power as a function of time from the alternating voltage supply, the S.....* received values comprising a set of waveform values representative of a cyclic waveform of the electrical current and/or power supply; generating a delta waveform by calculating a difference between a cyclic waveform and an earlier cyclic waveform, by subtracting the respective sets of waveform values; detecting an edge or edges in the delta waveform; and identif'ing at least one load based at least on information on the detected edge or edges.