JP6328036B2 - Weighing meter state change detection device and program - Google Patents

Description

  The present invention relates to an apparatus for detecting a change in state such as replacement or disconnection of an energy meter represented by electricity, gas or water.

  Analog metering meters are widely used for metering the usage of electricity, gas, and water. The metering meter can output an electric signal of one pulse per unit usage amount, and a building management system or the like receives the pulse signal. Thus, the meter value can be monitored by a building management system or the like without visually checking the value (hereinafter referred to as the meter value) indicated by the meter at the place where the metering meter is installed. The number of pulses per unit usage is defined as a pulse unit for each meter, for example, “pulse / kWh” in the case of electricity.

  As a prior example relating to the present invention, in Patent Document 1, a gas appliance connected to a metering meter is estimated from the flow rate of the metered gas.

  Patent Document 2 discloses a method in which a facility diagnosis apparatus detects a change in facility.

  Patent Documents 3 and 4 describe a method for detecting a target signal from past signals.

  Patent Document 5 discloses a method for detecting an abnormality from data.

JP 2003-149027 A JP 2014-49067 A JP 2008-145505 A International Publication 2006/009035 International Publication No. 2010/041447

  When the connection line connecting the meter and the monitoring device such as the building management system is disconnected or has poor contact due to deterioration over time, the pulse signal transmitted from the meter to the building management system is lost, or the number of pulse signals decrease. In this case, it becomes impossible to monitor the correct meter value in the building management system. As a result, it becomes difficult to correctly grasp the usage of electricity, gas, and water, and it becomes difficult to plan suitable energy saving and resource saving measures. In addition, if the tenant in the building is charged for electricity, gas, water, etc. based on the meter value monitored by the building management system, the wrong usage fee shall be charged. Could lead to Therefore, it is necessary to detect and repair defects such as disconnection at an early stage.

  Further, when the meter is replaced with another meter having a different pulse unit, the number of pulse signals transmitted from the meter to the building management system changes for the same amount of electricity, gas, and water usage. In this case, if the setting of the pulse unit on the building management system side is not corrected, it is impossible to monitor the correct meter value in the building management system. As a result, it becomes difficult to correctly grasp the usage of electricity, gas, and water, and it becomes difficult to plan suitable energy saving and resource saving measures. In addition, if the tenant in the building is charged for electricity, gas, water, etc. based on the meter value monitored by the building management system, the wrong usage fee will be charged. It can be connected. Therefore, it is necessary to detect the replacement of the metering meter at an early stage and to correct the setting of the pulse unit or pulse constant (kWh / pulse) of the building management system at an early stage.

  However, the number of pulse signals of the metering meter originally depends on the operating conditions of facilities using electricity, gas, water, and the like. For example, the number of pulse signals transmitted from the metering meter to the building management system changes due to the facility being stopped or the facility being increased. Therefore, even if the number of pulse signals increases or decreases, or the usage amount of electricity, gas, and water increases, it has been difficult for the building management system to detect disconnection or change of the meter.

  In patent document 1, the technique which estimates the gas appliance which uses gas from the flow volume of gas is shown. However, it is a technique on the premise that the gas flow rate is correctly measured, and the disconnection or change of the metering meter cannot be detected by the technique of Patent Document 1.

  Patent Document 2 discloses a technique in which an equipment diagnosis apparatus detects a change in equipment. However, in the first embodiment, since the equipment change is detected by comparing the equipment information of the equipment diagnosis device and the equipment management device, if the equipment management device does not detect the equipment change, The diagnostic device cannot detect a change in equipment. It does not show how the equipment management device detects equipment changes. Further, in the second embodiment, a method of detecting a change in equipment when “power amount <rated power amount−threshold” is satisfied, but the rated power amount differs for each facility. For this reason, it is necessary to set the rated power amount for each facility that uses the electricity metered by the metering meter, and the work load is large. In addition, as described above, the amount of power used varies depending on the operation status, and therefore, there is a possibility that a change in equipment is erroneously detected due to a temporary change in the amount of power depending on the setting of the threshold value.

  Patent Documents 3 and 4 disclose techniques for detecting a signal similar to a target signal from accumulated signals. However, when the signal to be detected (target signal) is indefinite, such as an increase or decrease in the number of pulse signals due to a change in the metering meter, or an increase or decrease in the usage amount of electricity, gas, or water, the technique of Patent Document 3 cannot be applied as it is. If it is detected that the latest signal does not exist in the signal accumulated in the past because the signal to be detected is indefinite, the amount of power used varies depending on the driving situation as described above. , May temporarily indicate a signal that does not exist among previously accumulated signals. Therefore, there is a need for a method for detecting that the signal does not exist in the signal accumulated in the past and is not a temporary change in the meter value due to a change in the driving situation. Not shown.

  Patent Document 5 further includes a reference data generation unit that generates normal data (reference data) in advance and a normal distribution storage unit that spontaneously generates a feature amount, and the feature amount generated by the normal distribution storage unit. Is a technique for detecting an abnormality of input data by comparing each of the reference data and the input data with a probability close to the reference data as a dissimilarity of the input data with respect to the reference data (normal data). It is shown. However, it is necessary to prepare reference data in advance.

  The present invention has been made to solve the above-described problems, and has an object to detect a change in state of a meter meter, a disconnection, or the like from a time-series change in energy consumption without adding a sensor or the like. And

A state change detection device for a meter according to the present invention is a state change detection device for connecting a plurality of meter meters, and is based on a pulse signal transmitted from each of the plurality of meter meters in accordance with the use of energy. A feature amount calculation means for calculating a feature amount obtained by normalizing the amount of increase in usage amount per unit time, a grouping means for classifying the plurality of weighing meters based on a certain rule, and a feature amount when the feature amount is normal. A normal learning means for calculating a threshold value that defines a range of possible values from the standard deviation of the feature quantity for each group of the weighing meters classified by the grouping means, the feature quantity computed by the feature quantity computation means, and the Detecting means for detecting a change in state of the weighing meter from a threshold value calculated by the normal learning means.

  Further, the feature amount calculating means calculates the feature amount from a value after application of the low-pass filter of an increase amount per unit time of energy usage.

  The feature amount calculating means includes a data length learning means for calculating a period of the amount of energy used for calculating the feature amount, and the data length learning means is a standard for the feature amount calculated by the normal learning means. The period is calculated from the deviation.

The program according to the present invention uses a computer mounted on a state change detection device connecting a plurality of metering meters based on a pulse signal transmitted from each of the plurality of metering meters according to the use of energy. A feature amount calculating means for calculating a feature amount obtained by normalizing an increase amount per unit time, a grouping means for classifying the plurality of measurement meters based on a certain rule, and a range of values that the feature amount can take when normal Normal learning means for calculating a threshold value to be defined from the standard deviation of the feature quantity for each group of the weighing meters classified by the grouping means, the feature quantity computed by the feature quantity computation means, and the normal learning means computed This is intended to function as detection means for detecting a change in the state of the metering meter from a threshold value.

  According to the present invention, it is possible to detect a change in the state of the measuring meter such as replacement of the measuring meter or disconnection of the measuring meter without adding a sensor or the like, and an operator does not check the measuring meter. become.

  Further, by calculating the feature amount from the value after the low pass filter is applied to the amount of increase in the amount of energy used per unit time, it is possible to reduce erroneous detection of a change in the state of the meter.

  Further, by providing the data length learning means, it is possible to shorten the detection time of the change in state of the metering meter.

It is a block block diagram of one Embodiment of the state change detection apparatus which concerns on this invention. It is a figure which shows an example of the data structure memorize | stored in the recording means in this Embodiment. It is a figure for demonstrating the operation | movement flow of the state change detection apparatus in this Embodiment. It is a figure for demonstrating operation | movement of the feature-value calculating means in this Embodiment. It is a figure for demonstrating operation | movement of the normal learning means in this Embodiment.

  FIG. 1 is a configuration diagram of a state change detection device for a metering meter according to a first embodiment of the present invention. The state change detection device 1 in the present embodiment is electrically connected to one or more metering meters 2 and receives a pulse signal. The state change detection apparatus 1 includes a communication unit 11, a usage amount calculation unit 12, a grouping unit 13, a feature amount calculation unit 14, a normal learning unit 15, an abnormality detection unit 16, a data length learning unit 17, an output unit 18, and a recording unit. 100. Each component 11-18,100 is electrically connected, respectively. Among these, the constituent elements 11 to 18 are realized by a cooperative operation of a computer such as a microcomputer having a CPU, a ROM, a RAM, and the like, and a program operating on the CPU mounted on the computer. The recording unit 100 is realized by a storage unit such as a RAM or an HDD mounted on a computer, or an external storage unit accessible via a network.

  Further, the program used in the present embodiment can be provided not only via the communication unit 11 but also stored in a computer-readable recording medium such as a CD-ROM or a USB memory. The program provided from the communication means or the recording medium is installed in the computer, and various processes are realized by the CPU of the computer sequentially executing the program.

It is assumed that items as shown in FIG. 2 are recorded in the recording unit 100. The serial number is an identifier of the weighing meter, and is an arbitrary numerical value. The object is a measurement object of the metering meter, and in the case of the present embodiment, energy such as electricity, gas, and water is applicable. Instead of recording “electricity”, “gas”, and “water supply”, it may be recorded by a corresponding symbol such as A or B. Units and is a unit of the measurement object, for example, if the electric kWh, etc. kWs and m ³ If gas, such as m ³ if the water is recorded. The pulse unit represents the number of pulses per unit usage and is recorded as a numerical value. The consumption facility name is the name of the facility that consumes the measurement target of the metering meter, and records air conditioning equipment, lighting, and the like. In addition, it may record with applicable symbols, such as A and B, instead of recording with air-conditioning cold / heat and illumination. The current accumulated value is a value calculated by a usage calculation means 12 described later, and represents the meter value of each metering meter calculated by the state change detection device 1 based on the pulse received from the metering meter. The m ₁ January d ₁ day h ₁ at accumulated value represents the accumulated value of the past m ₁ January d ₁ day h ₁ hour time point. Hereinafter, the same applies to the m ₂ d ₂ day h ₂ o m accumulated value and the m ₃ d d ₃ h h ₃ o m accumulated value.

  The communication means 11 receives a pulse signal from the weighing meter 2. An object to be measured by the meter 2 may be any of electricity, gas, and water, and may be a meter other than electricity, gas, and water as long as it is a meter that transmits a pulse.

  The usage amount calculation unit 12 receives the pulse signal from the communication unit 11 and calculates the usage amount of electricity, gas, water, or the like. For example, when one pulse is received from an electric meter whose pulse unit is 1 pulse / kWh, it is calculated that 1 kWh has been consumed. Further, the usage amount calculation unit 12 calculates the accumulated value of the usage amount of electricity, gas, and water and records it in the recording unit 100. For example, assuming that 10 kWh has been recorded as a cumulative value by receiving 10 pulses in the past from an electric meter with a pulse unit of 1 pulse / kWh, if one pulse is newly received in this state, the usage calculation means 12 The cumulative value is updated to 11 kWh. Further, the usage amount calculation unit 12 may record the time series change of the accumulated value in the recording unit 100. For example, if the cumulative value is recorded every hour and the cumulative value at 12:00 is recorded in the recording means 100, when the current time is 13:00, the usage amount calculating means 12 The current cumulative value is recorded in the recording means 100 as the cumulative value at the time. The current time is obtained from time measuring means (not shown). Here, the recording interval of the time series change is described as one hour, but any interval may be used. It is assumed that the recording interval is held in advance by the usage amount calculation unit 12.

  The grouping means 13 groups the weighing meters according to a certain rule. For example, an electric meter, a gas meter, and a water meter are classified into different groups. Moreover, even if it is an electric meter, the electric meter which measures the electric consumption of air-conditioning cooling / heating equipment, the electric meter which measures the electric consumption of water supply and sewage devices such as pumps, and the electric consumption of lighting equipment are measured. Each meter may be classified according to the type of electricity meter and the type of electricity, gas and water consumption equipment.

  The feature amount calculating unit 14 calculates a feature amount for detecting replacement / disconnection of the metering meter 2 from the time series change of the accumulated value calculated by the usage amount calculating unit 12.

  The normal learning means 15 calculates the normal range indicated by the feature value from the feature value calculated by the feature value calculation means 14.

  The abnormality detecting unit 16 is used for calculating the feature amount by detecting that the latest feature amount calculated by the feature amount calculating unit 14 exceeds the normal range defined by the normal learning unit 15. Detects changes in the status of the weighing meter 2 (disconnection or replacement).

  The data length learning unit 17 updates the period of the time-series cumulative value used by the feature amount calculating unit 14 for calculating the feature amount.

  The output means 18 outputs, when the abnormality detecting means 16 detects a change in the state of the meter, the detection of the change in the state of the meter 2 and the estimated time when the state has changed in a prescribed manner.

  Next, the operation of each means of the weighing meter state change detection device 1 according to the present invention will be described with reference to the flowchart of FIG.

  In step ST <b> 101, the communication unit 11 receives a pulse signal from the weighing meter 2.

  In step ST102, when the usage amount calculation means 12 receives that the communication means 11 has received the pulse signal, the usage amount calculation means 12 acquires the pulse unit and unit of the meter 2 that has transmitted the pulse from the recording means 100, and calculates the usage amount. . For example, when the pulse unit is 10 and the unit is kWh, it is calculated that the electric energy of 10 kWh is used by receiving the pulse. The calculated usage amount is added to the current cumulative value stored in the recording unit 100 to update the current cumulative value.

  In step ST103, the recording unit 100 stores the stored current cumulative value as a past history when the specified time comes. For example, when it is 10:00 on January 1 and the time is a predetermined time, the current accumulated value is separately stored as the accumulated value at 10:00 on January 1.

  In step ST104, the grouping means 13 classifies the weighing meters according to a certain rule. For example, an electric meter, a gas meter, and a water meter are classified into different groups depending on the type of measurement target. Moreover, even if it is an electric meter, the electric meter which measures the electric consumption of air-conditioning cooling / heating equipment, the electric meter which measures the electric consumption of water supply and sewage devices such as pumps, and the electric consumption of lighting equipment are measured. Each meter may be classified according to the type of equipment that consumes electricity, gas, and water. Alternatively, all the metering meters may be in the same group.

  In step ST <b> 105, the feature amount calculation unit 14 calculates a feature amount for detecting a state change from the time series change of the accumulated value stored in the recording unit 100. For example, the ratio of the increase amount of the cumulative value is calculated. As shown in FIG. 4A, let f (i) be the cumulative value at a certain time i. Let f (i−N) be the cumulative value of N (N ≠ 0) past from i. In this case, as shown in FIG. 4B, the increment g (i) of the cumulative value can be calculated as g (i) = f (i) −f (i−N). In this case, as shown in FIG. 4D, the ratio y of the increase amount of the cumulative value can be calculated as y (i) = g (i) / g (i−N).

  When the most recent increase value g (i) is close to the past increase value g (i−N), the ratio y (i) is close to 1. On the other hand, when the most recently accumulated increase amount g (i) and the past accumulated value increase amount g (i−N) are greatly different, the ratio y (i) is a value far from 1. For example, when disconnection occurs, the ratio y (i) becomes zero. In addition, when only one line is disconnected in a three-wire meter, the number of pulses received from the meter is halved, so the ratio y (i) is likely to be a value of 0.5 or less. In many cases, the pulse unit of the weighing meter is defined by an integer multiple, for example, a numerical value such as 1 or 10. For this reason, when the meter is changed to another weighing meter having a different pulse unit, the number of pulses received by the state change detection device 1 becomes an integral multiple, for example, 10 times or more, or 1/10 or less. In this case, the ratio y (i) is likely to be a value far from 1, such as 10 or more, or 0.1 or less. Therefore, when the ratio y (i) shows a value far from 1, it becomes possible to detect a change in the state of the weighing meter.

  The reason similar to the above is that the difference (g (i) −g (i−N)) between the increase amount g (i) of the latest cumulative value and the increase amount g (i−N) of the past cumulative value is large. Therefore, there is a possibility that a change in the state of the weighing meter can be detected. However, the magnitude of the difference (g (i) −g (i−N)) between the increase amount g (i) of the most recent cumulative value and the increase amount g (i−N) of the past cumulative value is It depends on. For example, when the difference (g (i) −g (i−N)) of the increase amount of the accumulated value is 1 in a weighing meter in which the increase amount g (i) of the accumulated value is usually about 100, The difference is not great. However, when the difference (g (i) −g (i−N)) of the increase amount of the cumulative value is usually 1 in the weighing meter in which the increase amount g (i) of the cumulative value is usually about 3, The difference is considered relatively large. Therefore, the state change of the meter is detected by the difference (g (i) −g (i−N)) between the increase amount g (i) of the latest cumulative value and the increase amount g (i−N) of the past cumulative value. In this case, since the determination of the magnitude of the difference value (g (i) −g (i−N)) cannot be handled equally by a plurality of weighing meters, it is difficult to detect the change in the state of the weighing meter. There is.

  As in the present embodiment, calculating the ratio of the increase amount of the accumulated value normalizes the average value of the values used for detecting the change in the state of the meter. By normalizing, it becomes possible to detect a change in the state of a plurality of weighing meters having different increments g (i) of the accumulated value based on the same rule.

として、ｇ（ｉ）から演算できる。したがって、累積値ｆ（ｉ）の増加量ｇ（ｉ）の移動平均値ｈ（ｉ）の比率ｙ（ｉ）は、

として、ｇ（ｉ）から演算できる。 Further, instead of calculating the ratio y (i) using the increase amount g (i) of the cumulative value f (i), the moving average value h (i) of the increase amount g (i) of the cumulative value f (i) is calculated. ) Ratio y (i) may be calculated. The moving average value h (i) in FIG.

Can be calculated from g (i). Therefore, the ratio y (i) of the moving average value h (i) of the increase amount g (i) of the cumulative value f (i) is

Can be calculated from g (i).

  Since electricity, gas, water supply, etc. may temporarily increase or decrease temporarily, the increase in cumulative value (g (i) = f (i) −f (i−N)) is also temporarily May increase or decrease rapidly. As a result, the rate of increase of the cumulative value (y (i) = g (i) / g (i−N)) also temporarily shows a value far from 1, and it is possible to erroneously detect a change in the state of the meter. There is sex. As in the present embodiment, by setting the ratio y (i) of the moving average value h (i) of the increase amount g (i) of the cumulative value f (i), the moving average value functions as a low-pass filter, Even if the usage amount of electricity, gas, water supply, etc. suddenly increases or decreases temporarily, false detection of changes in the state of the meter can be reduced.

  In the first embodiment, the moving average is described as an example of the low-pass filter. However, other low-pass filters such as a triangular moving average may be used.

を満たすとき、計量メータが状態変化（断線、交換）したものとして検出する。 In step ST106, the abnormality detection unit 16 detects a change in the state of the meter using the feature amount calculated by the feature amount calculation unit 14. For example, the ratio y (i) calculated by the feature amount calculation means 14 is compared with a specified threshold value α,

When the condition is satisfied, it is detected that the weighing meter has changed state (disconnected, replaced).

として演算できる。ここでＭとは、ｙ（ｉ）の総サンプル数である。ポアソン分布は、正規分布の優れた近似であることが知られている。正規分布の場合、平均値±σの範囲に全体の約６８％、平均値±２σの範囲に全体の約９５％、平均値±３σの範囲に全体の約９９％が含まれることが知られている。したがって、正常学習手段１５が、閾値α＝３σに更新することで、異常検出手段１６は、通常のポアソン分布の約９９％に収まらない比率ｙ（ｉ）を計量メータの状態変化として、適切に検出することが可能になる。なお、閾値αは、必ずしも３σである必要はなく、σの定数倍であればよい。 In step ST107, the normal learning means 15 updates the threshold value α used by the abnormality detection means 16. As shown in FIG. 5, the ratio y (i) calculated by the feature amount calculation means 14 is Poisson distributed around the average value 1. In this case, the standard deviation σ of the ratio y (i) is

Can be calculated as Here, M is the total number of samples of y (i). The Poisson distribution is known to be an excellent approximation of the normal distribution. In the case of the normal distribution, it is known that the average value ± σ range includes about 68% of the whole, the average value ± 2σ range of about 95% of the whole, and the average value ± 3σ of about 99% of the whole. ing. Therefore, when the normal learning means 15 updates the threshold value α = 3σ, the abnormality detection means 16 appropriately sets the ratio y (i) that does not fall within about 99% of the normal Poisson distribution as a change in the state of the meter. It becomes possible to detect. Note that the threshold α is not necessarily 3σ and may be a constant multiple of σ.

  Further, as the number of samples constituting the Poisson distribution increases, a more accurate standard deviation can be calculated. However, it takes time to store a large number of samples with one weighing meter. If the time for storing the number of samples is long, the change in the state of the meter cannot be detected properly during that time. Therefore, the grouping means 13 may calculate the standard deviation σ using the respective ratios y (i) of the weighing meters in the same group, and update the threshold value α.

  In step ST108, the data length learning unit 17 updates the period of the accumulated value f (i) used by the feature amount calculation unit 14 to calculate the feature amount. The feature amount calculation means 14 uses the latest cumulative value f (i) and the past cumulative value f (i−N) from i to N (N ≠ 0), and uses the ratio y (i) as the feature amount. Calculate. In this case, the feature quantity cannot be calculated unless N accumulated values are accumulated. Further, in order to detect the state change more accurately, the feature amount calculation means 14 calculates the feature amount y (i) from the moving average value h (i) of the increase amount g (i) of the cumulative value f (i). In this case, the feature quantity cannot be calculated unless 2N accumulated values are accumulated. In addition, when the feature value y (i) is calculated from the moving average value h (i), the low-pass filter functions, so y (i) changes gradually in time series. In this case, it takes a maximum of N times from when the state of the weighing meter changes until it is detected. Therefore, in order to shorten the time required to detect the state change, a smaller value of N is preferable. On the other hand, when N is decreased, the amount of data for calculating the moving average value h (i) decreases. , It becomes easy to detect a change in state easily due to a sudden increase / decrease in temporary usage. Therefore, the period of the accumulated value f (i) used for calculating the feature value by the feature value calculating means 14, that is, N is updated to an appropriate value.

For example, in order to prevent a state change from being erroneously detected due to a temporary increase / decrease in the temporary usage amount, N is minimized within a range in which the threshold value α to be compared with the feature amount does not change significantly. The threshold value α is calculated from the standard deviation σ. σ is calculated from the feature value y (i). The feature amount y (i) is calculated from the variable i and the constant N. Here, σ when N = M (M is a constant) and i is fixed at the current time is defined as β. A function of σ when i is fixed at the current time and N is a variable is σ (N). At this time, the absolute value of the difference between β and σ (N) is defined as a function q (N) as in the following equation.

  Since the threshold value α is calculated from the standard deviation σ, minimizing N within a range where the threshold value α to be compared with the feature amount does not change greatly means that the difference q (N) from the standard deviation β is minimized. N to be converted is determined in the range of 1 to M. A known method such as a gradient method is used as a method for obtaining N in a range of 1 to M that minimizes q (N). When the obtained minimum value of q (N) is larger than the prescribed value H (q (N)> H), N = M may be set. As a result, it is possible to reduce the time required to detect the state change while reducing erroneous detection of the state change.

  In step ST109, when the abnormality detection unit 16 detects a change in the state of the weighing meter, the output unit 18 outputs the detection of the change in state by a specified method. For example, when a display device such as a display (not shown) is connected to the state change detection device 1, the state change detection is displayed on the display device. Specifically, the serial number of the weighing meter in which the state change is detected is displayed. In order to notify the manager of the state change detection device 1 that the state change detection device 1 detects the state change and displays the detection result on the display device, a sound / announcement is transmitted from a speaker (not shown). May be. Alternatively, when the state change detection device 1 is connected to the public line network via a line (not shown), the output unit 18 may transmit the state change detection result to a preset mail address.

  Moreover, the output means 18 may output together the time estimated that the state of the metering meter changed. It is assumed that the abnormality detection means 16 can detect the time when it is estimated that the state of the weighing meter has changed. For example, as described above in the description of step ST108, it takes a maximum of N times from when the state changes until detection. Therefore, it is possible to output N times past from the time when the state change is detected as the time when it is estimated that the state has changed.

  When the calculation is completed up to step ST109, the process returns to step ST101, receives the next pulse signal, and thereafter repeats step ST101 to step ST109.

  In the first embodiment, as shown in FIG. 1, all the means 11 to 18 and 100 are described as being in the state change detection device 1, but each means is not necessarily the same device. There is no need to be. For example, the weighing meter 2 may be connected to a building management system and the accumulated value may be managed. Therefore, some or all of the means of the state change detection device 1 may be applied to the building management system. In this case, it is assumed that the state change detection device 1 is electrically connected to the building management system. Alternatively, a device that manages cumulative values, such as a building management system, may be connected to a server at a remote location via a public line network. In this case, since the server is electrically connected to the metering meter and the building management system through the public network, a part or all of each means of the state change detection device 1 is applied to the server. May be. In this case, it is assumed that the server functions as the state change detection device 1 or that the server and the state change detection device 1 are electrically connected.

  In addition, the server can be electrically connected to a building management system of a plurality of buildings through a public line network. In this case, the grouping means 13 can also classify the metering meters of a plurality of buildings according to a predetermined rule. As a result, the normal learning means 15 can prescribe the threshold value α more quickly and appropriately.

  As described above, in the present embodiment, it is possible to detect a change (change / disconnection) in the state of the meter without adding a sensor or the like and without an operator performing confirmation work on the meter. .

  DESCRIPTION OF SYMBOLS 1 State change detection apparatus, 2 Metering meter, 11 Communication means, 12 Usage amount calculation means, 13 Grouping means, 14 Feature-value calculation means, 15 Normal learning means, 16 Abnormality detection means, 17 Data length learning means, 18 Output means 100 Recording means.

Claims (4)

A state change detection device for connecting a plurality of weighing meters,
A feature amount calculating means for calculating a feature amount obtained by normalizing an increase amount per unit time of the amount of energy used based on a pulse signal transmitted from each of the plurality of metering meters according to the use of energy;
Grouping means for classifying the plurality of weighing meters based on a certain rule;
Normal learning means for calculating a threshold value defining a range of values that can be taken when the feature quantity is normal from a standard deviation of the feature quantity for each group of the weighing meters classified by the grouping means;
Detecting means for detecting a change in the state of the meter from the feature quantity calculated by the feature quantity calculating means and the threshold value calculated by the normal learning means;
A state change detection apparatus comprising: The state change detection device according to claim 1,
The state change detecting device, wherein the feature amount calculating means calculates the feature amount from a value after application of a low-pass filter of an increase amount per unit time of energy use amount. The state change detection device according to claim 1 or 2,
A data length learning means for calculating a period of the amount of energy used by the feature quantity computing means for computing the feature quantity;
The state change detection device, wherein the data length learning means calculates the period from a standard deviation of feature quantities calculated by the normal learning means. A computer installed in a state change detection device that connects a plurality of weighing meters ,
A feature amount calculating means for calculating a feature amount obtained by normalizing an increase amount per unit time of energy usage based on a pulse signal transmitted from each of the plurality of metering meters in accordance with the use of energy,
Grouping means for classifying the plurality of weighing meters based on a certain rule;
Normal learning means for calculating a threshold value defining a range of values that can be taken when the feature quantity is normal from a standard deviation of the feature quantity for each group of the weighing meters classified by the grouping means;
Detecting means for detecting a change in the state of the meter from the feature value calculated by the feature value calculating means and the threshold value calculated by the normal learning means;
Program to function as.

Images