WO2018198210A1

WO2018198210A1 - Abnormal device discrimination device, method, and program

Info

Publication number: WO2018198210A1
Application number: PCT/JP2017/016446
Authority: WO
Inventors: 滋河本; 鈴木　亮太; ムルトゥザペトラードワラー
Original assignee: 日本電気株式会社
Priority date: 2017-04-25
Filing date: 2017-04-25
Publication date: 2018-11-01

Abstract

The present invention makes it possible to discriminate in which device an abnormality has occurred when the abnormality is detected in the combined waveform of a plurality of devices. An abnormal device discrimination device is provided with: a means for detecting an abnormal waveform from the combined waveform of a plurality of devices; a means for removing the abnormal waveform from the combined waveform; a means for separating the combined waveform from which the abnormal waveform is removed into the waveform of each device and identifying the state of said each device; a means for calculating the degree of coincidence between time during which said each device is in the on-state and time when the abnormal waveform has occurred; and a means for discriminating, on the basis of the degree of coincidence, a device in which an abnormality has occurred.

Description

Abnormal equipment discrimination device, method, program

The present invention relates to an abnormal device determination device, method, and program.

The state of equipment (abnormality) is obtained by acquiring current consumption, power, vibration, etc. of electrical equipment deployed in factories, stores, offices, houses, etc. with sensors, etc., and analyzing the acquired information using various methods. There is known a method for discriminating the presence / absence or the like).

As a related technique for estimating the operating status by monitoring the power supply current and power of an electrical device, for example, Patent Document 1 discloses a measurement for acquiring measurement data indicating a time change of at least one of current consumption and power consumption of an electrical device. A configuration is disclosed that includes a data acquisition unit, a fluctuation component extraction unit that extracts fluctuation components related to fluctuations in current consumption and power consumption from measurement data, and a feature amount acquisition unit that acquires feature amounts of the fluctuation components.

Further, Patent Document 2 discloses a configuration in which combined power consumption data of a plurality of electric devices is separated into devices, and when the power consumption is significantly increased or decreased, the increase or decrease of the devices is estimated.

Furthermore, Patent Document 3 discloses a measured harmonic component calculation result obtained by waveform analysis from the amount of electricity obtained from each component of the device connected to the power system, a failure of each component stored in advance, and an abnormality that is a precursor thereof. Calculate the harmonic component coincidence with multiple harmonic anomalous harmonic components consisting of the harmonic anomaly pattern that is estimated to be a harmonic anomaly. A failure estimation device that identifies harmonic abnormality patterns and outputs them as harmonic abnormality estimation results, and a plurality of failure probability conversion processing units that convert the harmonic abnormality estimation results output from each failure estimation device into failure probabilities, A power quality evaluation system is disclosed that includes a failure probability estimation processing unit that estimates an overall abnormality estimation value of the device from the failure probability of each converted part.

As a related technique for determining an abnormality from a waveform, for example, Patent Document 4 includes a current sensor that detects a load current of a spindle motor of a machine tool, and a normal tool of the machine tool is set for setting a reference value before the abnormality detection is determined. Is processed multiple times, and an allowable range setting means for setting the allowable range based on the reference value obtained from the measured value of the load current by the current sensor at this time is provided. A configuration is provided that includes an abnormality detecting means for determining that the tool is normal if the measured value is within the allowable range, and that the tool is abnormal if the measured value is outside the allowable range.

In addition, in HEMS (Home Energy Management System), BEMS (Building Energy Management System), SEMS (Store Energy Management System), etc., the controller acquires the current waveform, voltage waveform, etc. from the measuring instrument installed in the electrical equipment in real time. In addition to the configuration to be observed, the combined current waveforms of multiple electrical devices flowing through the main board of the distribution board are observed and transferred to the cloud server etc. via the communication network, and the waveforms are separated for each device on the cloud server, The use of device separation technology that estimates the power consumption for each device and the on / off status of each device is also under consideration.

For example, Non-Patent Document 1 acquires a current waveform (instantaneous waveform for one cycle) flowing in the main line using one current sensor attached to a distribution board, and current waveform information unique to each device is obtained. A system that estimates the power consumption of each device and discriminates the state of the device by analyzing the waveform in light of the provided waveform database is disclosed.

International Publication No. 2016/157926 Japanese Unexamined Patent Publication No. 2016-163443 JP 2010-273481 A Japanese Patent Application Laid-Open No. 5-116056

The following is an analysis of related technologies.

According to the related technology described above, when an abnormality occurs in any of a plurality of devices to be separated, it is detected that an abnormality has occurred in a plurality of devices from a composite waveform such as current and power of the plurality of devices. I can.

However, there is a problem that it is impossible to determine which device has an abnormality.

The present invention was devised in view of the above problems, and its purpose is to detect which device has an abnormality when an abnormality is detected in a composite waveform of a plurality of devices. It is to provide a discrimination device, a method, and a program.

According to one aspect of the present invention, means for detecting an abnormal waveform from a combined waveform of a plurality of devices, and removing the abnormal waveform from the combined waveform and separating the waveform into a device-specific waveform, A means for identifying; a degree of coincidence calculating means for calculating a degree of coincidence between the time when each of the devices is turned on and the occurrence time of the abnormal waveform; and a means for determining which device has an abnormality based on the degree of coincidence Is provided.

According to another aspect of the present invention, an abnormal waveform is detected from a combined waveform of a plurality of devices, the abnormal waveform is removed from the combined waveform, and then separated into waveforms for each device, and the separated waveform An abnormal device that identifies the state of the device from each other, calculates the degree of coincidence between the time when each device is on and the occurrence time of the abnormal waveform, and determines the device in which the abnormality occurred based on the degree of coincidence A determination method is provided.

According to another aspect of the present invention, a process of detecting an abnormal waveform from a combined waveform of a plurality of devices, and removing the abnormal waveform from the combined waveform and separating the waveform into each device, A process for identifying a state, a coincidence calculation process for calculating the degree of coincidence between the time when each device is turned on and the occurrence time of the abnormal waveform, and a device in which an abnormality has occurred is determined based on the degree of coincidence. A discrimination process and a program to be executed by a computer are provided.

According to the present invention, a semiconductor storage such as a computer-readable recording medium (for example, RAM (Random Access Memory), ROM (Read Only Memory), or EEPROM (Electrically Erasable and Programmable ROM) ROM) storing the above program, HDD (Hard Disk Drive), CD (Compact Disk), DVD (Digital Versatile Disk) and other non-transitory computer computer readable recording media are provided.

According to the present invention, when an abnormality is detected in a composite waveform of a plurality of devices, it is possible to determine which device has an abnormality.

(A) thru | or (C) is a figure explaining the analysis of the related technique by this invention. It is a figure explaining the structure of one exemplary embodiment of this invention. It is a figure explaining one exemplary embodiment of the present invention. 5 is a flowchart illustrating the processing procedure of an exemplary embodiment of the present invention. (A) thru | or (F) is a figure explaining one illustrative embodiment of this invention. (A) thru | or (D) is a figure explaining one illustrative embodiment of this invention. It is a figure explaining one exemplary embodiment of the present invention. It is a figure explaining the structure of another exemplary embodiment of this invention.

Embodiments of the present invention will be described. First, analysis of related technology according to the present invention will be described. According to related technology equipment separation technology, abnormality detection technology, etc., if an abnormality occurs in any of multiple devices to be separated from the combined waveforms of current and power of multiple devices, multiple devices are abnormal. It is possible to detect that has occurred.

However, it is impossible to determine which device has an abnormality. Hereinafter, a description will be given with reference to FIG.

Referring to FIG. 1A, an abnormal waveform 2 is superimposed on a combined waveform 1 of current or power of an electric device (simply called “device”) A and device B. Assume that the synthesized waveform 1 is separated into the waveform 3 of the device A and the waveform 4 of the device B by the device separation technique. In this case, the abnormal waveform 2 may belong to the waveform 3 of the device A separated from the device, or may belong to the waveform 4 of the device B.

That is, as shown in FIG.
(Device A waveform + Abnormal waveform) + Device B waveform => (Composite waveform including abnormal waveform)
As shown in FIG. 1C,
Waveform of device A + (waveform of device B + abnormal waveform) ⇒ (composite waveform including abnormal waveform)
Holds.

As described above, it is not possible to determine whether the abnormal waveform 2 belongs to the waveform of the device A or the device B only by the device separation.

In addition, when determining which device has an abnormality due to device separation, it is necessary to provide teacher data for normality and abnormality for each device.

Also, when there are multiple types of abnormalities, there is a problem that it is necessary to provide teacher data for each device according to the type of abnormality.

As a result, there is a problem that the capacity of the database increases due to an increase in teacher data, and the accuracy of device separation decreases.

According to one aspect of the present invention for solving the above problems,
(A) An abnormal waveform is detected from a composite waveform of a plurality of devices,
(B) The abnormal waveform is removed from the synthesized waveform and then separated into waveforms for each device to identify the state of the device,
(C) calculating the degree of coincidence between the time when each device is turned on and the time when the abnormal waveform occurs;
(D) Based on the degree of coincidence, a device in which an abnormality has occurred is determined.
An apparatus for executing the above-described series of processing with a processor is provided.

According to the present invention, it is only necessary to provide teacher data for normal operation of each device, and the operation time (on time) of each device separated from the device after removing the abnormal waveform and the occurrence time of the abnormal waveform It is possible to determine which device is abnormal based on the degree of coincidence.

In one embodiment of the present invention, a device in which an abnormality has occurred may be determined based on the degree of coincidence. For example, the ratio of the number of times that the occurrence time of the abnormal waveform overlaps the on-state period of the device is calculated as the degree of coincidence with respect to the number of times the abnormal waveform has occurred, You may make it discriminate | determine from the apparatus which generate | occur | produced.

In one aspect of the present invention, the degree of coincidence is calculated based on a sum of lengths of time in which the on-state period of the device and the period in which the abnormal waveform occurs overlap on the time axis, and the degree of coincidence is maximum. This device may be determined as a device in which an abnormality has occurred. Alternatively, when there are a plurality of devices having substantially the same degree of coincidence, it may be selected as a device suspected of being abnormal. In this case, for example, a device having the highest degree of coincidence may be a device in which an abnormality has occurred, and another device may be a device suspected of being abnormal. Exemplary embodiments of the present invention are described below.

FIG. 2 is a diagram illustrating the functional configuration of the abnormal device determination apparatus 100 according to an exemplary embodiment of the present invention. Referring to FIG. 2, the abnormal device determination apparatus 100 includes a composite waveform acquisition unit 101, an abnormal waveform generation detection unit 102, a storage unit 103, an abnormal waveform removal unit 104, a device separation unit 105, and a state identification unit 106. A degree-of-matching calculation unit 107, a determination unit 108, and an output unit 109. Hereinafter, the outline of each part will be described.

The composite waveform acquisition unit 101 acquires composite waveform data of a plurality of devices that are subject to abnormality determination.

The abnormal waveform generation detection unit 102 detects an abnormal waveform from the combined waveform data acquired by the combined waveform acquisition unit 101 and stores the generation time of the abnormal waveform in the storage unit 103.

The abnormal waveform removing unit 104 removes abnormal waveforms from the combined waveform data.

The device separation unit 105 separates the composite waveform from which the abnormal waveform has been removed into a waveform for each device.

The state identification unit 106 estimates the time transition of the state of each device (for example, on or off state) from the waveform (time series data) separated for each device by the device separation unit 105, and obtains the time series data of the state. Output. In FIG. 2, the device separation unit 105 and the state identification unit 106 are shown as separate units for the purpose of simply explaining the functions, but the functions of the device separation unit 105 and the state identification unit 106 are combined into one unit (for example, Of course, the device separation unit 105) may be combined.

The coincidence calculation unit 107 inputs time series data of the state (for example, on or off state) for each device identified by the state identifying unit 106, inputs the abnormal waveform generation time from the storage unit 103, and outputs the abnormal waveform generation time. And the ratio of the time when the device is on is calculated. For example, for m occurrence times t ₁ ,..., T _{n of} an abnormal waveform of n times, when the number of devices that overlap with the ON state of the device is m, (m / n) × 100 (%) It is good.

FIG. 3 is a diagram illustrating the coincidence degree calculation unit 107 in FIG. 3A and 3B show time transitions of the on / off states of the devices A and B, which are separated by the device separation unit 105 and identified by the state identification unit 106. FIG.

FIG. 3C shows the abnormality as a rectangular wave (pulse waveform) according to the abnormal waveform generation time stored in the storage unit 103. It should be noted that the abnormal waveform generation time may be specified by the period of the start time and end time of the abnormal waveform, as will be described later, instead of the instantaneous time such as t ₁ , t ₂ ,.

In the example of FIG. 3, since the devices A are all in the on state (matched three times) at the times of occurrence of the abnormal waveform three times t ₁ , t ₂ , t ₃ , the matching degree of the device A is ( 3/3) × 100 = 100%.

In addition, in the device B, since the device B is in the on state only at the time t ₂ at the occurrence times t ₁ , t ₂ , and t ₃ of the abnormal waveform three times, the coincidence degree of the device B is (1/3) × 100 = 33%.

The determining unit 108 in FIG. 2 determines that an abnormality has occurred in, for example, a device having the highest degree of matching among a plurality of devices. In the example of FIG. 3, it is determined that an abnormality has occurred in the device A with the higher degree of matching.

FIG. 4 is a flowchart illustrating the operation of an exemplary embodiment of the present invention. Referring to FIG. 4, the synthesized waveform acquisition unit 101 acquires synthesized waveform data of a plurality of devices (step S1).

Next, the abnormal waveform generation detection unit 102 detects an abnormal waveform from the combined waveform data acquired by the combined waveform acquisition unit 101, and stores the generation time of the abnormal waveform in the storage unit 103 (step S2).

Next, the abnormal waveform removing unit 104 removes the abnormal waveform from the combined waveform data (step S3).

Next, the device separation unit 105 separates the synthesized waveform from which the abnormal waveform has been removed into a waveform for each device of the device (step S4).

Next, the state identifying unit 106 identifies the state (for example, on or off) of each device from the waveform (time series data) separated for each device, and the time transition of the state of each device (time series of state) Data) is output (step S5).

Next, the coincidence calculation unit 107 inputs the time series data of the devices identified by the state identification unit 106, inputs the abnormal waveform occurrence time from the storage unit 103, and determines the number of occurrences of the abnormal waveform. On the other hand, the ratio of the number of times the occurrence time of the abnormal waveform overlaps with the ON state period of the device is calculated as the matching degree of the device (step S6).

Next, the determination unit 108 determines, for example, a device having the highest degree of coincidence among a plurality of devices as a device in which an abnormality has occurred (step S7).

Next, the output unit 109 outputs a determination result (information on a device in which an abnormality has occurred) (step S8).

Next, the details of some functional units in FIG. 2 will be described. First, the composite waveform acquisition unit 101 in FIG. 2 will be described.

The combined waveform acquisition unit 101 may use an arbitrary sensor or the like as long as it is a device that acquires a combined waveform of a plurality of devices that are subject to abnormality determination (for example, a waveform obtained by combining power supply currents and power waveforms of a plurality of devices). .

Alternatively, the composite waveform acquisition unit 101 may be configured to acquire a composite waveform via a communication unit from a sensor or the like that acquires composite waveforms of a plurality of devices that are subject to abnormality determination.

FIG. 5A is a diagram schematically illustrating a configuration in which the composite waveform acquisition unit 101 acquires composite current waveforms of a plurality of devices from a controller such as HEMS / BEMS / SEMS / FEMS (Factory Energy Management System). is there. Referring to FIG. 5A, in a building 21 such as a store, business office, or general house, a communication device (HEMS / BEMS / SEMS / FEMS controller) 24 reads meter reading data (power consumption, etc.) of the smart meter 25. For example, it is acquired from the B route. Note that the high-voltage power receiving facility 26 is suitable for high-voltage consumers, and the high-voltage power receiving facility 26 is not necessary for low-voltage consumers such as ordinary houses. The meter reading data (power consumption, current value, etc.) acquired by the communication device 24 from the smart meter 25 through the B route includes information on the power consumption of the entire building. Alternatively, at least one breaker (not shown) of the main breaker (not shown) and the branch breaker (not shown) to which the main power line of the distribution board 22 is connected may be connected to the main breaker or the branch breaker. A current sensor 23 that detects a flowing current (for example, a combined current consumption current of the devices A, B, and C) is provided, and current waveform data is transmitted from the current sensor 23 to the communication device 24 by wireless transmission or the like. May be. The current sensor 23 may be configured by a CT (Current-Transformer) (for example, a zero-phase current sequence (ZCT)), a Hall element, or the like. The current sensor 23 samples a current waveform (analog signal) with an analog-digital converter (not shown), converts it into a digital signal, compresses and encodes it with an encoder (not shown), and then sends it to the communication device 24 with Wi-SUN (Wireless Wireless transmission may be performed by Smart (Utility Network) or the like. The current waveform from the communication device 24 is received by the communication unit 115 of the combined waveform acquisition unit 101. In FIG. 5A, for the sake of simplicity, the device A, the device B, and the device C are connected to the distribution board 22, but it is needless to say that the number of devices is not limited to three. FIG. 5B illustrates a power supply current waveform (total power supply current waveform) acquired by a current sensor 23 connected to a main breaker (not shown) or a branch breaker of the distribution board 22 of FIG. 5A. It is. Note that the combined waveform acquisition unit 101 may acquire a combined power waveform from the communication device 24.

The combined waveform acquisition unit 101 acquires time information when the combined waveform is acquired by the current sensor 23 together with the combined waveform from the communication device 24. When the synthesized waveform is a power supply current waveform, the time information at the sampling time of the synthesized waveform can be calculated from the start time of the synthesized waveform and the sampling frequency of the current sensor 23. The composite waveform acquisition unit 101 may associate the sample value of the composite waveform with the time information and store them in a memory or a buffer (not shown) such as a RAM.

A part of the function of the abnormal device discriminating apparatus 100 is mounted on the communication device (HEMS / BEMS / SEMS / FEMS controller) 24, and the combined waveform acquisition unit 101 includes the current sensor 23 shown in FIG. It is good also as a structure.

Next, the abnormal waveform generation detection unit 102 in FIG. 2 will be described. The abnormal waveform generation detection unit 102 reads the synthetic waveform data from the memory or buffer (for example, RAM) (not shown) of the synthetic waveform acquisition unit 101 and detects an abnormal waveform. An arbitrary method is used to detect the abnormal waveform.

For example, the abnormal waveform occurrence detection unit 102 in FIG. 2 may detect the abnormal waveform by determining whether or not the amplitude of the composite waveform is within an allowable range of a preset upper limit value and lower limit value.

Even if it is determined whether the peak of the abnormal waveform 2 as shown in FIG. 6A is less than the maximum amplitude of the composite waveform 1 and the amplitude of the waveform is within the allowable range, 2 cannot be detected. Therefore, for example, the method illustrated in FIG. 6B may be used as a method for detecting an abnormal waveform whose amplitude is within the allowable range of the upper limit value and the lower limit value set in advance (however, other methods such as frequency spectrum analysis) may be used. Method may be used).

FIG. 6B shows a differential waveform of the composite waveform 1 including the abnormal waveform 2 of FIG. The abnormal waveform generation detection unit 102 in FIG. 2 obtains a differential coefficient (difference value) by differentiating (differing) the composite waveform 1 including the abnormal waveform 2 in FIG. 6A, and obtaining this as a threshold value (upper limit value or lower limit value). ) To detect an abnormal waveform section. Note that. For simplicity, the abnormal waveform 2 has a substantially triangular waveform in FIG. 6A, and the differential waveform of the abnormal waveform 2 is represented by a rectangular wave (pulse) in FIG. 6B. In FIG. 6A, for ease of explanation, the combined waveform 1 of a plurality of devices is a periodic waveform composed of a sine waveform, and the differential waveform of FIG. 6B is a cosine waveform.

The slope (derivative coefficient) (positive value) of the rise to the peak of the abnormal waveform 2 (substantially triangular wave) in FIG. 6A is larger than the maximum slope (derivative coefficient) of the original composite waveform 1 and is the upper limit value. The slope (derivative coefficient) (negative value) of the falling from the peak exceeding UL (positive value) is larger in absolute value than the negative absolute value of the slope (derivative coefficient) of the original composite waveform 1 Yes, below the lower limit LL (negative value).

For this reason, the abnormal waveform generation detection unit 102 crosses the generation time of the abnormal waveform 2 (the differential value exceeds the upper limit value and the lower limit value) depending on whether or not the differential value (difference value) of the composite waveform 1 is within the allowable range. Time) can be detected. That is, the differential value (difference value) of the composite waveform 1 changes within a predetermined range in a time interval in which a plurality of devices operate normally and the composite waveform changes smoothly. If an abnormal waveform appears while the composite waveform is moving smoothly, the difference between the previous sample value and the sample value of abnormal waveform 2 will exceed the upper limit or lower limit, and an abnormal waveform will be generated. The time (abnormal section) will be detected.

Alternatively, the abnormal waveform generation detection unit 102 may predict the next sample value based on the predetermined number of sample values up to that point of the composite waveform 1 in FIG. . For example, the predicted value ^ Y (n) of the composite waveform 1 at the sampling point n is used as the value of the composite waveform 1 at a predetermined number of sampling points s of the composite waveform 1 so far: Y (n−1), Y (n -2), ..., Y (ns), and if the absolute value of the difference between the predicted value ^ Y (n) and the actual sample value Y (n) is greater than or equal to a predetermined value, it is determined as abnormal You may make it do. Although not particularly limited, as the predicted value ^ Y (n) of the composite waveform 1 at the sampling point n, for example, the following linear prediction:
^ Y (n) = a ₁ × Y (n−1) +... + A _s × Y (ns)
May be used.

Alternatively, the abnormal waveform generation detection unit 102 divides the combined waveform into predetermined time intervals, performs FFT (Fast Fourier Transform) or the like in each time interval, converts the waveform into the frequency domain, and is derived from the frequency spectrum component. The amount (for example, the square root of the square addition of the even-order harmonic component, the square root of the square addition of the odd-order harmonic component, or THD (Total Harmonic Distortion)) is obtained, and the feature amount is determined. The abnormal section of the composite waveform may be determined by comparing with the threshold value. Alternatively, in the frequency spectrum, abnormality may be determined based on the magnitude of the power spectrum of a specific high frequency component.

Or the abnormal waveform generation | occurrence | production detection part 102 learns the synthetic | combination waveform at the time of normal, and the Mahalanobis distance showing the deviation | decrease degree of the data at the time of the test | inspection with respect to the distribution of the electric synthesis waveform at the time of normal with respect to the synthetic | combination waveform at the time of an inspection. Or the like may be used to detect an outlier or the like and detect it as an abnormal waveform.

As shown in FIG. 6C, when the device changes from the off state to the on state (time t ₂ ), the power consumption waveform rises rapidly. In addition, when the device changes from the on state to the off state (time t ₁ ), the power consumption waveform suddenly falls. The same applies to the state transition between the standby state and the on state that suppresses the power consumption of the device. The abnormal waveform generation detection unit 102 may not detect such a change in the composite waveform accompanying a change in the operation mode of the device as an abnormal waveform. For example, the abnormal waveform generation detection unit 102 may include (a) a transition from an off state to an on state, (b) a transition from an on state to an off state, (c) a transition from a standby state to an on state, (d ) Among the transitions from the on state to the standby state, (a) and (b), or (c) and (d), or (a) to (d) transition of the current or power composite waveform For example, learning may be performed as teacher data of a normal operation waveform, and a change in the composite waveform at the time of transition may not be detected as an abnormal waveform at the time of evaluation.

Next, the abnormal waveform removing unit 104 in FIG. 2 will be described. The abnormal waveform removing unit 104 removes the abnormal waveform 2 from the synthesized waveform 1. The abnormal waveform removing unit 104 applies a low-pass filter having a predetermined cutoff frequency to the composite waveform 1 including the abnormal waveform 2 (for example, FIG. 6A), thereby removing the abnormal waveform 2 (FIG. 6 (FIG. 6 (A)). D) may be acquired. The low-pass filter may be a FIR (Finite Impulse Response) filter. The abnormal waveform removal unit 104 converts time series data of the composite waveform 1 into a frequency domain using FFT or the like, sets a frequency component equal to or higher than a predetermined cutoff frequency to 0, and IFFT (Inverse FFT: Inverse Fast Fourier Transform) ) Etc. to return to the time domain, low pass filter calculation may be performed.

Alternatively, the abnormal waveform removing unit 104 may generate a composite waveform from which the abnormal waveform 2 is removed by interpolating a time interval corresponding to the abnormal waveform 2 from the composite waveform 1. .

Next, the device separation unit 105 in FIG. 2 will be described. The device separation unit 105 separates the synthesized waveform from which the abnormal waveform has been removed into a waveform for each device using any method described in Non-Patent Document 1 or the like. The device separation unit 105 separates the synthesized waveform from which the abnormal waveform has been removed into a waveform for each device by referring to a waveform database (not shown) that stores a waveform pattern during operation for each device and its feature amount. You may do it. That is, the device separation unit 105 calculates a feature amount of the synthesized waveform for a predetermined time interval (for example, one cycle (20 milliseconds) or a plurality of cycles of the commercial power supply frequency, or a time interval in which one cycle is divided into a plurality of times). The waveform may be separated for each device based on the feature amount or the waveform pattern unique to the device. The waveform database may be provided in the device separation unit 105 or may be provided outside the device separation unit 105. For example, the device separation unit 105 may connect to a data server or the like via a communication network and access the waveform database of the data server. A waveform shape (waveform peak value, average value, effective value (root （mean 等 square), peak value, etc.) may be used as the feature amount of the waveform. Alternatively, a Fourier transform (fast Fourier transform or discrete Fourier transform, or short-time fast Fourier transform or short-time discrete Fourier transform) or the like is performed on a predetermined time window of the composite waveform 1 and the like, Value derived from square addition of amplitude of frequency spectrum component (for example, square root of square addition of even-order harmonic component, square root of square addition of even-order harmonic component, or THD (Total Harmonic Distortion)) ) May be the feature amount. FIG. 5C to FIG. 5E schematically show current waveforms separated for each device by the device separation unit 105 for each of the

devices

20C, 20B, and 20A.

Next, the state identification unit 106 in FIG. 2 will be described. Note that, as described above, the device separation unit 105 and the state identification unit 106 may be combined into one unit. The state identification unit 106 estimates and identifies the time transition of the state of each device (for example, on state, off (standby) state) from the waveform data (time series) data separated for each device by the device separation unit 105. , Output time-series data of the state of each device. In the waveform database (not shown), waveform patterns and feature amounts corresponding to the states of the respective devices are stored, and the state identification unit 106 refers to the waveform database (not shown) to check the state of the waveform data in each time interval. May be estimated. FIG. 5F schematically shows time-series data of the state of the device A output by the state identification unit 106, for example, for the waveform of the device A 20A of FIG. 5A (FIG. 5E).

For a plurality of devices (for example, device A, device B, device C, etc. in FIG. 5A) (device A is on, device B, device C is off), device B is on, device A, device (C is off), (device C is on, device A, device B is off), (device A, device B is on, device C is off), ..., (device A, device B, device C are all ON) waveform pattern or feature value for combination, etc., and fluctuation amount are stored in the waveform database in advance, or the waveform pattern or feature value for combination, and machine learning result of fluctuation amount etc. are stored in waveform database Then, waveform separation for each device from the synthesized waveform in the device separation unit 105 and state estimation of each device in the state identification unit 106 may be performed in cooperation.

Further, depending on the device, a waveform pattern or a feature amount corresponding to each operation mode in the ON state (for example, the air volume of the air conditioner is strong, medium, weak, etc.) is stored in the waveform database. You may make it discriminate | determine the operation mode of the apparatus of an ON state from the waveform isolate | separated for every apparatus. In this case, it is possible to determine in which operation mode of the device an abnormal waveform occurs.

Next, the coincidence degree calculation unit 107 in FIG. 2 will be described. As described above, the degree-of-match calculation unit 107 inputs the time series data of the devices identified by the state identification unit 106, inputs the abnormal waveform generation time from the storage unit 103, and sets the abnormal waveform generation time and Calculate the proportion of the devices in the on-state time.

Alternatively, the degree-of-match calculation unit 107 may calculate a ratio between a time in which the period in which the device is on and the period in which the abnormal waveform occurs overlap on the time axis and the period in which the device is on as the degree of coincidence. .

Referring to FIG. 7, the number of times that the abnormal waveform occurrence time coincides with the on-state time of device A and device B is three for both device A and device B (depending on how the abnormal waveform occurrence time is determined). The number of matches will be 3).

In this case, a case where the time when the device A and the device B are in the ON state in a part of the time period of the abnormal waveform generation period is also regarded as coincident. In the device A, the ON period is completely overlapped in time with _three abnormal waveform generation periods T ₁ = t ₁₁ -t ₁₂ , T ₂ = t ₂₁ -t ₂₂ , and T ₃ = t ₃₁ -t ₃₂ However, in the device B, the ON period only partially overlaps any of the _three abnormal waveform generation periods T ₁ , T ₂ , and T ₃ .

For this reason, when calculating the degree of coincidence of the devices based on the sum of the time over which the period in which the abnormal waveform occurs and the on-state period of the device overlap on the time axis, in device A, the sum of the times is
T ₁ + T ₂ + T ₃
It becomes.

The degree of coincidence of the equipment may be a value obtained by dividing the sum of the time in which the abnormal waveform occurs and the period in which the equipment is on on the time axis by dividing the sum by the period in which the abnormal waveform has occurred, as a percentage. . In this case, the matching degree of the device A is
(T ₁ + T ₂ + T ₃ ) / (T ₁ + T ₂ + T ₃ ) × 100 = 100%
It is. On the other hand, the degree of coincidence of device B is less than 100%.

Next, the determination unit 108 in FIG. 2 will be described. As described above, the determination unit 108 determines the device having the highest degree of coincidence as a device in which an abnormality corresponding to the abnormal waveform has occurred. At this time, if there are multiple devices that have almost the same degree of coincidence, or a slight difference, or the difference is not statistically significant, these devices, together with the device with the highest degree of coincidence, are suspected of being abnormal. It is good also as a structure selected as an apparatus to be called. At this time, the device having the highest degree of coincidence may be selected as a device in which an abnormality has occurred, and another device may be selected as a device suspected of being abnormal. In this case, a maintenance person or the like may individually check the presence or absence of each abnormality for a plurality of devices determined to be suspected of abnormality. Similarly, for equipment determined to be abnormal, maintenance personnel or the like may investigate whether there is an abnormality and take necessary measures (for example, repair / part replacement).

Next, the output unit 109 in FIG. 2 will be described. The output unit 109 may output the determination result of the presence or absence of abnormality in the determination unit 108 to a display device or the like. Alternatively, the output unit 109 may be configured to notify the maintenance person's terminal or the HEMS / BEMS / SEMS / FEMS controller of FIG. 5A via a communication network.

FIG. 8 is a diagram illustrating a configuration in which the abnormal device determination device 100 is mounted on a computer system as another embodiment of the present invention. Referring to FIG. 8, a computer system such as a server computer includes a processor (CPU (Central Processing Unit), data processing device) 111, a semiconductor memory (for example, RAM (Random Access Memory), ROM (Read Only Memory), or EEPROM. (Electrically Erasable and mProgrammable ROM), HDD (Hard Disk Drive), CD (Compact Disc), DVD (Digital Versatile Disc), etc., a storage device 112, a display device 113, and a communication interface 114 I have. The communication interface 114 may function as the communication unit 115 in FIG. 2 may output the determination result to the display device 113, for example. The storage device 112 may be the same device as the storage unit 103 in FIG. The storage device 112 stores a program that realizes the function of the abnormal device determination device 100 of FIG. 2, and the processor 111 reads and executes the program, thereby realizing the abnormal device determination device 100 of the above-described embodiment. You may make it do. The computer system of FIG. 8 may be implemented as a cloud server that provides an abnormal device determination service to a client as a cloud service.

In the above-described embodiment, the combined waveform of the plurality of devices is not limited only to the current waveform and the power waveform, but at least a sensor (vibration sensor, acoustic sensor, etc.) that obtains a combined waveform of signals from a plurality of signal sources. Of course, the synthesized signal sensed by one may be used.

In the above embodiment, for the sake of simplicity, a transient waveform such as a single triangular wave (such as a glitch) has been described as an example of an abnormal waveform. However, AC (Alternating Current) ripple (ripple noise) or the like repeatedly appears. Of course, it may be a waveform such as noise, or may be an abnormal waveform such as DC level fluctuation, such as DC (Direct Current) offset noise.

It should be noted that the disclosures of Patent Documents 1-4 and Non-Patent Document 1 described above are incorporated herein by reference. Within the scope of the entire disclosure (including claims) of the present invention, the embodiments and examples can be changed and adjusted based on the basic technical concept. Various combinations or selections of various disclosed elements (including each element of each claim, each element of each embodiment, each element of each drawing, etc.) are possible within the scope of the claims of the present invention. . That is, the present invention of course includes various variations and modifications that could be made by those skilled in the art according to the entire disclosure including the claims and the technical idea.

The above embodiment is appended as follows (however, it is not limited to the following).

(Appendix 1)
Means for detecting an abnormal waveform from a composite waveform of a plurality of devices;
Means for removing the abnormal waveform from the synthesized waveform;
Means for separating the synthesized waveform from which the abnormal waveform is removed into waveforms for each device and identifying the state of the device;
A degree of coincidence calculating means for calculating a degree of coincidence between the time when each of the devices is on and the occurrence time of the abnormal waveform;
Discrimination means for discriminating a device in which an abnormality has occurred based on the degree of coincidence;
An abnormal device discriminating apparatus comprising:

(Appendix 2)
The abnormal device determination apparatus according to appendix 1, wherein the determination unit determines a device in which an abnormality has occurred based on the magnitude of the degree of coincidence.

(Appendix 3)
The degree of coincidence calculation means calculates, as the degree of coincidence, the ratio of the number of times the occurrence time of the abnormal waveform overlaps the on-state period of the device with respect to the number of times the abnormal waveform has occurred,
3. The abnormal device determination device according to appendix 1 or 2, wherein the determination unit determines that the device having the highest degree of coincidence is a device in which an abnormality has occurred.

(Appendix 4)
The degree of coincidence calculating means calculates the degree of coincidence based on the sum of lengths of time in which the period in which the device is on and the period in which the abnormal waveform occurs overlap on the time axis,
The abnormal device discriminating apparatus according to appendix 1 or 2, wherein the determination unit discriminates a device having the highest degree of coincidence from a device in which an abnormality has occurred.

(Appendix 5)
The abnormal device discriminating apparatus according to any one of appendices 1 to 4, wherein the discriminating unit selects a device suspected of being abnormal when there are a plurality of devices having substantially the same degree of coincidence.

(Appendix 6)
Detect abnormal waveforms from composite waveforms of multiple devices,
Removing the abnormal waveform from the combined waveform;
Separate the synthesized waveform from which the abnormal waveform has been removed into a waveform for each device, and identify the state of the device,
Calculate the degree of coincidence between the on-state time of each device and the occurrence time of the abnormal waveform,
An abnormal device determination method for determining a device in which an abnormality has occurred based on the degree of coincidence.

(Appendix 7)
The abnormal device determination method according to appendix 6, wherein the determination unit determines a device in which an abnormality has occurred based on the magnitude of the degree of coincidence.

(Appendix 8)
In calculating the degree of coincidence, the ratio of the number of times the occurrence time of the abnormal waveform overlaps the ON state period of the device is calculated as the degree of coincidence with respect to the number of times the abnormal waveform has occurred.
8. The abnormal device determination method according to appendix 6 or 7, wherein the device having the highest degree of coincidence is determined as a device in which an abnormality has occurred.

(Appendix 9)
In calculating the degree of coincidence, the degree of coincidence is calculated based on the length of time in which the on-state period of the device and the period in which the abnormal waveform occurs overlap on the time axis,
8. The abnormal device determination method according to appendix 6 or 7, wherein the device having the highest degree of coincidence is determined as a device in which an abnormality has occurred.

(Appendix 10)
The abnormal device determination method according to any one of appendices 6 to 9, wherein when there are a plurality of devices having substantially the same degree of coincidence, the device is selected as a device suspected of being abnormal.

(Appendix 11)
A process of detecting an abnormal waveform from a composite waveform of multiple devices;
Processing to identify the state of the device by removing the abnormal waveform from the composite waveform and then separating the waveform into a device-specific waveform;
A degree of coincidence calculation process for calculating a degree of coincidence between the on-time of each device and the occurrence time of the abnormal waveform;
A discrimination process for discriminating a device in which an abnormality has occurred based on the degree of coincidence;
A program that causes a computer to execute.

(Appendix 12)
The program according to appendix 11, wherein the discrimination process discriminates a device in which an abnormality has occurred based on the degree of coincidence.

(Appendix 13)
The degree of coincidence calculation processing calculates, as the degree of coincidence, the ratio of the number of times the occurrence time of the abnormal waveform overlaps the on-state period of the device with respect to the number of times the abnormal waveform has occurred,
The program according to appendix 11 or 12, wherein the determination processing determines that the device having the highest degree of coincidence is a device in which an abnormality has occurred.

(Appendix 14)
The degree of coincidence calculation processing calculates the degree of coincidence based on the sum of lengths of time in which the period in which the device is on and the period in which the abnormal waveform occurs overlap on the time axis,
The program according to appendix 11 or 12, wherein the determination processing determines that the device having the highest degree of coincidence is a device in which an abnormality has occurred.

(Appendix 15)
The program according to any one of appendices 11 to 14, wherein the determination unit selects a device suspected of being abnormal when there are a plurality of devices having substantially the same degree of coincidence.

1 Composite Waveform 2 Abnormal Waveform 3 Device A Waveform 4 Device B Waveform 20A Device A
20B Equipment B
20C Equipment C
21 Building 22 Distribution board 23 Current sensor 24 Communication device (HEMS / BEMS / SEMS / FEMS controller)
25 Smart Meter 26 High Voltage Power Receiving Equipment 100 Abnormal Device Discriminating Device 101 Composite Waveform Acquisition Unit 102 Abnormal Waveform Generation Detection Unit 103 Storage Unit 104 Abnormal Waveform Removal Unit 105 Device Separation Unit 106 State Discrimination Unit 107 Matching Level Calculation Unit 108 Discrimination Unit 109 Output Unit 111 Processor 112 Storage Device 113 Display Device 114 Communication Interface 115 Communication Unit

Claims

Means for detecting an abnormal waveform from a composite waveform of a plurality of devices;
Means for removing the abnormal waveform from the synthesized waveform;
Means for separating the synthesized waveform from which the abnormal waveform is removed into waveforms for each device and identifying the state of the device;
A degree of coincidence calculating means for calculating a degree of coincidence between the time when each of the devices is on and the occurrence time of the abnormal waveform;
Discrimination means for discriminating a device in which an abnormality has occurred based on the degree of coincidence;
An abnormal device discriminating apparatus comprising:
The abnormal device determination device according to claim 1, wherein the determination unit determines a device in which an abnormality has occurred based on the magnitude of the degree of coincidence.
The degree of coincidence calculation means calculates, as the degree of coincidence, the ratio of the number of times the occurrence time of the abnormal waveform overlaps the on-state period of the device with respect to the number of times the abnormal waveform has occurred,
The abnormal device determination apparatus according to claim 1, wherein the determination unit determines that the device having the highest degree of coincidence is a device in which an abnormality has occurred.
The degree of coincidence calculating means calculates the degree of coincidence based on the sum of lengths of time in which the period in which the device is on and the period in which the abnormal waveform occurs overlap on the time axis,
The abnormal device determination apparatus according to claim 1, wherein the determination unit determines that the device having the highest degree of coincidence is a device in which an abnormality has occurred.
5. The abnormal device determination device according to claim 1, wherein, when there are a plurality of devices having substantially the same degree of coincidence, the determination unit selects the device as suspected of being abnormal.
Detect abnormal waveforms from composite waveforms of multiple devices,
Removing the abnormal waveform from the combined waveform;
Separate the synthesized waveform from which the abnormal waveform has been removed into a waveform for each device, and identify the state of the device,
Calculate the degree of coincidence between the on-state time of each device and the occurrence time of the abnormal waveform,
An abnormal device determination method for determining a device in which an abnormality has occurred based on the degree of coincidence.
The abnormality determination method according to claim 6, wherein the determination unit determines an apparatus in which an abnormality has occurred based on the magnitude of the degree of coincidence.
In calculating the degree of coincidence, the ratio of the number of times the occurrence time of the abnormal waveform overlaps the ON state period of the device is calculated as the degree of coincidence with respect to the number of times the abnormal waveform has occurred.
The abnormal device determination method according to claim 6 or 7, wherein the device having the highest degree of coincidence is determined as a device in which an abnormality has occurred.
In calculating the degree of coincidence, the degree of coincidence is calculated based on the length of time in which the on-state period of the device and the period in which the abnormal waveform occurs overlap on the time axis,
The abnormal device determination method according to claim 6 or 7, wherein the device having the highest degree of coincidence is determined as a device in which an abnormality has occurred.
10. The abnormal device determination method according to claim 6, wherein when there are a plurality of devices having substantially the same degree of coincidence, the device is selected as a device suspected of being abnormal.
A process of detecting an abnormal waveform from a composite waveform of multiple devices;
Processing to identify the state of the device by removing the abnormal waveform from the composite waveform and then separating the waveform into a device-specific waveform;
A degree of coincidence calculation process for calculating a degree of coincidence between the on-time of each device and the occurrence time of the abnormal waveform;
A discrimination process for discriminating a device in which an abnormality has occurred based on the degree of coincidence;
A program that causes a computer to execute.