Electrical equipment inspection device

The electrical equipment inspection device addresses the oversight of component placement defects by using a control unit to compare actual installation data with design data, enhancing detection accuracy and efficiency.

JP2026181945APending Publication Date: 2026-11-19MITSUBISHI ELECTRIC CORP
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
JP2025077782
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2025-05-08
Publication Date
2026-11-19

AI Technical Summary

Technical Problem

Conventional AI-based inspection methods for electrical equipment overlook defects in the placement of components during assembly, despite using machine learning to detect defects in the components themselves.

Method used

An electrical equipment inspection device that uses a control unit to detect abnormalities in component arrangement by reading unique identification information and actual installation data, comparing it with pre-recorded design data to identify discrepancies.

Benefits of technology

Accurately detects abnormal arrangements of components, reducing the risk of overlooking placement errors and improving inspection efficiency compared to visual methods.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2026181945000001_ABST
    Figure 2026181945000001_ABST
Patent Text Reader

Abstract

The present invention provides an electrical equipment inspection device capable of detecting abnormal placement of components that make up an electrical system. [Solution] The control unit (50) of the electrical equipment inspection device (100) includes an extraction unit (51) that reads equipment identification information, enclosure identification information, area identification information, and actual position information of each component from the electrical equipment as actual installation information of the electrical equipment, and a recording unit (54) that records equipment identification information, enclosure identification information, area identification information, first parent identification information that specifies the enclosure to which the area should belong, and second parent identification information that specifies the area to which the equipment should belong as design data (D). Using the extracted actual installation information, the control unit (50) derives area identification information for the area where the equipment is installed, and detects an abnormality in the installation of the equipment based on the derived area identification information, the first parent identification information, and the second parent identification information of the equipment that are recorded in advance in the design data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present disclosure relates to an inspection apparatus for electrical equipment.

Background Art

[0002] Conventionally, in the inspection of switchboards as electrical equipment used in buildings, factories and the like, after the assembly of the switchboard is completed, an inspector visually determines whether each component constituting the switchboard is in a normal state and whether each component is arranged in the switchboard as designed in the drawing. However, when the number of inspection items is large, or when components with similar appearances are arranged, there is a risk that the inspector may overlook an abnormality even if the inspection is performed visually. Therefore, conventionally, there has been proposed a defective product determination apparatus that performs the following defective product determination method, which executes inspection of electrical equipment using AI (Artificial Intelligence).

[0003] That is, the conventional defective product determination apparatus inputs data of a plurality of determination targets into a network to extract features of the determination targets, determines whether the distribution of the features of the determination targets follows a normal distribution, minimizes the error in feature extraction, calculates the anomaly score of the determination targets, performs threshold processing on the anomaly score, and determines whether the determination target is a normal product or a defective product (see, for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of Invention

Problems to be Solved by the Invention

[0005] The conventional defective product detection device described above uses AI-based machine learning to determine if each component constituting the distribution panel is defective or not. However, in most cases, the electrical equipment and other components built into the distribution panel are purchased as finished products. Therefore, defects in the components themselves are often not detected as defects because they are considered normal, but there is a problem in that defects in the placement of components during the assembly of the distribution panel may be overlooked.

[0006] This disclosure provides technology to solve the above-mentioned problems and aims to provide an electrical equipment inspection device that can accurately detect abnormal arrangement of components constituting electrical equipment. [Means for solving the problem]

[0007] The electrical equipment inspection device described herein is An electrical equipment inspection device that uses a control unit to detect abnormalities in the arrangement of one or more devices as components, in electrical equipment provided by arranging one or more devices as components in a region within one or more boxes as components, The control unit, As actual installation information for the aforementioned electrical equipment, An extraction unit reads from the electrical equipment the following: equipment identification information as unique identification information for the equipment, enclosure identification information as unique identification information for the enclosure, area identification information as unique identification information for the area, and actual position information for each of the components. The system includes a recording unit that pre-records the following as design data: the equipment identification information, the enclosure identification information, the area identification information, a first parent identification information specifying the enclosure to which the area belongs, and a second parent identification information specifying the area to which the equipment belongs. Using the actual installation information extracted by the extraction unit, the region identification information of the region in which the equipment is installed in the electrical equipment is derived. The derived area identification information in which the equipment is installed, Based on the first parent identification information and the second parent identification information of the device, which are pre-recorded in the design data, an abnormality in the placement of the device is detected. It is. [Effects of the Invention]

[0008] According to the electrical equipment inspection device of this disclosure, an electrical equipment inspection device is obtained that can accurately detect abnormal arrangements of components constituting the electrical equipment. [Brief explanation of the drawing]

[0009] [Figure 1] This is a block diagram showing the schematic configuration of an electrical equipment inspection device according to Embodiment 1. [Figure 2] This figure shows the schematic configuration of the design data for the switchboard included in the electrical equipment inspection device according to Embodiment 1. [Figure 3] This is a schematic diagram illustrating the configuration of the switchboard corresponding to the design data shown in Figure 2. [Figure 4] This figure shows the control flow of the inspection device according to Embodiment 1. [Figure 5] This figure shows the schematic configuration of a power distribution panel to be inspected by the inspection device according to Embodiment 1. [Figure 6] This figure shows an example of the results obtained by the inspection device according to Embodiment 1, which identifies each component of the power distribution panel shown in Figure 5: the enclosure, area, equipment, and nameplate. [Figure 7] This figure shows another schematic configuration of the power distribution panel that the inspection device according to Embodiment 1 performs inspection on. [Figure 8] This figure shows an example of the results obtained by the inspection device according to Embodiment 1, which identifies each component of the power distribution panel shown in Figure 7, nameplate, and enclosure. [Figure 9] This figure shows a schematic configuration of the learning device for an electrical equipment inspection device according to Embodiment 2. [Figure 10] This figure shows an example of training data for an electrical equipment inspection device according to Embodiment 2. [Figure 11]This is a diagram showing the neural network of the electrical equipment inspection apparatus according to the second embodiment. [Figure 12] This is a flowchart relating to the learning processing of the electrical equipment inspection apparatus according to the second embodiment. [Figure 13] This is a block diagram showing the configuration of the determination device of the electrical equipment inspection apparatus according to the second embodiment. [Figure 14] This is a flowchart showing the processing of the inference unit of the electrical equipment inspection apparatus according to the second embodiment. [Figure 15] This is a diagram showing the recognition image of the inference unit of the electrical equipment inspection apparatus according to the second embodiment. [Figure 16] This is a diagram showing the recognition result of the inference unit of the electrical equipment inspection apparatus according to the second embodiment. [Mode for Carrying Out the Invention]

[0010] Embodiment 1. Figure 1 is a block diagram showing the schematic configuration of an electrical equipment inspection apparatus 100 according to the first embodiment. Figure 2 is a diagram showing the schematic configuration of design data D of a switchboard 20, which is provided in the electrical equipment inspection apparatus 100 according to the first embodiment. Figure 3 is a schematic diagram for explaining the configuration of the switchboard 20 corresponding to the design data D shown in Figure 2.

[0011] The electrical equipment inspection apparatus 100 of the present embodiment (hereinafter referred to as inspection apparatus 100) is configured to detect assembly anomalies of each component constituting the switchboard 20, which is electrical equipment disposed in buildings, factories and the like to distribute received electric power to a plurality of loads. First, the schematic configuration of the switchboard 20 will be described with reference to Figure 3. The switchboard 20 shown in Figure 3 has an exemplary configuration in which each constituent part such as components is accurately assembled as shown in the design drawing, and is a switchboard that serves as a reference for the design data D to be described later.

[0012] The switchboard 20 has two boxes, namely a box 1 and a box 2, as constituent parts. The boxes 1 and 2 are, for example, metal casings and have multiple regions as components that divide their internal space. In this power distribution panel 20, box 1 has region 11 at the top and region 12 at the bottom. Box 2 has region 21 at the top, region 22 in the middle, and region 23 at the bottom. Each area houses components that make up the distribution panel 20, such as electrical equipment including power distribution equipment and control equipment, mechanical equipment such as shutters and heat sinks, and nameplates that display information about each component. Note that cables and other components used to connect the devices are not shown in the diagram.

[0013] As shown in Figure 3, equipment 1 and nameplate 11 are arranged within area 11 of the enclosure 1. Equipment 2 and nameplate 12 are arranged within area 12 of the enclosure 1. Furthermore, equipment 3 and nameplate 21 are located within area 21 of box body 2. Nameplate 22 is located within area 22 of box body 2. Equipment 4 and equipment 5 are located within area 23 of box body 2.

[0014] Next, the inspection device 100 will be described. As shown in Figure 1, the inspection device 100 of this embodiment includes a control unit 50 and an alarm device 60 as an alarm unit. The control unit 50 includes an image data extraction unit 51 as an extraction unit, a design data acquisition unit 52, an input data acquisition unit 53, a recording unit 54, and a decision device 55.

[0015] The image data extraction unit 51 captures images such as photographs of the power distribution board to be inspected, reads the unique identification information of each component based on the identifier described later in the captured image, and also reads their actual location information. The read identification information and actual location information of each component are transmitted to the judgment device 55 as actual installation information of the power distribution board. The unique identification information for each component specifically includes the unique enclosure identification information for each enclosure, the unique area identification information for each region, and the unique device identification information for each device.

[0016] The design data acquisition unit 52 is an input unit that acquires the design data D shown in Figure 2, which will be described later, from an external source. The input data acquisition unit 53 is an input unit for inputting unique information, such as identification information, from an external source for boxes, areas, equipment, and nameplates whose identification information could not be recognized from their identifiers. The recording unit 54 is a recording device that stores the design data D acquired by the design data acquisition unit 52.

[0017] The judgment device 55, as will be described in detail later, determines assembly defects such as incorrect placement of each component of the distribution board based on the design data D recorded in the recording unit 54 and the actual installation information of the distribution board under inspection, and outputs the judgment result to the alarm device 60.

[0018] The hardware configuration of the control unit 50 is such that the decision-making device 55 is a processor and the recording unit 54 is a storage device. The recording unit 54, which is a storage device, includes a volatile storage device such as a random access memory (not shown) and a non-volatile auxiliary storage device such as a flash memory. Alternatively, a hard disk may be provided as an auxiliary storage device instead of flash memory. The decision-making device 55, which is a processor, executes the program input from the recording unit 54, which is a storage device. In this case, the program is input to the decision-making device 55 from the auxiliary storage device via the volatile storage device. The decision-making device 55 may also output data such as calculation results to the volatile storage device of the recording unit 54, which is a storage device, or it may save the data to the auxiliary storage device via the volatile storage device.

[0019] Next, we will explain the design data D shown in Figure 2. The design data D recorded in the recording unit 54 corresponds to the switchboard 20, which is a model device assembled precisely according to the design drawings, and shows the precise configuration information in which each component, such as the enclosure, area, equipment, and nameplate, should be arranged. As shown in Figure 2, design data D records identification information for the enclosure, area, equipment, and nameplate.

[0020] In this embodiment, as unique identification information corresponding to each of the box bodies 1 and 2, box body identification information, namely Jbox1 and Jbox2, is recorded in association with each box body. Furthermore, each of the devices 1, 2, 3, 4, and 5 is associated with and recorded as unique identification information, specifically device identification information such as JDevice1, JDevice2, JDevice3, JDevice4, and JDevice5. Furthermore, as unique identification information corresponding to each of the nameplates 11, 12, 21, and 22, the nameplate identification information that identifies each nameplate, nameplate 11, nameplate 12, nameplate 21, and nameplate 22, is recorded in association with each nameplate. Furthermore, as unique identification information corresponding to each of the regions 11, 12, 21, 22, and 23, region identification information that identifies each region, namely J-region 11, J-region 12, J-region 21, J-region 22, and J-region 23, is recorded in association with each region.

[0021] Furthermore, since the area belongs to the enclosure, and the equipment and nameplate also belong to the area, they have a parent-child relationship, and the attribute relationship of this area to the enclosure is indicated by the parent identification information. In Figure 2, the identification information for boxes 1 and 2, namely P-box 1 and P-box 2, is recorded as the first parent identification information that identifies the box to which each region 11, 12, 21, 22, and 23 should belong. Furthermore, as second-parent identification information that identifies the area to which each device 1, 2, 3, 4, and 5 should belong, the identification information for area 11 and area 12, namely P-area 11 and P-area 12, are recorded in association with each other. Furthermore, as third-parent identification information that identifies the area to which each nameplate 11, 12, 21, and 22 should belong, the identification information for area 11 and area 12, nameplates P-area 11 and P-area 12, are associated and recorded.

[0022] Furthermore, each component, such as the enclosure, area, equipment, and nameplate, has its own unique name, and if the names are the same, it indicates that they have the same appearance and dimensions. Furthermore, the nameplate has printed information about the nameplates of each component that makes up the distribution panel 20. While the example shown uses identification information and parent identification information to indicate parent-child relationships, any notation is acceptable as long as the parent-child relationships between each component are clear, such as representing them hierarchically.

[0023] The control operations performed by the inspection device 100 configured as described above during the inspection of a power distribution board will now be explained. Figure 4 shows the control flow of the inspection device 100 in this embodiment. Figure 5 shows a schematic configuration of the power distribution panel 20R1 that the inspection device 100 of this embodiment performs inspection on. Figure 6 shows an example of the results obtained by the inspection device 100 of this embodiment, which identifies each component of the distribution board 20R1, including the box, area, equipment, and nameplate.

[0024] As shown in Figure 5, the distribution panel 20R1, which is the subject of inspection, has a barcode C as an identifier that shows unique identification information for each of its components: the enclosure, area, equipment, and nameplate. Each barcode C is affixed to a designated location, such as the upper left corner of each box, area, piece of equipment, or nameplate.

[0025] First, the image data extraction unit 51 captures image data of the distribution panel 20R shown in Figure 5, and extracts identification information of the box, area, equipment, and nameplates in the photograph, as well as their actual location information, as actual installation information (step S001). This identification is performed by image processing of the image data, including the barcode C.

[0026] In step S001, first, the image data extraction unit 51 identifies the box 1 based on the box identification information (H1) read from the barcode C. Then, the judgment device 55 retrieves the design data D corresponding to the identified box 1 from the recording unit 54. Furthermore, if the judgment device 55 identifies multiple identical boxes, or if there is no design data for the corresponding box, it sends an NG signal to the alarm device 60.

[0027] Furthermore, in step S001, the image data extraction unit 51 identifies regions 11 and 12 based on region identification information (region J11, region J12) read from barcode C. As shown in Figure 6, the determination device 55 then records the region identification information (J region 11, J region 12) of the identified regions 11 and 12, along with their position coordinates, in the recording unit 54.

[0028] Furthermore, the image data extraction unit 51 identifies equipment 1, equipment 2, and nameplates 11 and 12 based on the equipment identification information (J equipment 1, J equipment 2) and nameplate identification information (J nameplate 11, J nameplate 12) read from barcode C. Next, the determination device 55 records the identified device identification information (K1, K2) of device 1 and device 2, and the nameplate identification information (J nameplate 11, J nameplate 12) of nameplate 12, along with their position coordinates, in the recording unit 54.

[0029] In the case of a nameplate, the image data extraction unit 51 identifies and records the image of the displayed characters at a predetermined position using OCR (Optical Character Recognition). The OCR algorithm can be kNN (k-nearest neighbor algorithm), SVM (Support Vector Machine), or deep learning, but is not particularly limited to any specific algorithm. Furthermore, the inscriptions on the nameplates that have been read (RYOIKI11, RYOIKI12) are also recorded in the recording unit 54 in correspondence with the nameplate identification information (J nameplate 11, J nameplate 12). Furthermore, for boxes, areas, equipment, and nameplates to which barcode C could not be affixed or identified, unique information such as identification information is input from an external source via the input data acquisition unit 53 and recorded in the recording unit 54.

[0030] Next, the determination device 55 derives parent identification information (step S002). The parent identification number for each area will be the box itself. Furthermore, the location of the equipment and nameplates will be used to determine which area they are actually located within, and the parent identification number for that area will be assigned accordingly.

[0031] For example, in the case of the distribution panel 20R1 shown in Figure 5, device 1 has a y-coordinate of 700, which is between the y-coordinate of area 11 (900) and the y-coordinate of area 12 (600). Since the position of barcode C on device 1 is determined to be the upper left, it can be seen that device 1 is located in area 11. From this, it can be seen that the parent identification number of device 1 in the distribution panel 20R1 is area 11. The parent identification numbers of other devices and nameplates can be derived in the same way.

[0032] As shown in Figure 6, the identification information, parent identification information, and, in the case of a nameplate, the display characters for each component of the distribution panel 20R1, such as the box, area, equipment, and nameplate, are recorded in the recording unit 54.

[0033] Next, the decision device 55 performs a comparison with the design data D (step S003). When comparing design data D with the identification results in Figure 6, the same identification information should correspond to the same parent identification information and the same notation. If there are discrepancies, differences in parent identification numbers, or different notations, an inconsistency will occur, and the judgment device will send an NG signal to the alarm device. In the case of design data D shown in Figure 2 and extracted data shown in Figure 6, the box body 1 matches, so no NG signal is sent to the alarm device 60. Thus, the judgment device 55 sends an OK signal to the alarm device 60 and ends the inspection process of the distribution board 20R1 (step S004).

[0034] The following describes examples of cases where the matching process in the judgment device 55 fails. Figure 7 shows a schematic configuration of the power distribution panel 20R2 that the inspection device 100 of this embodiment performs inspection on. Figure 8 shows an example of the results obtained by the inspection device 100 of this embodiment, which identifies each component of the distribution board 20R2, including the box, area, equipment, and nameplate.

[0035] As shown in Figures 7 and 8, in the distribution panel 20R2, the installation locations of equipment 1 and equipment 2 differ from the area indicated by the parent identification number specified in the design data D shown in Figure 2, resulting in an inconsistency. In this case, the determination device 55 determines that the placement areas of equipment 1 and equipment 2 are reversed in area 11 and area 12, and sends an NG signal to the alarm device 60.

[0036] In this way, the inspection device 100 of this embodiment can affix barcodes C indicating individual identification information to the box, area, equipment, and nameplate, and by photographing the box, area, equipment, and nameplate together, it can perform a comparison with the design data D. In this way, it is possible to determine whether the equipment and nameplates in the distribution panel under inspection are located in the area specified in design data D, and if they are not located there, an alarm can be issued to warn the user.

[0037] In this embodiment, the distribution panel is shown as having two enclosures, each enclosure containing two or three compartments, and each compartment containing multiple pieces of equipment and nameplates; however, the embodiment is not limited to this. The number and combination of boxes, areas, and devices are flexible. While barcodes were used to identify the boxes, areas, and devices, anything that can identify them is acceptable, such as letters, symbols, color coding, specific shapes, or RF tags.

[0038] Furthermore, while barcodes were used as an identifier to show identification information, the system is not limited to them. For example, a tag that emits identification information via radio waves could be attached as an identifier, and the extraction unit could read these radio waves. Furthermore, while electrical and mechanical equipment and nameplates have been given as examples of components constituting a distribution panel, the panel is not limited to these. For example, identifiers may be assigned to connecting conductors such as cables that connect each piece of equipment as components to detect abnormalities in their arrangement.

[0039] The electrical device inspection apparatus of this embodiment, configured as described above, An electrical equipment inspection device that uses a control unit to detect abnormalities in the arrangement of one or more devices as components, in electrical equipment provided by arranging one or more devices as components in a region within one or more boxes as components, The control unit, As actual installation information for the aforementioned electrical equipment, An extraction unit reads from the electrical equipment the following: equipment identification information as unique identification information for the equipment, enclosure identification information as unique identification information for the enclosure, area identification information as unique identification information for the area, and actual position information for each of the components. The system includes a recording unit that pre-records the following as design data: the equipment identification information, the enclosure identification information, the area identification information, a first parent identification information specifying the enclosure to which the area belongs, and a second parent identification information specifying the area to which the equipment belongs. Using the actual installation information extracted by the extraction unit, the region identification information of the region in which the equipment is installed in the electrical equipment is derived. The derived area identification information in which the equipment is installed, Based on the first parent identification information and the second parent identification information of the device, which are pre-recorded in the design data, an abnormality in the placement of the device is detected. It is.

[0040] This allows, when inspecting electrical equipment, to determine whether the equipment, nameplates, etc., indicated on the design drawing are correctly placed in the area indicated on the design drawing by comparing them with design data created in advance based on the design drawing. This allows for accurate detection of abnormalities in the arrangement of components that make up electrical equipment, and enables faster detection of abnormalities compared to visual inspection, thereby reducing the labor required for inspection work.

[0041] Embodiment 2. The following describes Embodiment 2 of this book, focusing on the differences from Embodiment 1, using diagrams. Parts similar to those in Embodiment 1 are denoted by the same reference numerals and their descriptions are omitted. In Embodiment 2, unlike Embodiment 1, barcode C is not added to the area or device, and identification is performed by recognizing the photographic data, which is the captured image.

[0042] The configuration of the inspection device 100 in Embodiment 2 is the same as in Embodiment 1, but the barcode C is not affixed to the power distribution board to be inspected, and the processing content of the judgment device 55 is different. The following describes the decision-making device 55 and the learning process by the learning device, which is a preliminary step before the decision-making device 55 makes a decision. <Learning Phase>

[0043] Figure 9 shows a schematic configuration of the learning device 256 of the electrical equipment inspection device 100 according to Embodiment 2. Figure 10 shows an example of training data for the electrical equipment inspection device 100 according to Embodiment 2. Figure 11 shows the neural network of the electrical equipment inspection device 100 according to Embodiment 2. Figure 12 is a flowchart showing the learning process of the electrical equipment inspection device 100 according to Embodiment 2. Figure 13 is a block diagram showing the configuration of the judgment device of the electrical equipment inspection device 100 according to Embodiment 2. Figure 14 is a flowchart showing the processing of the inference unit 257T of the electrical equipment inspection device 100 according to Embodiment 2. Figure 15 shows the recognition image of the inference unit 257T of the electrical equipment inspection device 100 according to Embodiment 2. Figure 16 shows the recognition results of the inference unit 257T of the electrical equipment inspection device 100 according to Embodiment 2.

[0044] As shown in Figure 9, the control unit 50 of this embodiment includes a learning device 256. Note that in Figure 9, some of the components of the control unit 50 shown in Figure 1 are omitted from the illustration. The learning device 256 comprises a data acquisition unit 256I, a model generation unit 256G, and a trained model storage unit 256M. The learning device 256 may be located in the same place as the image data extraction unit 51, or it may be located in a different place from the image data extraction unit 51, for example, it may reside on a cloud server.

[0045] The data acquisition unit 256I acquires photographic data of each of the multiple areas, equipment, and nameplates (hereinafter collectively referred to as "objects") that constitute the components of the power distribution panel, as well as the names of the objects (hereinafter referred to as "name data") as identification information, as training data. Figure 10 shows an example of the acquired training data.

[0046] The model generation unit 256G trains based on training data created from a combination of photographic data and name data output from the data acquisition unit 256I. In other words, it generates a trained model from photographic data and name data of the target object.

[0047] The model generation unit 256G can use any known learning algorithm, such as supervised learning, unsupervised learning, or reinforcement learning. As an example, the case where a neural network is applied will be described.

[0048] The model generation unit 256G learns, for example, according to a neural network model, using so-called supervised learning. Here, supervised learning is a method in which pairs of input and result data are provided to the learning device 256, thereby learning features in the training data and inferring results from the input.

[0049] A neural network consists of an input layer made up of multiple neurons, an intermediate layer (hidden layer) made up of multiple neurons, and an output layer made up of multiple neurons. The intermediate layer can be one or more layers. For example, in a three-layer neural network as shown in Figure 11, when multiple inputs are input to the input layer (X1-X3), these values ​​are multiplied by weights W1 (w11-w16) and input to the hidden layer (Y1-Y2). The result is then multiplied again by weights W2 (w21-w26) and output from the output layer (Z1-Z3). This output varies depending on the values ​​of weights W1 and W2.

[0050] In this embodiment, the neural network learns through so-called supervised learning, according to training data created based on a combination of photographic data and name data acquired by the data acquisition unit 256I. In other words, a neural network learns by inputting photographic data into the input layer and adjusting the weights W1 and W2 so that the result output from the output layer approaches the name data. The model generation unit 256G generates and outputs a trained model by performing the training described above. The trained model storage unit 256M stores the trained model output from the model generation unit 256G.

[0051] Next, we will explain the learning process performed by the learning device 256 using Figure 12. In step b1, the data acquisition unit 256I acquires photographic data and name data as identification information. Although photographic data and name data are acquired simultaneously, it is sufficient if the photographic data and name data can be input in association, and the photographic data and name data may be acquired at different times.

[0052] In step b2, the model generation unit 256G generates a trained model by so-called supervised learning, according to the training data created based on the combination of photographic data and name data acquired by the data acquisition unit 256I. In step b3, the trained model storage unit 256M stores the trained model generated by the model generation unit 256G. In this way, a trained model is generated that outputs name data as identification information for each component that makes up the power distribution panel, based on the input image data.

[0053] <Utilization Phase> Figure 13 is a block diagram showing the configuration of the decision-making device 55 in this embodiment. The decision-making device 55 comprises a data acquisition unit 257I, an inference unit 257T, and a decision-making unit 257D. The data acquisition unit 257I acquires photographic data of the box.

[0054] The inference unit 257T infers the name data output obtained using the trained model. That is, by inputting the photographic data acquired by the data acquisition unit 256I into this trained model, it is possible to obtain the output of name data, which is the identification information of each component, inferred from the photographic data.

[0055] Next, using Figure 14, we will explain the process for obtaining name data output using the inference unit 257T. In step c1, the data acquisition unit 257I acquires photographic data.

[0056] In step c2, the inference unit 257T retrieves the trained model from the trained model storage unit 256M, inputs the photo data into the retrieved trained model, and obtains output of name data for multiple objects and coordinate data on the photo where they were recognized. Figure 13 shows an example of the recognition of box 1, and Figure 15 shows the recognition result obtained.

[0057] In step c3, the inference unit 257T outputs a combination of name data output and coordinate data output obtained by the trained model to the determination unit.

[0058] Next, using Figure 16, we will explain the process for determining whether or not to send an alarm to the alarm device 60 using the determination unit 257D. The determination unit 257D, similar to step S002 in Figure 4 shown in Embodiment 1, uses the combination of output name data and coordinate data to derive parent identification information indicating the attribute relationship between the area of ​​the target box and the equipment and nameplate. The method for determining parent-child relationships is the same as in Embodiment 1. Furthermore, the nameplate markings are identified using OCR, and this method is also the same as in Embodiment 1.

[0059] The resulting recognition data can be compared with the design data to determine whether they match or not. The method of determination is the same as in Embodiment 1. If there is a match, the determination device sends an NG signal to the alarm unit. In this embodiment, the identification information of the box body and objects that could not be recognized by the inference unit 257T are input via an input device.

[0060] In this embodiment, we have described the case where supervised learning is applied to the learning algorithm used by the model generation unit 256G, but this is not the only possible case. In addition to supervised learning, reinforcement learning, unsupervised learning, or semi-supervised learning can also be applied to the learning algorithm.

[0061] Furthermore, the learning algorithm used in the model generation unit 256G can be deep learning, which learns to extract the features themselves, or machine learning can be performed according to other known methods, such as genetic programming, inductive logic programming, or support vector machines.

[0062] Therefore, according to this second embodiment, in addition to the effects of the first embodiment, each component can be recognized by pre-training photographic data of the area, equipment, and nameplate without attaching a barcode. This improves the accuracy of inspection and reduces the labor involved in inspection. Furthermore, regarding the area, instead of pre-training photographic data, the area may be identified from photographic data of the box body during the utilization phase using image processing to determine edge features such as the density of the image at the boundary of the area.

[0063] While this disclosure describes various exemplary embodiments and examples, the various features, aspects, and functions described in one or more embodiments are not limited to the application of a particular embodiment, but are applicable individually or in various combinations to the embodiments. Accordingly, countless variations not illustrated are conceivable within the scope of the art disclosed in this specification. These include, for example, modifying, adding or omitting at least one component, or even extracting at least one component and combining it with components of other embodiments.

[0064] The various aspects of this disclosure are summarized below as an appendix.

[0065] (Note 1) An electrical equipment inspection device that uses a control unit to detect abnormalities in the arrangement of one or more devices as components, in electrical equipment provided by arranging one or more devices as components in a region within one or more boxes as components, The control unit, As actual installation information for the aforementioned electrical equipment, An extraction unit reads from the electrical equipment the following: equipment identification information as unique identification information for the equipment, enclosure identification information as unique identification information for the enclosure, area identification information as unique identification information for the area, and actual position information for each of the components. The system includes a recording unit that pre-records the following as design data: the equipment identification information, the enclosure identification information, the area identification information, a first parent identification information specifying the enclosure to which the area belongs, and a second parent identification information specifying the area to which the equipment belongs. Using the actual installation information extracted by the extraction unit, the region identification information of the region in which the equipment is installed in the electrical equipment is derived. The derived area identification information in which the equipment is installed, Based on the first parent identification information and the second parent identification information of the device, which are pre-recorded in the design data, an abnormality in the placement of the device is detected. An inspection device for electrical equipment. (Note 2) The electrical equipment is provided with one or more nameplates as components on which nameplate information of the electrical equipment is displayed, The extraction unit uses the following as the actual installation information: The nameplate identification information, which is the unique identification information of the nameplate, and the actual position information of the nameplate are read from the electrical equipment. The design data includes third parent identification information that specifies the area to which the nameplate should belong, The control unit, Based on the actual installation information extracted by the extraction unit, the area identification information for the area where the nameplate is installed in the electrical equipment is derived. The derived area identification information on which the nameplate is installed, Based on the first parent identification information and the third parent identification information recorded in the design data, The system detects abnormalities in the placement of the aforementioned nameplate. An inspection device for electrical equipment as described in Appendix 1. (Note 3) The control unit, From the actual position information of the region, the equipment, and the nameplate extracted by the extraction unit, the identification information of the box to which the region belongs, the region where the equipment and the nameplate are arranged is derived. The derived identification information for each of the box and the region, Based on the first parent identification information of the region, the second parent identification information of the device, and the third parent identification information of the nameplate, which are recorded in advance in the design data, The system detects abnormalities in the arrangement of the aforementioned equipment and the aforementioned nameplate. An inspection device for electrical equipment as described in Appendix 2. (Note 4) Each of the above components has an identifier that indicates the identification information, The extraction unit reads each of the identification pieces indicated by the identifier. An inspection device for electrical equipment as described in any one of the appendices 1 to 3. (Note 5) The extraction unit captures an image of each component and reads the identification information of each component from the captured image. An electrical equipment inspection device as described in any one of the items from Appendix 1 to Appendix 4. (Note 6) The control unit, A trained model for inferring the identification information of each component from the images of each component within the electrical equipment, The system includes an inference unit that outputs the identification information of each component as the actual installation information of the electrical equipment from the captured image, An inspection device for electrical equipment as described in Appendix 5. (Note 7) The aforementioned electrical equipment is a distribution panel that distributes the received power to multiple loads. An electrical equipment inspection device as described in any one of the appendices 1 through 6. [Explanation of symbols]

[0066] 50 Control unit, 51 Image data extraction unit (extraction unit), 54 Recording unit, 257T Inference Unit, 20, 20R1, 20R2 Distribution Panel, D Design Data, 100. Inspection device for electrical equipment.

Claims

1. An electrical equipment inspection device that uses a control unit to detect abnormalities in the arrangement of one or more devices as components, in electrical equipment provided by arranging one or more devices as components in a region within one or more boxes as components, The control unit, As actual installation information for the aforementioned electrical equipment, An extraction unit reads from the electrical equipment the following: equipment identification information as unique identification information for the equipment, enclosure identification information as unique identification information for the enclosure, area identification information as unique identification information for the area, and actual position information for each of the components. The system includes a recording unit that pre-records the following as design data: the equipment identification information, the enclosure identification information, the area identification information, a first parent identification information specifying the enclosure to which the area should belong, and a second parent identification information specifying the area to which the equipment should belong. Using the actual installation information extracted by the extraction unit, the region identification information of the region in which the equipment is installed in the electrical equipment is derived. The derived area identification information in which the equipment is installed, Based on the first parent identification information and the second parent identification information of the device, which are recorded in advance in the design data, an abnormality in the placement of the device is detected. An inspection device for electrical equipment.

2. The electrical equipment is provided with one or more nameplates as components on which nameplate information of the electrical equipment is displayed, The extraction unit uses the following as the actual installation information: The nameplate identification information, which is the unique identification information of the nameplate, and the actual position information of the nameplate are read from the electrical equipment. The design data includes third parent identification information that specifies the area to which the nameplate should belong, The control unit, Based on the actual installation information extracted by the extraction unit, the area identification information for the area where the nameplate is installed in the electrical equipment is derived. The derived area identification information on which the nameplate is installed, Based on the first parent identification information and the third parent identification information recorded in the design data, The system detects abnormalities in the placement of the aforementioned nameplate. An inspection device for electrical equipment according to claim 1.

3. The control unit, From the actual position information of the region, the equipment, and the nameplate extracted by the extraction unit, the identification information of the box to which the region belongs, the region where the equipment and the nameplate are arranged is derived. The derived identification information for each of the box and the region, Based on the first parent identification information of the region, the second parent identification information of the device, and the third parent identification information of the nameplate, which are recorded in advance in the design data, The system detects abnormalities in the arrangement of the aforementioned equipment and the aforementioned nameplate. An inspection device for electrical equipment according to claim 2.

4. Each of the above components has an identifier that indicates the identification information, The extraction unit reads each of the identification pieces indicated by the identifier. An inspection device for electrical equipment as described in claim 3.

5. The extraction unit captures an image of each component and reads the identification information of each component from the captured image. An inspection device for electrical equipment as described in claim 3.

6. The control unit, A trained model for inferring the identification information of each component from the images of each component within the electrical equipment, The system includes an inference unit that outputs the identification information of each component as the actual installation information of the electrical equipment from the captured image, An inspection device for electrical equipment according to claim 5.

7. The aforementioned electrical equipment is a distribution panel that distributes the received power to multiple loads. An inspection device for electrical equipment according to any one of claims 1 to 6.

Citation Information

Patent Citations

  • Abnormal product determination method

    JP2020173496A