Program, information processing method, and information processing device

Abstract

To efficiently output information relating to an internal condition of a resin member used in a power transmission / distribution facility.SOLUTION: A program causes a computer to execute the processing of: obtaining an image captured in a state where resin member used in a power transmission / distribution facility is irradiated with transmitted light; inputting the obtained image including the resin member into a learned model that has learned to output information relating to an internal condition of the resin member when an image including resin member that has been irradiated with transmitted light is input, thereby obtaining information relating to an internal condition of the resin member from the learned model; and outputting the obtained information relating to the internal condition of the resin member.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a program, an information processing method, and an information processing device. [Background technology]

[0002] For example, Patent Document 1 describes matters relating to electrical equipment such as transformers, cables, circuit breakers, and capacitors, and further refers to a method for evaluating the deterioration of electrical insulating oil used in these electrical equipment. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-65858 Summary of the Invention [Problem to be solved by the invention]

[0004] However, the evaluation method disclosed in Patent Document 1 does not take into consideration the efficient output of information relating to the internal state of resin members used in power transmission and distribution facilities.

[0005] The present invention has been made in consideration of the above circumstances, and its purpose is to provide a program or the like that can efficiently output information regarding the internal state of resin components used in power transmission and distribution equipment. [Means for solving the problem]

[0006] A program according to one aspect of the present disclosure causes a computer to acquire an image captured when a resin member used in power transmission and distribution equipment is irradiated with transmitted light, and input the acquired image including the resin member into a learning model that has been trained to output information regarding the internal state of the resin member when an image including the resin member irradiated with transmitted light is input, thereby acquiring information regarding the internal state of the resin member from the learning model and outputting the acquired information regarding the internal state of the resin member.

[0007] In this embodiment, for example, resin members are used as various insulating materials in power transmission and distribution facilities such as transformers. When the resin members are manufactured as injection-molded products using a resin mold, it is expected that air bubbles (voids) will occur inside the molded product (resin member). The resin member of the power transmission and distribution facility is placed under a light source that irradiates transmitted light, such as visible light, and is imaged by an imaging device located on the opposite side of the light source. By arranging the resin member between the light source and the imaging device in this manner, a portion of the light irradiated from the light source passes through the resin member, thereby enabling the resin member to be imaged in a state where the transmitted light is irradiated. In this case, the optical axis (imaging direction) of the imaging device and the optical axis (illumination direction) of the light source may form a predetermined imaging angle, for example, of approximately 5° to 60°. A control unit of the information processing device executes a program stored in a storage unit to acquire the image captured by the imaging device (an image of the resin member irradiated with transmitted light). The information processing device and the imaging device may be connected to each other via a wired or wireless connection, and the control unit of the information processing device may acquire the image by communicating with the imaging device. Alternatively, the control unit of the information processing device may acquire the image captured by the imaging device offline via a storage medium such as a USB memory or an SD card. The control unit of the information processing device inputs the acquired image (an image of a resin member irradiated with transmitted light) to a learning model (void detection model). The learning model is trained to output information regarding the internal state of a resin member when an image including the resin member irradiated with transmitted light is input. If an air bubble (void) is present inside the resin member in the image of the resin member irradiated with transmitted light, the air bubble (void) is imaged in a lighter color than other portions, i.e., portions without air bubbles. That is, if the resin member is generally yellow, the portion with the air bubble (void) is imaged as a shadow in a light yellow color, which is a lighter color than the yellow. The bubbles (voids) imaged in a pale yellow shade can be visually recognized over a wide range, such as elongated or wide, depending on the shape of the bubble.Furthermore, when viewed from the opposite side of the resin component, the air bubbles (voids) appear as shadows at the same position. That is, the shadows can be seen at the same position when viewed from one side (the imaging device side) and the other side (the light source side). The information (information about the internal state of the resin component) output by the learning model (void detection model) trained in this way includes, for example, information about the presence or absence of air bubbles (voids). If air bubbles (voids) are present in the resin component, the learning model may indicate (identify) the position and size of the air bubbles (voids) in the image, for example, by superimposing a bounding box on the image. Alternatively, the information output by the learning model (void detection model) may be a judgment result (pass / fail judgment) indicating whether the resin component is normal (good) or abnormal (defective) based on the information about the presence or absence of air bubbles (voids). The control unit of the information processing device acquires the information (information about the internal state of the resin component) output by the learning model and outputs the acquired information to, for example, a display unit such as a monitor, or an information terminal such as a smartphone connected for communication. By using this learning model, it is possible to efficiently determine the presence or absence of bubbles (voids) formed inside various resin materials used, for example, as insulating materials in power transmission and distribution facilities. This makes it possible to shorten the inspection time, eliminate concerns about X-ray exposure for inspectors, reduce the cost of the inspection equipment, and reduce the cost of ensuring redundancy that accompanies this cost reduction, compared to when an X-ray CT scanner is used to inspect the presence or absence of bubbles (voids).

[0008] In a program according to one aspect of the present disclosure, information regarding the internal state of the resin member includes information regarding the presence or absence of air bubbles formed inside the resin member, and if air bubbles are present inside the resin member, it is determined that the resin member is unusable, and if no air bubbles are present inside the resin member, it is determined that the resin member is usable.

[0009] In this aspect, the information output by the learning model (information regarding the internal state of the resin component) includes information regarding the presence or absence of bubbles (voids) formed within the resin component, and functions as a void detection model. If the information output by the learning model indicates the presence of bubbles (voids) within the resin component, the control unit of the information processing device derives and outputs a message indicating that the resin component is unusable. If the information output by the learning model indicates the absence of bubbles (voids) within the resin component, the control unit of the information processing device derives and outputs a message indicating that the resin component is usable. Alternatively, if the information output by the learning model indicates that the resin component is normal (no voids) or abnormal (voids present) based on the presence or absence of bubbles (voids) formed within the resin component, the control unit of the information processing device may use the output results from the learning model to derive and output a message indicating that the resin component is usable (normal: no voids) or unusable (abnormal: voids present). Since the information output by the learning model includes information on the presence or absence of bubbles (voids) formed inside the resin component, inspection during the manufacturing process of the resin component can be performed efficiently based on the presence or absence of the bubbles (voids). Therefore, production efficiency can be improved in a resin component production method that includes an inspection process for the presence or absence of bubbles (voids).

[0010] In a program according to one aspect of the present disclosure, the resin member is a coil bobbin used in a power transmission and distribution transformer, and the image includes a side wall or a partition wall of the coil bobbin.

[0011] In this aspect, the resin member is a coil bobbin used in a transformer installed in a power transmission line, and is molded from, for example, polyamide resin. The coil bobbin is used for the primary coil or secondary coil of the transformer, and the windings that constitute these coils are wound around it. The coil bobbin has partition walls (wing portions, partition plates) that divide the space of the recessed groove portion for the winding in the width direction. The partition walls are formed in an annular plate shape that protrudes radially from the outer circumferential surface of the cylindrical coil bobbin and function as partition plates that separate the windings wound around the outer circumferential surface. Among the multiple partition walls formed in this annular plate shape, the wall surfaces of adjacent partition walls are arranged so as to be parallel to each other. The multiple partition walls may have different thicknesses, for example, approximately 3 mm to 10 mm. Side walls are provided at each axial end of the cylindrical coil bobbin to sandwich one or more of the partition walls. That is, one or more partition walls are interposed between two side walls provided at both axial ends of a cylindrical coil bobbin. A cutout may be formed in the side wall, and a partition wall adjacent to the side wall may be exposed through the cutout. In this case, the partition wall adjacent to the side wall may have a thickness greater (thicker) than the other partition walls. The image of the resin member irradiated with transmitted light may include these plate-shaped side walls or partition walls. That is, transmitted light may pass through these plate-shaped side walls or partition walls. In this case, if the side walls or partition walls have different thicknesses, the image of the resin member irradiated with transmitted light may include the wall (side wall or partition wall) with the largest thickness. In this way, by inputting an image including a plate-shaped side wall or partition wall into a learning model, it is possible to efficiently detect voids formed inside the partition walls (vanes, partition plates) of the coil bobbin in a transformer installed in a power transmission line.

[0012] A program according to one aspect of the present disclosure corrects the acquired image according to the imaging angle at which the image was captured, and inputs the corrected image into the learning model.

[0013] In this aspect, when capturing an image of a resin member such as a coil bobbin illuminated with transmitted light from a light source, it is assumed that the optical axis (imaging direction) of the imaging device and the optical axis (illumination direction) of the light source form a predetermined imaging angle of, for example, approximately 5° to 60°. Even in such a case, the control unit of the information processing device performs angle correction, such as tilt correction, keystone correction, or distortion correction, on the image acquired from the imaging device according to the imaging angle. The control unit of the information processing device inputs the image (corrected image) that has been subjected to angle correction according to the imaging angle into the learning model. By inputting the image (corrected image) that has been subjected to angle correction into the learning model, the inference accuracy of the learning model can be improved.

[0014] In the program according to one aspect of the present disclosure, the light source of the transmitted light is a white LED.

[0015] In this embodiment, the light source of the transmitted light is a white LED, which emits white light having a color temperature of 5500 K and an illuminance of 2800 lux as transmitted light, for example. The power source for driving the white LED may be, for example, a DC power supply with an output of approximately 10 W. In this case, the control unit of the information processing device may increase or decrease the illuminance of the white light emitted from the white LED by increasing or decreasing the output wattage of the DC power supply. By using a white LED that emits white light (natural light) with a relatively high illuminance as the light source of the transmitted light, the white light can be transmitted through a plate-like wall portion, such as a side wall or a partition wall, of a resin member such as a coil bobbin, and an image of an area showing voids formed inside the wall portion can be efficiently captured.

[0016] A program according to one aspect of the present disclosure derives a physical quantity of the transmitted light based on the acquired image of the resin member, and performs processing to irradiate the transmitted light with the derived physical quantity.

[0017] In this aspect, the control unit of the information processing device identifies the type of object (inspection target) to be inspected based on the appearance, such as the shape, material, or color, of the resin member, based on an image of the resin member (e.g., a coil bobbin) acquired from the imaging device. The control unit of the information processing device may use edge detection, pattern recognition, or an object recognition model, such as YOLO, on the acquired image to identify the object. The control unit of the information processing device identifies the type of object to be inspected for the presence or absence of bubbles (voids), based on the resin member (subject) included in the image, and derives physical quantities, such as color temperature, color wavelength (frequency), illuminance, or luminance, of transmitted light according to the identified type of object. The derived physical quantities may be a single type of physical quantity or a combination of multiple types of physical quantities. The storage unit of the information processing device may store definition information (an irradiation table) that defines physical quantities corresponding to each type of object (resin member), and the control unit of the information processing device may derive the physical quantities of transmitted light according to the type of object (resin member) by referring to the irradiation table. The control unit of the information processing device performs processing to cause the light source, such as a white LED, to irradiate transmitted light at the derived physical quantity. When the light source (LED light), such as a white LED, is connected to a drive power source, such as a DC power supply, the control unit of the information processing device may increase or decrease the illuminance of the transmitted light by increasing or decreasing the output wattage of the DC power supply. When the light source (LED light), such as a white LED, is a light source device with color temperature variable control, the control unit of the information processing device may perform color temperature variable control by outputting a control signal to the light source device, thereby increasing or decreasing the color temperature of the transmitted light. By identifying the type of inspection object based on the resin member (subject) included in the image in this way and irradiating transmitted light at a physical quantity corresponding to the identified type of inspection object, the inspection process for bubbles (voids) can be automated, thereby improving the efficiency of the entire production process, including the inspection.

[0018] An information processing method according to one aspect of the present disclosure causes a computer to acquire an image of a resin member used in power transmission and distribution equipment while irradiating it with transmitted light, input the acquired image including the resin member into a learning model that has been trained to output information regarding the internal state of the resin member when an image including the resin member irradiated with transmitted light is input, and acquire information regarding the internal state of the resin member from the learning model, thereby executing a process to output the acquired information regarding the internal state of the resin member.

[0019] In this aspect, it is possible to provide an information processing method that can efficiently output information relating to the internal state of a resin member used in a power transmission and distribution facility.

[0020] An information processing device according to one aspect of the present disclosure includes an image acquisition unit that acquires an image captured when a resin member used in a power transmission and distribution facility is irradiated with transmitted light, an information acquisition unit that acquires information regarding the internal state of the resin member from a learning model that has been trained to output information regarding the internal state of the resin member when an image including the resin member irradiated with transmitted light is input, by inputting the acquired image including the resin member into the learning model, and an output unit that outputs the acquired information regarding the internal state of the resin member.

[0021] In this aspect, it is possible to provide an information processing device that can efficiently output information relating to the internal state of a resin member used in a power transmission and distribution facility. [Effects of the Invention]

[0022] It is possible to provide a program or the like that efficiently outputs information regarding the internal state of resin members used in power transmission and distribution facilities. [Brief explanation of the drawings]

[0023] [Figure 1] 1 is a schematic diagram for explaining an overview of a coil bobbin determination system according to a first embodiment. [Figure 2] FIG. 1 is a block diagram showing a configuration of an information processing device. [Figure 3] FIG. 10 is an explanatory diagram regarding the generation process of a learning model (void detection model). [Figure 4] 10 is a flowchart illustrating a processing procedure performed by a control unit of an information processing device. [Figure 5] FIG. 10 is a schematic diagram for explaining an overview of a coil bobbin determination system according to a second embodiment (automation). [Figure 6] FIG. 2 is an explanatory diagram illustrating an irradiation table. [Figure 7] 10 is a flowchart illustrating a processing procedure performed by a control unit of an information processing device. DETAILED DESCRIPTION OF THE INVENTION

[0024] (Embodiment 1) Hereinafter, embodiments will be described with reference to the drawings. Fig. 1 is a schematic diagram for explaining an overview of a coil bobbin determination system S according to embodiment 1. Fig. 2 is a block diagram showing the configuration of an information processing device 1. The coil bobbin determination system S is configured with the information processing device 1 as a main device, and the information processing device 1 acquires an image of a coil bobbin B captured by an imaging device C.

[0025] The coil bobbin B is an insulating member used in a transformer provided in a power transmission line, and corresponds to an example of a resin member used in a power transmission and distribution facility. In this embodiment, the resin member used in the power transmission and distribution facility will be described based on the coil bobbin B used in a transformer, but the resin member is not limited thereto, and may be, for example, a resin insulating member such as a bushing, a joint, or the like provided in various types of power transmission and distribution facility.

[0026] The coil bobbin B is molded from, for example, polyamide resin, and is used as the primary or secondary coil of a transformer, with the windings that make up these coils wound around it. The coil bobbin B has a recessed groove space for winding the winding, and is provided with dividing walls (wing portions, partition plates) that divide the recessed groove space in the width direction. Furthermore, side walls are provided on both ends of the coil bobbin B so as to sandwich these dividing walls. Injection-molded products manufactured using a resin mold, such as the coil bobbin B, are expected to have air bubbles (voids) inside the molded product (resin member), and there is a particular concern that air bubbles (voids) are more likely to occur inside plate-shaped walls (dividing walls, side walls).

[0027] By irradiating such plate-shaped wall portions (dividing walls, side walls) with relatively high illuminance white light from a light source device 2 such as a white LED, the white light can be transmitted through the wall portions of the coil bobbin B. The imaging device C captures an image of the coil bobbin B so as to include the coil bobbin B, particularly the wall portions (dividing walls, side walls) through which the white light (transmitted light) has passed, and outputs the captured image (image data) to the information processing device 1. As will be described in detail later, the optical axis of the light source device 2 and the optical axis of the imaging device C are offset, and a predetermined imaging angle is formed due to the difference between these optical axes. The information processing device 1 may perform angle correction on the image acquired from the imaging device C in accordance with the imaging angle.

[0028] An image of the coil bobbin B captured under irradiation with transmitted light includes the wall portions (dividing walls, side walls) of the coil bobbin B. When the wall portions (dividing walls, side walls) are the inspection target, if there are bubbles (voids) inside the wall portions (in the case of an abnormality), the bubbles (voids) are photographed as shadows. If there are no bubbles (voids) inside the wall portions (in the case of a normal state), the shadow-like bubbles (voids) are not photographed, and the image becomes monochromatic, reflecting the color of the wall portions. The information processing device 1 acquires the images thus photographed by the imaging device C and stores them in the memory unit 12.

[0029] The information processing device 1 inputs an image of the coil bobbin B acquired from the imaging device C or an angle-corrected image into the learning model 101, thereby determining whether or not air bubbles (voids) exist inside the coil bobbin B (presence or absence of voids). The information processing device 1 outputs the determination result to a display device such as a display, or to an information terminal T such as a smartphone that is communicably connected, thereby notifying an inspector who performs quality inspection of the coil bobbin B in the production process of the coil bobbin B of information regarding whether or not the coil bobbin B is usable.

[0030] The imaging device C is, for example, a CMOS camera or the like, and captures an image of the coil bobbin B under a visible light source such as a white LED. The imaging device C may be, for example, a camera built into an information terminal T such as a smartphone carried by an inspector of the coil bobbin B, or may be connected via the input / output I / F 14 of the information processing device 1.

[0031] The information terminal T of an inspector of the coil bobbin B and the information processing device 1 are communicatively connected via an external network such as the Internet, a LAN, etc. The image of the coil bobbin B captured by the imaging device C is acquired by the information processing device 1 by transmitting it from the information terminal T of the inspector, acquiring it via the input / output I / F 14 of the information processing device 1, or acquiring it offline via a storage medium such as a memory card.

[0032] The light source device 2 is an LED light using a white LED. The physical quantity of the white light emitted from the light source device 2 may be, for example, a color temperature of 5500K and an illuminance of 2800 lux.

[0033] The control unit 11 has an arithmetic processing device with a timing function, such as one or more central processing units (CPUs), micro-processing units (MPUs), graphics processing units (GPUs), etc., and performs various information processing, control processing, etc. by reading and executing a program P (program product) stored in the storage unit 12. By executing the program P, the control unit 11 operates as functional units corresponding to each process, such as an acquisition unit (image acquisition unit, information acquisition unit) that acquires information (estimation results) from an image or a learning model 101, an output unit that outputs information regarding the internal state of the resin member, and a derivation unit that derives a determination result regarding the usability of the resin member.

[0034] The storage unit 12 includes a volatile storage area such as a static random access memory (SRAM), a dynamic random access memory (DRAM), or a flash memory, and a non-volatile storage area such as an EEPROM or a hard disk. The storage unit 12 pre-stores a program P (program product) and data referenced during processing, and further stores various data including image data acquired during processing and intermediate data generated during processing. The program P (program product) stored in the storage unit 12 may be a program P (program product) read from a recording medium M readable by the information processing device 1. Alternatively, the program P (program product) may be downloaded from an external computer (not shown) connected to a communication network (not shown) and stored in the storage unit 12.

[0035] The communication unit 13 is a communication module or communication interface for communicating with an information terminal T or the like by wired means such as Ethernet (registered trademark) or wirelessly, and is, for example, a short-range wireless communication module such as Wi-Fi (registered trademark) or Bluetooth (registered trademark), or a wide-area wireless communication module such as 4G or 5G. The control unit 11 communicates with an information terminal T such as a smartphone carried by an inspector of the coil bobbin B or the like via an external network such as the Internet or a LAN, via the communication unit 13.

[0036] The input / output I / F 14 is a communication interface that complies with a communication standard such as RS232C or USB. For example, an imaging device C is connected to the input / output I / F 14, and an input device such as a keyboard or a display device such as a liquid crystal display may also be connected to the input / output I / F 14.

[0037] FIG. 3 is an explanatory diagram regarding the generation process of the learning model 101 (void detection model). The control unit 11 of the information processing device 1 uses training data for the learning model 101 to train a neural network, such as a CNN, and generates the learning model 101 (void detection model) that receives an image including a resin member (coil bobbin B) irradiated with transmitted light as input and outputs information regarding the internal state of the resin member. The image including the resin member irradiated with transmitted light may be an image that has been angle-corrected according to the shooting angle of the imaging device C. The information regarding the internal state of the resin member output by the learning model 101 may include, for example, whether or not air bubbles (voids) exist inside the resin member. Alternatively, the information regarding the internal state of the resin member output by the learning model 101 may be information indicating a determination result of whether the resin member is normal or abnormal depending on the presence or absence of air bubbles (voids) inside the resin member.

[0038] The training data includes question data and answer data, and an image of a resin member illuminated with transmitted light corresponds to the question data, and the result of determining whether the resin member is normal or abnormal depending on the presence or absence of air bubbles (voids) inside the resin member corresponds to the answer data. The data set of question data and answer data included in the training data for learning the learning model 101 is synonymous with the data set of input data and output data when using the learning model 101 (void detection model), and if it is defined in one data set, it naturally applies to the other data set as well.

[0039] The neural network (learning model 101) trained using training data is expected to be used as a program module that is part of artificial intelligence software. The learning model 101 (void detection model) is used in the information processing device 1 that includes the control unit 11 (CPU, etc.) and memory unit 12 as described above, and is executed by the information processing device 1 having such calculation processing capabilities to form a neural network system. That is, the control unit 11 of the information processing device 1 performs calculations to extract feature amounts of an image including a resin member irradiated with transmitted light that is input to the input layer in accordance with instructions from the learning model 101 stored in the memory unit 12, and outputs information regarding the internal state of the resin member from the output layer.

[0040] The learning model 101 (void detection model) is configured, for example, by a convolutional neural network (CNN) and includes an input layer that receives input of an image containing a resin component irradiated with transmitted light, an intermediate layer that extracts features of the image containing the resin component irradiated with transmitted light, and an output layer that outputs information related to the internal state of the resin component. The input layer includes multiple neurons that receive input of an image containing the resin component irradiated with transmitted light and passes the input values to the intermediate layer. The intermediate layer is defined using an activation function such as a ReLU function or a sigmoid function and includes multiple neurons that extract features of each input value and passes the extracted features to the output layer. Parameters such as weighting coefficients and bias values of the activation function are optimized using an error backpropagation algorithm. The output layer is configured, for example, by a fully connected layer and outputs information related to the internal state of the resin component based on the features output from the intermediate layer. Alternatively, the output layer may be, for example, composed of a softmax layer, and may output information regarding the internal state of the resin part, such as the probability of the presence or absence of air bubbles (voids), or the probability that coil bobbin B (resin part) is usable or unusable.

[0041] In the present embodiment, the learning model 101 is a CNN or the like. However, the learning model 101 is not limited to this. It may be constructed using other machine learning algorithms, such as object detection neural networks (e.g., R-CNN, YOLO), transformers, BERT, GPT, recurrent neural networks (RNNs), long-short term models (LSTMs), support vector machines (SVMs), Bayesian networks, linear regression, regression trees, multiple regression, random forests, and ensembles. Alternatively, the learning model 101 may be constructed using an artificial intelligence chatbot (e.g., ChatGPT). In this case, ChatGPT may be fine-tuned to efficiently output information about the internal state of the resin component. Alternatively, a question generated using an external database (e.g., WebDB) may be input to a prompt, which is the input interface of ChatGPT, along with an image of the resin component irradiated with transmitted light.

[0042] 4 is a flowchart illustrating a processing procedure performed by the control unit 11 of the information processing device 1. The control unit 11 of the information processing device 1 receives an operation from an operator using, for example, a keyboard connected to the input / output I / F 14, and performs the following processing based on the received operation.

[0043] The control unit 11 of the information processing device 1 acquires an image of a resin member used in a power transmission and distribution facility, which is irradiated with transmitted light (S101). For example, a resin member such as a coil bobbin B is placed under irradiation with white light from a light source device 2 (LED light) such as a white LED. Regarding the physical quantities of the white light from the light source device 2 (LED light), for example, the color temperature is 5500K and the illuminance is 2800 lux.

[0044] The coil bobbin B, which is molded from polyamide resin, is provided with a plurality of plate-like wall portions such as dividing walls and side walls, and the thickness of these plate-like wall portions is, for example, about 2 mm to 10 mm. The white light irradiated from the light source device 2 has an illuminance sufficient to pass through the walls (dividing walls, side walls) of the coil bobbin B, and corresponds to transmitted light that passes through the walls.

[0045] The imaging device C, which is configured with a CMOS camera or the like, is disposed on the opposite side of the coil bobbin B from the light source device 2, and captures an image of the wall portion (dividing wall, side wall) of the coil bobbin B with white light (transmitted light) passing through it. The image captured by the imaging device C may include the dividing wall or side wall closest to the light source device 2. In this case, the inspection target portion of the coil bobbin B may be the wall portion (dividing wall, side wall) closest to the light source device 2. Alternatively, the coil bobbin B may be fixed, and the light source device 2 may be brought close to the inspection target portion of the coil bobbin B. In other words, the light source device 2 may be configured to be movable relative to the coil bobbin B, which is the subject of the inspection. By making the position of the light source device 2 movable in this way, it is possible to increase the number of applicable locations on the coil bobbin B that are the inspection target portion. In this way, the wall portion that is the inspection target portion of the coil bobbin B may be thicker (having a larger plate thickness) than the other wall portions. The imaging device C and the information processing device 1 are connected so as to be able to communicate with each other, and the control unit 11 of the information processing device 1 acquires the image (image data) that the imaging device C captures and outputs.

[0046] The control unit 11 of the information processing device 1 performs angle correction on the acquired image (S102). The axial direction of the cylindrical coil bobbin B and the optical axis indicating the irradiation direction of the white light from the light source device 2 are generally aligned, whereas the axial direction of the coil bobbin B and the optical axis indicating the imaging direction of the imaging device C are misaligned. That is, the imaging device C is positioned with respect to the coil bobbin B and the light source device 2 so that the optical axis (imaging direction) of the imaging device C and the optical axis (irradiation direction) of the light source form a predetermined imaging angle, for example, of approximately 5° to 60°. When the coil bobbin B is imaged at such an imaging angle, the coil bobbin B included in the image output from the imaging device C has a tilt or distortion according to the imaging angle. In response to this, the control unit 11 of the information processing device 1 performs angle correction such as tilt correction, keystone correction, or distortion correction according to the imaging angle.

[0047] The control unit 11 of the information processing device 1 inputs the angle-corrected image to the learning model 101 (S103). The control unit 11 of the information processing device 1 acquires information about the internal state of the resin member from the learning model 101 (S104). The control unit 11 of the information processing device 1 inputs the angle-corrected image to the learning model 101, thereby acquiring information about the internal state of the coil bobbin B (resin member) from the learning model 101.

[0048] The information output by the learning model 101 may include, for example, whether or not air bubbles (voids) are present inside the wall portions (dividing walls, side walls) of the coil bobbin B. Alternatively, the information output by the learning model 101 may include a determination result as to whether the coil bobbin B is normal or abnormal, depending on the presence or absence of air bubbles (voids) inside the wall portions (dividing walls, side walls) of the coil bobbin B. The learning model 101, which has been trained to output information regarding the internal state of the coil bobbin B when an image including the coil bobbin B irradiated with transmitted light is input, outputs abnormality if air bubbles (voids) are present inside the coil bobbin B, and outputs normality if no air bubbles (voids) are present inside the coil bobbin B.

[0049] The control unit 11 of the information processing device 1 outputs a determination result regarding whether the resin member is usable or not based on the acquired information regarding the internal state of the resin member (S105). The control unit 11 of the information processing device 1 outputs a determination result (usable or not) indicating whether the coil bobbin B is normal (good) or abnormal (defective) based on the information regarding the internal state of the coil bobbin B acquired from the learning model 101.

[0050] When the learning model 101 outputs information regarding the presence or absence of air bubbles (voids) inside the coil bobbin B, the control unit 11 of the information processing device 1 outputs an abnormality if air bubbles (voids) are present inside the coil bobbin B, and outputs a normality if no air bubbles (voids) are present inside the coil bobbin B. When the learning model 101 outputs a determination result as to whether the coil bobbin B is normal or abnormal depending on the presence or absence of air bubbles (voids) inside the coil bobbin B, the control unit 11 of the information processing device 1 may output the determination result by the learning model 101 itself. The control unit 11 of the information processing device 1 can notify the coil bobbin B inspector of the determination result by outputting the determination result as to whether the coil bobbin B is normal or abnormal, for example, to a display device such as a display, or to an information terminal T connected for communication.

[0051] (Embodiment 2) 5 is a schematic diagram for explaining an overview of a coil bobbin determination system S according to a second embodiment (automation). The coil bobbin determination system S includes an information processing device 1, an imaging device C, and a light source device 2, similar to the first embodiment. The light source device 2 is further connected to a driving power source 3, such as a DC power supply, whose output wattage can be variably controlled. The information processing device 1 is communicably connected to the light source device 2 and the driving power source 3, for example, via an input / output I / F 14, and outputs various control signals to the light source device 2 and the driving power source 3, thereby performing control to change physical quantities such as the color temperature, color wavelength, or illuminance of the transmitted light irradiated from the light source device 2.

[0052] 6 is an explanatory diagram illustrating an example of an irradiation table. The storage unit 12 of the information processing device stores definition information in the form of, for example, a table (irradiation table), in which physical quantities corresponding to each type of object to be inspected (resin member) are defined. The irradiation table includes, as management items (fields), items indicating the types of physical quantities, such as the object to be inspected, color temperature, color wavelength, and illuminance.

[0053] The management items of the object to be inspected may store, for example, the type of insulating part that is the object to be inspected. Alternatively, if the object to be inspected is a coil bobbin B, classification information such as a model or part number according to the shape, size, and thickness of the wall parts (dividing walls, side walls) of the coil bobbin B may be stored.

[0054] The color temperature management item stores the color temperature value of the transmitted light, which is set according to the type of inspection object stored in the same record. The color wavelength management item stores the color wavelength value of the transmitted light, which is set according to the type of inspection object stored in the same record. The illuminance management item stores the illuminance value of the transmitted light, which is set according to the type of inspection object stored in the same record. The control unit 11 of the information processing device 1 can efficiently derive the physical quantity of the transmitted light corresponding to the type of inspection object (resin member) by referring to the irradiation table.

[0055] 7 is a flowchart illustrating a processing procedure performed by the control unit 11 of the information processing device 1. The control unit 11 of the information processing device 1 receives an operation from an operator using, for example, a keyboard connected to the input / output I / F 14, and performs the following processing based on the received operation.

[0056] The control unit 11 of the information processing device 1 acquires an image of a resin member used in a power transmission and distribution facility (S201). The control unit 11 of the information processing device 1 acquires an image (image data) of the resin member from the imaging device C, similar to the first embodiment. The image acquired by the control unit 11 of the information processing device 1 in this process is used as data for object recognition.

[0057] The control unit 11 of the information processing device 1 identifies the type of the resin member based on the acquired image (S202). The control unit 11 of the information processing device 1 identifies the type of the object (inspection subject) to be inspected based on the image of the resin member acquired from the imaging device C and the appearance of the resin member, such as its shape, material, or color.

[0058] When performing the process of identifying the object to be inspected, the control unit 11 of the information processing device 1 may use edge detection, pattern recognition, or an object recognition model such as YOLO on the acquired image. If the resin member includes a coil bobbin B and insulating parts other than the coil bobbin B, the control unit 11 of the information processing device 1 may identify the type of the object to be inspected according to the type of insulating parts. If the resin member is a coil bobbin B, the control unit 11 may identify the type (model, etc.) of the object to be inspected according to the shape, size, and thickness of the wall parts (dividing walls, side walls) of the coil bobbin B.

[0059] The control unit 11 of the information processing device 1 derives the physical quantity of the transmitted light according to the identified type of resin member (S203). When deriving the physical quantity of the transmitted light according to the identified type of resin member, the control unit 11 of the information processing device 1 may refer to an irradiation table stored in the storage unit 12. By referring to the irradiation table, the control unit 11 of the information processing device 1 derives the physical quantity corresponding to the identified type of inspection object, i.e., the color temperature, color wavelength (frequency), illuminance, brightness, etc. of the transmitted light.

[0060] The control unit 11 of the information processing device 1 executes a process for irradiating transmitted light with the derived physical quantity (S204). When executing the process for irradiating transmitted light with the derived physical quantity, the control unit 11 of the information processing device 1 may output a control signal to the light source device 2 according to the color temperature variable control of the light source device 2. Alternatively, when executing the process for irradiating transmitted light with the derived physical quantity, the control unit 11 of the information processing device 1 may output a control signal for increasing or decreasing the output wattage to the driving power supply 3 connected to the light source device 2. In this way, depending on the type of resin member being inspected, it is possible to irradiate light in a wavelength range with low absorption in the inspected resin member, thereby enabling adaptation to resin members made of various materials. Alternatively, by increasing or decreasing the intensity (illuminance) of transmitted light depending on the resin thickness of the coil bobbin B being inspected, it is possible to adapt to coil bobbins B having various resin thicknesses.

[0061] Furthermore, the control unit 11 of the information processing device 1 may identify an inspection target portion of the resin member based on an image of the resin member acquired from the imaging device C. In addition, the light source device 2 may be mounted on, for example, a moving device configured to move the light source device 2 relative to the resin member, and the control unit 11 of the information processing device 1 may output a control signal to the moving device to move the position of the light source device 2. As a result, the moving device may move the light source device 2 so that it approaches or faces the inspection target portion of the resin member in response to the control signal from the control unit 11 of the information processing device 1. In this way, by mounting the light source device 2 on the moving device and controlling the moving device to move the position of the light source device 2, it is possible to increase the number of applicable locations that are inspection target portions of resin members such as coil bobbins B, while automating the inspection.

[0062] The control unit 11 of the information processing device 1 acquires an image captured in a state in which the transmitted light is irradiated at the derived physical quantity onto a resin member used in a power transmission and distribution facility (S205). The control unit 11 of the information processing device 1 controls the light source device 2 or the driving power supply 3 to irradiate the transmitted light at the derived physical quantity, and performs control to irradiate the transmitted light onto a resin member such as a coil bobbin B that is an inspection object. Then, the control unit 11 of the information processing device 1 acquires an image captured by the imaging device C in the same manner as in step S101 of the first embodiment.

[0063] The control unit 11 of the information processing device 1 performs angle correction on the acquired image (S206). The control unit 11 of the information processing device 1 inputs the angle-corrected image to the learning model 101 (S207). The control unit 11 of the information processing device 1 acquires information on the internal state of the resin member from the learning model 101 (S208). The control unit 11 of the information processing device 1 outputs a determination result on whether the resin member is usable or not based on the acquired information on the internal state of the resin member (S209). The control unit 11 of the information processing device 1 performs processes from S206 to S209, similar to processes S102 to S105 in the first embodiment.

[0064] The embodiments disclosed herein are to be considered in all respects as illustrative and not restrictive. The scope of the present invention is defined by the claims, not by the above meaning, and is intended to include all modifications within the meaning and scope of the claims.

[0065] Multiple claims in the claims section may be combined with each other regardless of the form of reference. Multiple dependent claims are defined in the claims section, depending on multiple claims. Multiple dependent claims may not be defined in the claims section, but multiple dependent claims may be defined that depend on multiple dependent claims. [Explanation of symbols]

[0066] S coil bobbin determination system, B coil bobbin (resin member), T information terminal, C imaging device, 1 information processing device, 11 control unit, 12 memory unit, P program (program product), M recording medium, 13 communication unit, 14 input / output I / F, 101 learning model (void detection model), 2 light source device (white LED), 3 driving power supply (DC power supply)

Claims

1. On the computer, Obtaining an image of a resin member used in a power transmission and distribution facility while irradiating it with transmitted light; By inputting the acquired image including the resin member into a learning model that has been trained to output information regarding the internal state of the resin member when an image including the resin member irradiated with transmitted light is input, information regarding the internal state of the resin member is acquired from the learning model; Outputting the acquired information about the internal state of the resin member A program that executes a process.

2. the information about the internal state of the resin member includes information about the presence or absence of bubbles formed inside the resin member; If there are bubbles inside the resin member, it is determined that the resin member is unusable; If there are no air bubbles inside the resin member, it is determined that the resin member is usable. The program according to claim 1.

3. the resin member is a coil bobbin used in an electric power transmission and distribution transformer, The image includes a side wall or a dividing wall of the coil bobbin. The program according to claim 2.

4. Correcting the acquired image according to the imaging angle when capturing the image; Inputting the corrected image into the learning model The program according to claim 2.

5. The light source of the transmitted light is a white LED. The program according to claim 2.

6. deriving a physical quantity of the transmitted light based on the acquired image of the resin member; Processing is performed to irradiate transmitted light using the derived physical quantity. The program according to claim 2.

7. On the computer, Obtaining an image of a resin member used in a power transmission and distribution facility while irradiating it with transmitted light; By inputting the acquired image including the resin member into a learning model that has been trained to output information regarding the internal state of the resin member when an image including the resin member irradiated with transmitted light is input, information regarding the internal state of the resin member is acquired from the learning model; Outputting the acquired information about the internal state of the resin member An information processing method for executing a process.

8. an image acquisition unit that acquires an image captured while irradiating a resin member used in the power transmission and distribution equipment with transmitted light; an information acquisition unit that inputs an acquired image including the resin member into a learning model that has been trained to output information regarding the internal state of the resin member when an image including the resin member irradiated with transmitted light is input, and acquires information regarding the internal state of the resin member from the learning model; an output unit that outputs the acquired information about the internal state of the resin member; An information processing device comprising:

Images