JP7045030B2 - Inspection system, inspection method, server equipment, and program - Google Patents

Description

The present invention is to photograph an inspection object such as a solar module placed in a solar power plant or the like with a camera mounted on a drone, and analyze the image obtained by the photography by using artificial intelligence technology. It relates to an inspection system, an inspection method, a server device, and a program for detecting abnormalities in cells, modules, clusters, and strings.

Conventionally, with respect to a solar module arranged in a photovoltaic power plant or the like, a cell abnormality, a module abnormality, a cluster abnormality, a string abnormality, etc. have a serious influence on the power generation efficiency. For example, so-called hot spots are a phenomenon in which a part of the module is damaged due to heat generation due to manufacturing defects such as soldering defects and adhesion of fallen leaves. It causes a big loss.

Therefore, it is extremely important to detect such various abnormalities at an early stage and deal with them at an early stage, and inspection of these abnormalities is an indispensable item in the periodic inspection of photovoltaic power plants.

Generally, in order to find a hot spot generated in a solar module, an infrared camera is used to take a picture of the solar module to find an abnormal temperature part of the cell.

Regarding this type of technology, for example, in Patent Document 1, a method for inspecting a solar cell array including a plurality of solar cell panels, the solar cell array is energized, an image is acquired, the image is analyzed, and an index is calculated. However, a method of estimating the output power from the output characteristics with respect to the index and determining whether or not the replacement is possible is disclosed.

Japanese Unexamined Patent Publication No. 2013-36747

However, in the prior art disclosed in Patent Document 1 and the like, an inspection object such as a solar panel such as a solar power plant is photographed by a camera mounted on a drone, and the image obtained by the photographing is artificially captured. Analysis using intelligent technology failed to detect cell, module, cluster, and string anomalies.

The present invention has been made in view of such a problem, and an inspection object such as a solar power plant is photographed by remotely controlling an unmanned aerial vehicle such as a drone, and the photographed information is obtained by using artificial intelligence technology. The purpose is to detect abnormalities in the inspection target based on the above easily and quickly.

In order to solve the above problems, the inspection system according to the first aspect of the present invention is an inspection system provided with an unmanned aircraft, a command terminal, and a server device to inspect an inspection object, and is the unmanned aircraft. Is provided with a first communication unit that communicates with the command terminal, a navigation control unit that performs navigation control based on the route control information, and a camera that captures an actual image and an infrared image based on the photographing control information. The command terminal generates the navigation control information based on the route information from the unmanned aircraft and the second communication unit that communicates with the server device, and the unmanned aircraft via the second communication unit. The server includes an unmanned aircraft control unit for transmission, a camera control unit for generating the image shooting control information for controlling image capture by the camera of the unmanned aircraft, and transmitting the image control information to the unmanned aircraft via the second communication unit. The apparatus includes a third communication unit that communicates with the command terminal, a first storage unit that stores the route information, a second storage unit that stores the image analysis result, and an image that analyzes an image included in the shooting information. The unmanned aircraft includes an analysis unit, and the unmanned aircraft transmits the captured image obtained by the imaging to the command terminal via the first communication unit, and the command terminal transmits the captured image to the second communication unit. The image is transferred to the server device via the communication unit, the third communication unit receives the captured image, the image analysis unit classifies the captured image into a module area, and the module area is classified. The threshold image of the above is generated, a region where the captured image has a larger RGB value than the threshold image is detected as a difference image, noise is removed from the difference image, and a region to be classified is specified. By excluding the area where the short film of the specially projected area is less than a predetermined pixel, the abnormal area in the captured image is specified, and the abnormal area is referred to while referring to the past image analysis result stored in the second storage unit. Identify aspects.

The inspection method according to the second aspect of the present invention is an inspection method using an inspection system including an unmanned aircraft, a command terminal, and a server device to inspect an inspection object, and the command terminal is the server device. Generates navigation control information based on the navigation information from, generates image control information for controlling the camera, controls the navigation and image capture of the unmanned aircraft based on the navigation control information and the image control information, and controls the navigation and image capture of the unmanned aircraft. The unmanned aircraft navigates and shoots under the control of the command terminal, transmits the captured image to the command terminal, the command terminal transfers the captured image to the server device, and the server device receives the captured image. The module area is classified for the captured image, a threshold image of the module region is generated, a region where the captured image has a larger RGB value than the threshold image is detected as a difference image, and noise is removed from the difference image. By performing the above, specifying the area to be classified, and excluding the area where the short film of this specially projected area is less than a predetermined pixel, the abnormal area in the captured image is specified, and the past image analysis result is obtained. With reference to it, identify the aspect of the anomaly.

The server device according to the third aspect of the present invention is a server device that inspects an inspection object by freely communicating with an unmanned aircraft and a command terminal, and stores a communication unit that communicates with the command terminal and route information. A storage unit, a second storage unit for storing an image analysis result, and an image analysis unit for analyzing an image taken by an unmanned aircraft are provided, and the image analysis unit is provided via the communication unit. Upon receiving the captured image from the command terminal, the captured image is classified into module regions, a threshold image of the module region is generated, and the captured image has a larger RGB value than the threshold image. By detecting as a difference image, removing noise from the difference image, specifying an area to be classified, and excluding an area in which the short film of this specially projected area is less than a predetermined pixel, the captured image is included. The abnormal region is specified, and the mode of the abnormality is specified with reference to the past image analysis results stored in the second storage unit.

The program according to the fourth aspect of the present invention is a program executed by a server device that inspects an inspection object by freely communicating with an unmanned aircraft and a command terminal, and the server device is used by the unmanned aircraft. It functions as an image analysis unit that analyzes the captured image obtained by shooting, and the image analysis unit classifies the captured image into module regions, generates a threshold image of the module region, and uses the captured image. Detects a region having a larger RGB value than the threshold image as a difference image, removes noise from the difference image, identifies an region to be classified, and the short story of this specially projected region is less than a predetermined pixel. By excluding the region, the abnormal region in the captured image is specified, and the mode of the abnormality is specified while referring to the past image analysis results.

According to the present invention, an inspection object such as a solar power plant is photographed by remotely controlling an unmanned aerial vehicle such as a drone, and an abnormality of the inspection object based on the photographed information is easily detected by using artificial intelligence technology. It is possible to provide a rapid inspection system, inspection method, server device, and program.

It is a block diagram of the inspection system which concerns on one Embodiment of this invention. It is a block diagram of the drone of the inspection system. It is a block diagram of the station terminal of the inspection system. It is a block diagram of the server device of the data center of the inspection system. It is a flowchart which shows the processing procedure of the inspection system. It is a figure which shows an example of the analysis result display screen. It is a figure which shows an example of the analysis result display screen. It is a figure which shows an example of an inspection report. It is a figure which shows an example of an inspection report. It is a flowchart which shows the detailed processing procedure of the image analysis processing. It is a figure explaining the image analysis process. It is a figure which shows the example of the detected abnormality.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 shows and describes a configuration of an inspection system according to an embodiment of the present invention.

As shown in the figure, the inspection system includes a drone 1 as an unmanned aerial vehicle equipped with a camera, a station terminal 2 as a command terminal for remotely controlling the drone 1, and a server device 3 in a data center. is doing. The station terminal 2 is capable of wireless communication with the drone 1, and is capable of communicating with the server device 3 via a network 4 such as the Internet. Here, as the inspection target 5, a solar module installed in a photovoltaic power plant will be illustrated as an example for explanation.

In such a configuration, the server device 3 of the data station has received registration of a plurality of power plants in advance, and holds the route information of the drone 1 calculated based on the configuration of the solar module of each power plant. .. When the route information is transmitted from the server device 3 to the station terminal 2 standby at the power plant via the network 4, the station terminal 2 generates navigation control information based on the route information and navigates the drone 1. Implement control. More specifically, the station terminal 2 transmits the route control information to the drone 1 by wireless communication. At this time, the shooting control information for controlling the shooting conditions and the shooting timing related to the shooting by the camera is also transmitted by wireless communication.

When the drone 1 receives the navigation control information and the shooting control information transmitted from the station terminal 2, the drone 1 starts automatic navigation based on the navigation control information and takes a picture based on the shooting control information in the process of the navigation. In this example, the camera mounted on the drone 1 can acquire an infrared image in addition to the actual image, and at the time of imaging, the position information is acquired in real time and the shooting consisting of the actual image and the infrared image is taken. Link to information. The drone 1 transmits shooting information and position information to the station terminal 2 by wireless communication in real time in the process of automatic navigation. When the automatic navigation is completed, it may be transmitted in a batch.

The station terminal 2 associates the shooting information sent from the drone 1 with the location information and transfers the shooting information to the server device 3 of the data center. The server device 3 analyzes the infrared image included in this shooting information and detects an abnormality in the solar module. Then, the analysis result is transmitted to the station terminal 2 via the network 4. When the station terminal 2 receives this analysis result, it displays it so that the operator can confirm the mode and location of the abnormality.

FIG. 2 shows and describes a detailed configuration of the drone.

As shown in the figure, the drone 1 includes a control unit 11 that controls the entire control, and the control unit 11 includes a communication unit 12, a camera 13, a GPS 14, a drive system 15, a gyro unit 16, and a storage unit. It is free to communicate with the unit 17. The communication unit 12 performs wireless communication with the station terminal 2. The camera 13 is capable of acquiring an actual image and an infrared image, and is mounted on the drone 1 via an anti-vibration damper (not shown).

The GPS 14 measures the position information (latitude, longitude) of the drone 1. The drive system 15 performs automatic takeoff and landing during automatic navigation, and realizes navigation at an arbitrary position and altitude by hovering. The gyro unit 16 measures the altitude, flight attitude, and the like of the drone 1 during automatic navigation. The storage unit 17 includes a memory, a hard disk drive, and the like for temporarily holding shooting information, position information, and the like.

By operating based on the control program, the control unit 11 functions as a main control unit 11a, a shooting control unit 11b, a navigation control unit 11c, a position information acquisition unit 11d, and a shooting information acquisition unit 11d.

In such a configuration, when the navigation control information and the photographing control information from the station terminal 2 are received via the communication unit 12, the navigation control unit 11c decodes the navigation control information and sends a control signal to the drive system 15. Controls to carry out automatic navigation up to feed, automatic takeoff, hovering, and automatic landing. Further, the shooting control unit 11b decodes the shooting control information and controls the camera 13 so as to perform shooting under the shooting conditions specified at a specific timing and position. The shooting information acquisition unit 11e acquires an actual image and an infrared image from the camera 13. In this shooting process, the position information acquisition unit 11d acquires the position information (latitude, longitude) of the drone 1 from the GPS 14. The position information is stored in the storage unit 17 in association with the shooting information.

The position information includes information on the altitude, flight attitude, and shooting time measured by the gyro unit 16 in addition to the position information positioned by the GPS 14. In this way, the main control unit 11a transmits the position information and the shooting information stored in the storage unit 17 to the station terminal 2 via the communication unit 12. Of course, a fence function may be added to the navigation control unit 11c to automatically return to the flight start point and land at an arbitrary altitude and distance for safety. Further, a wireless telementor function may be further installed to appropriately display and record the flight position, altitude, etc. of the drone 1 aircraft. The drone 1 adopted in this embodiment has a flight time of about 20 minutes (during hovering) and a flight distance of about 10 km (during horizontal movement), but is not limited thereto.

Here, the camera 13 mounted on the drone 1 will be described in more detail. The camera 13 is lightweight and can be driven by a battery for a long time. Further, it has an autofocus function, and as described above, it is possible to obtain a real image by normal shooting and an outside line image by infrared shooting. It also has a plurality of recording modes (interval recording, one-shot recording, and moving image recording) as functions. The number of pixels is not particularly limited, but for example, a high resolution of 240 pixels with 320 pixels can be used.

FIG. 3 shows and describes a detailed configuration of the station terminal.

As shown in the figure, the station terminal 2 includes a control unit 21 that controls the whole. The control unit 21 is freely connected to the communication unit 22, the operation unit 23, the display unit 24, and the storage unit 25. By executing the control program, the control unit 21 functions as a main control unit 21a, a navigation information acquisition unit 21b, a shooting information / position information acquisition unit 21c, a drone control unit 21d, a camera control unit 21e, and a display control unit 21f. do.

The communication unit 22 has a function of performing wireless communication with the drone 1 and a function of communicating with the server device 3 via the network 4. For example, the communication unit 22 includes a 3G / LTE communication module. As the operation unit 23, a keyboard, a mouse, or the like can be adopted. The display unit 24 is composed of a liquid crystal display or the like, and can appropriately display shooting information or the like. The storage unit 25 is composed of a memory, a hard disk, or the like that stores the route information transmitted from the server device 3 of the data center, the shooting information, the position information, and the like transmitted from the drone 1.

In such a configuration, when the communication unit 22 receives the route information from the server device 3 of the data center, the route information acquisition unit 21b acquires the route information and stores it in the storage unit 25. Then, the drone control unit 21d generates navigation control information for automatically navigating the drone 1 based on the route information. Further, the camera control unit 21e receives settings such as shooting conditions based on the input by the operation of the operation unit 23 to the setting screen displayed on the display unit 24, and generates shooting control information. The main control unit 21a transmits the navigation control information and the photographing control information to the drone 1 via the communication unit 22. On the drone 1 side, automatic navigation is carried out based on this navigation control information, and shooting is carried out based on the shooting control information.

Then, when the shooting information and the position information transmitted from the drone 1 are received via the communication unit 22, the shooting information / position information acquisition unit 21c acquires this and stores it in the storage unit 25. The main control unit 21a reads out the shooting information and the position information from the storage unit 25 and transmits them to the server device 3 of the data center via the communication unit 22. The server device 3 performs image analysis based on the infrared image of the shooting information, and the analysis result is transmitted to the station terminal 2. When the station terminal 2 receives the data of the analysis result via the communication unit 22, it stores it in the storage unit 25 and displays it on the display unit 24 under the control of the display control unit 21f. As a result, the operator can confirm the mode and location of the abnormality of the inspection target.

The station terminal 2 may have a portable configuration including a duralumin case for storing the drone.

FIG. 4 shows and describes a detailed configuration of a server device in a data center.

As shown in the figure, the server device 3 of the data center includes a control unit 31 that controls the entire control. The control unit 31 is freely connected to the communication unit 32 and the storage unit 37. Then, by executing the control program, the control unit 31 functions as a main control unit 31a, an inspection target registration unit 31b, a drone route setting unit 31c, an image analysis unit 31d, and an inspection report generation unit 31e. The storage unit 37 is composed of a memory such as RAM or ROM, a hard disk drive (HDD), or the like, and stores an inspection target registration DB 33, a route information DB 34, an analysis result DB 35, and an inspection report DB 36.

The communication unit 32 executes communication with the station terminal 2 via the network 4. Information on the inspection target is stored in the inspection target DB 33. In this example, since the inspection target is a power plant, the ID and location of the power plant to be inspected, the arrangement information of the solar module in the power plant, the worker ID, and the like are stored in association with each other. The route information DB 34 stores the route information of the drone 1 which is associated with the ID of each power plant and matches the arrangement information of the solar modules of each power plant. The inspection result DB 35 stores information on past inspection results in association with the ID of each power plant. A huge amount of past inspection results are stored in this inspection result DB 35, and the image analysis unit 31d functions as a library for image analysis using artificial intelligence technology. The inspection report DB 36 stores the inspection report associated with the ID of each power plant.

In addition to the above, the storage unit 37 temporarily stores the shooting information, the position information, and the like transmitted from the station terminal 2.

In such a configuration, when the station terminal 2 obtains a predetermined authentication and accesses the server device 31, the server device 31 identifies the inspection target with reference to the inspection target registration DB 33. The drone route setting unit 31c refers to the route information DB 34, reads out the route information suitable for the inspection target, and the main control unit 31a transmits the route information to the station terminal 2 via the communication unit 32. The station terminal 2 executes automatic navigation by the drone 1 following this route information. Then, when the photographing information (actual image and infrared image) and the position information from the station terminal 2 are received via the communication unit 32, the image analysis unit 31d refers to the analysis result DB 35 as a library and artificial intelligence. Using technology, infrared images are analyzed to identify abnormalities and locations.

The analysis result is stored in the inspection result DB 35 in association with the ID of the power plant which is the inspection target. The main control unit 31a transmits this analysis result to the station terminal 2 via the communication unit 32. This encourages the station terminal 2 to view the analysis results. Subsequently, the inspection report generation unit 31e reads the inspection result information from the inspection result DB 35, and the inspection report for generating the inspection report is stored in the inspection report DB 36 in association with the power plant ID. ..

The individual authentication of the drone 1 is performed by the electronic authentication certificate pre-installed (registered) in the drone 1. Further, for the encryption of the data communication between the drone 1 and the station terminal 2 by the digital certificate, the digital certificate for SSL encrypted communication is used. Further, regarding the code signing to the software module of the drone 1, the code signing (electronic signature) using the electronic signature certificate (code signing certificate) is applied to the software module installed in the drone 1. , Authenticate the software distributor correctly and prove that it has not been spoofed or tampered with.

Hereinafter, the flow of processing by the inspection system according to the embodiment of the present invention will be described with reference to the flowchart of FIG. This also corresponds to the inspection method according to the embodiment of the present invention. Here, the description will proceed with reference to FIGS. 6 to 9 as appropriate.

In the server device 31, when the station terminal 2 obtains a predetermined authentication and accesses the server device 31, the main control unit 31a identifies the inspection target with reference to the inspection target registration DB 33, and the drone route setting unit 31c determines the route information DB 34. The route information suitable for the inspection target is read out, and the main control unit 31a transmits the route information to the station terminal 2 via the communication unit 32 (S1).

In the station terminal 2, when the communication unit 22 receives the route information from the server device 3, the route information acquisition unit 21b acquires it and stores it in the storage unit 25 (S2). Then, the drone control unit 21d generates navigation control information for automatically navigating the drone 1 based on the route information, and the camera control unit 21e operates the operation unit 23 on the setting screen displayed on the display unit 24. Based on the input by, the shooting conditions and the like are set, and the shooting control information is generated (S3). Then, the navigation control information and the shooting control information are transmitted to the drone 1 (S4).

In the drone 1, when the navigation control information and the shooting control information from the station terminal 2 are received via the communication unit 12 (S5), the navigation control unit 11c decodes the navigation control information, performs automatic navigation, and performs shooting control. The unit 11b decodes the shooting control information and controls the camera 13 so as to perform shooting under the shooting conditions specified at a specific timing and position (S6).

More specifically, in the route control in step S6, the route information of the drone 1 is set by using the drawing of the photovoltaic power plant and the map having the geocode information such as Google MAP. Specifically, the point point is set by the following method. That is, by setting the latitude and longitude of the point point, the current position information is compared with the position information set in the automatic route based on the position information of the GPS 14 mounted on the drone 1, and the difference is detected. , Control the set route. Then, the distance / direction from the coordinates of the automatic route start point is set at the point point, and automatic navigation control is performed to the point point according to the measured direction and distance. In this way, the route control that travels linearly between the points at a constant speed and a constant altitude is executed, and at the point point, the control for changing the direction toward the next point is executed.

Further, in the shooting control in step S6, the shutter interval time of the camera 13 is set according to the traveling speed of the drone 1, and the control for continuously connecting the images is performed. For example, when the drone 1 advances at a speed of 5 m / s, and when shooting is performed once every 20 m, the shutter interval time of the camera 13 is set to 4 seconds.

Subsequently, the shooting information acquisition unit 11e acquires an actual image and an infrared image transmitted from the camera 13. In the process of this shooting, the position information acquisition unit 11d acquires the position information (latitude, longitude) of the drone 1 from the GPS 14. The position information is stored in the storage unit 17 in association with the shooting information (S7). In this way, the shooting information and the position information are transmitted to the station terminal 2 (S8).

When the station terminal 2 receives the shooting information and the position information (S9), the station terminal 2 transfers the shooting information to the server device 3 (S10).

When the server device 3 receives the shooting information (actual image and infrared image) and the position information from the station terminal 2 via the communication unit 32 (S11), the image analysis unit 31d refers to the analysis result DB 35 as a library. While doing so, the infrared image is analyzed using artificial intelligence technology to detect the mode and location of the abnormality (S12). The details of the image analysis process will be described later. This analysis result is stored in the analysis result DB 35 in association with the ID of the power plant which is the inspection target.

In the server device 3, the main control unit 31a transmits this analysis result to the station terminal 2 via the communication unit 32 (S13). This encourages the station terminal 2 to view the analysis results. When the station terminal 2 receives the data of the analysis result via the communication unit 22 (S14), it stores it in the storage unit 25 and displays it on the display unit 24 under the control of the display control unit 21f (S15). ..

The state of display on the display unit 24 by the display control unit 21f is as shown in FIGS. 6 and 7. That is, the captured image display window 100a and the analysis result display window 100b are displayed side by side on the display screen 100, and on the right side thereof, the defective portion is displayed on the display based on the map data provided by the general-purpose map software. The marking display window 100c marked with is displayed. Then, when "output" of the tab is selected, the screen is switched to the screen as shown in FIG. 7, and the map display screen 101 in which the defective portion is marked on the recording map is superimposed and displayed by a pop-up window or the like. When the play button 100d is selected, the moving image in the captured image display window 100a is played. At this time, the analysis result display window 100b is also linked.

On the other hand, on the server device 3 side, the inspection report generation unit 31e reads the inspection result information from the inspection result DB 35 and generates the inspection report (S16).

An example of an inspection report is as shown in FIGS. 8 and 9. In the example of FIG. 8, a drawing in which defective parts are marked on a recording map and a table showing the number and contents of abnormal spots in each area are printed. On the other hand, in the example of FIG. 9, an image showing the defect portion marked in the infrared image and a table showing the temperature difference between the high temperature portion and the peripheral portion, which are the marking points, are printed. There is.

Here, the electronic network is further supplemented. The electronic network is a safe navigation mechanism that prevents unmanned aerial vehicles such as drones from flying outside the assumed navigation area by electronically constructing boundaries where they can fly. This mechanism works independently of the unmanned aerial vehicle system to recognize the electronic network even in an abnormal situation of the unmanned aerial vehicle, and if it goes out of the electronic network, the navigation is stopped urgently. The electronic network system consists of an electronic network control device and an electronic network beacon. The electronic network control device is installed in an unmanned aerial vehicle such as a drone, receives a signal from a beacon, and constantly confirms that the unmanned aerial vehicle is in the electronic network.

The electronic network controller receives signals from three or more electronic network beacons and constantly confirms that the unmanned aerial vehicle is within the navigation area. The electronic network control device is equipped with a wireless device in the same frequency band as the control signal of the unmanned aerial vehicle, and if it is determined that the unmanned aerial vehicle has gone out of the electronic network, the control and communication of the unmanned aerial vehicle are predetermined. Shut off for a while and stop the flight of the unmanned aerial vehicle. If the unmanned aerial vehicle still moves away from the electronic network, the electronic network controller shuts off the main power of the unmanned aerial vehicle and stops navigation.

A plurality of electronic network beacons are installed along the boundary line of the navigation area on the ground, and a plurality of electronic network beacons and one or more electronic network beacons are installed in the navigation area on the boundary line to determine the navigation area. The electronic beacon transmits a short-range wireless communication signal in the sub GHz band and conveys the position of the electronic network to the electronic network control device. The signal of the electronic beacon has an electronic beacon identification number, and even if a plurality of electronic networks are adjacent to each other, they are identified and distinguished from their own electronic network.

Next, with reference to the flowchart of FIG. 10, the image analysis process using the artificial intelligence technique by the image analysis unit 31d executed in step S12 of FIG. 5 will be described in detail. Here, the description will proceed with reference to FIGS. 11 and 12 as appropriate.

When this process is started, the image analysis unit 31d first classifies the module area as shown in FIG. 11B based on the input image as shown in FIG. 11A (S21). Subsequently, the image analysis unit 31d generates a threshold image of the module region as shown in FIG. 11 (c) (S22). In this embodiment, the threshold value is generated based on the temperature of the peripheral image + 2 degrees. For areas not on the module, the RGB value threshold is set to 255.

Next, as shown in FIG. 11D, the image analysis unit 31d detects a region where the input image has a larger RGB value than the threshold image as a difference image (S23). Next, as shown in FIG. 11 (e), noise reduction of the difference image is performed (S24). As a result, the area to be classified can be obtained. Then, the area where the short story of the detected area is less than a predetermined pixel (for example, 8 pixels) is excluded (S25), and the candidate area is classified by using the deep learning technique while referring to the analysis result DB35 as a library (S26). ), This process will be returned. The state of the output image is shown in FIG. 11 (f).

In the classification of the candidate region in step S26 described above, it is necessary to classify which of the abnormalities shown in FIGS. 12 (a) to 12 (d) corresponds to. That is, FIG. 12A shows an abnormality in the cell, and in detail, a cell crack, a stain on the module, a shadow, an abnormality in the junction box, and the like correspond to this. FIG. 12B shows an abnormality in the module, which corresponds to glass breakage, backsheet abnormality, and the like. FIG. 12 (c) shows an abnormality of the cluster, which corresponds to an interconnector failure, a solder failure, a bypass diode short circuit, and the like. FIG. 12D shows an abnormality in the string, which corresponds to connector damage, cable damage, junction box opening, and the like.

The image analysis unit 31d will classify which of the above abnormalities corresponds to each region while referring to the analysis result DB 35 as a library based on the size and shape of the narrowed region. Since the analysis result is stored in the analysis result DB 35 and the analysis result has a mechanism for constructing big data each time the detection is performed, it is expected that the detection accuracy will be improved every time the detection is repeated.

Therefore, according to the image analysis by the image analysis unit 31d, it is possible to detect and identify the mode (cell abnormality, module abnormality, cluster abnormality, string abnormality, etc.) and location of the module abnormality. More specifically, it is possible to identify whether the anomaly is a cluster, a hotspot, an entire module, or a string, and where it is on the image.

As described above, according to one embodiment of the present invention, the inspection system includes a drone 1 as an unmanned aerial vehicle, a station terminal 2 as a command terminal, and a server device 3 of a data center, and inspects an inspection object. Therefore, the unmanned aerial vehicle 1 captures an actual image and an infrared image based on the first communication unit 12 that communicates with the command terminal 2, the navigation control unit 11c that performs navigation control based on the route control information, and the photographing control information. The command terminal 2 generates the navigation control information based on the route information from the second communication unit 22 that communicates with the unmanned aerial vehicle 1 and the server device 3, and the server device 3, and the command terminal 2 generates the navigation control information. 2 An unmanned aerial vehicle control unit 21d that transmits to the unmanned aerial vehicle 1 via the communication unit 22 and a shooting control information that controls shooting by the camera 13 of the unmanned aerial vehicle 1 are generated and sent to the unmanned aerial vehicle 1 via the second communication unit 22. The server device 3 includes a camera control unit 21e for transmission, and the server device 3 stores a third communication unit 32 that communicates with the command terminal 2, a route information DB 34 as a first storage unit that stores route information, and an image analysis result. The unmanned aerial vehicle 1 includes an analysis result DB 35 as a second storage unit and an image analysis unit 31d for analyzing an image included in the shooting information, and the unmanned aerial vehicle 1 transmits the shot image obtained by shooting via the first communication unit 12. The command terminal 2 transfers the captured image to the server device 3 via the second communication unit 22, and the third communication unit 32 receives the captured image in the server device 3 for image analysis. Unit 31d provides an inspection system that identifies an abnormal region in a captured image and identifies an abnormality mode with reference to past image analysis results stored in the second storage unit 35.

Here, the image analysis unit 31d may specify whether or not it is an abnormal region by comparing it with a threshold value of an RGB value. Further, the image analysis unit 31d may specify the mode of abnormality in the abnormal region based on the size and shape of the abnormal region.

Therefore, by remotely controlling the drone, the mode and location of the abnormality of the inspection target can be detected in real time, and it can be confirmed in real time on the station terminal, and the inspection report can be created easily, quickly and automatically. It will be possible to provide.

Although one embodiment of the present invention has been described above, it goes without saying that the present invention is not limited to this and various improvements and changes can be made without departing from the spirit of the present invention.

For example, the types of objects to be detected and the types of abnormalities that can be detected are not limited to those described above, and can correspond to various types.

1 ... Drone, 2 ... Station terminal, 3 ... Server device, 4 ... Network, 11 ... Control unit, 11a ... Main control unit, 11b ... Shooting control unit, 11c ... Navigation control unit, 11d ... Position information acquisition unit, 11e ... Shooting information acquisition unit, 12 ... communication unit, 13 ... camera, 14 ... GPS, 15 ... drive system, 16 ... gyro unit, 17 ... storage unit, 21 ... control unit, 21a ... main control unit, 21b ... route information acquisition unit , 21c ... Shooting information / position information acquisition unit, 21d ... Drone control unit, 21e ... Camera control unit, 21f ... Display control unit, 22 ... Communication unit, 23 ... Operation unit, 24 ... Display unit, 25 ... Storage unit, 31 ... Control unit, 31a ... Main control unit, 31b ... Inspection target registration unit, 31c ... Drone route setting unit, 31d ... Image analysis unit, 31e ... Inspection report generation unit, 32 ... Communication unit, 33 ... Inspection target registration DB, 34 ... Route information DB, 35 ... Analysis result DB, 36 ... Inspection report DB, 37 ... Storage unit, 38 ... Program.

Claims (4)

An inspection system equipped with an unmanned aerial vehicle, a command terminal, and a server device to inspect an object to be inspected.
The unmanned aerial vehicle
The first communication unit that communicates with the command terminal and
A navigation control unit that controls navigation based on navigation control information,
It is equipped with a camera that shoots real images and infrared images based on shooting control information.
The command terminal is
A second communication unit that communicates with the unmanned aerial vehicle and the server device,
An unmanned aerial vehicle control unit that generates the navigation control information based on the route information from the server device and transmits the navigation control information to the unmanned aerial vehicle via the second communication unit.
A camera control unit that generates the shooting control information that controls shooting by the camera of the unmanned aerial vehicle and transmits the shooting control information to the unmanned aerial vehicle via the second communication unit is provided .
The server device is
A third communication unit that communicates with the command terminal,
The first storage unit that stores the route information and
A second storage unit that stores image analysis results,
It is equipped with an image analysis unit that analyzes the image included in the shooting information.
The unmanned aerial vehicle transmits the photographed image obtained by the photographing to the command terminal via the first communication unit, and the command terminal transmits the photographed image to the server via the second communication unit. The server device is transferred to the device, and the third communication unit receives the captured image in the server device.
The image analysis unit classifies the module region of the captured image, generates a threshold image of the module region, and detects a region where the captured image has a larger RGB value than the threshold image as a difference image. By removing noise from the difference image, identifying an area to be classified, and excluding an area where the short film of this specially projected area is less than a predetermined pixel, an abnormal area in the captured image is specified. An inspection system that identifies an abnormality mode with reference to past image analysis results stored in the second storage unit. It is an inspection method using an inspection system that is equipped with an unmanned aerial vehicle, a command terminal, and a server device, and inspects the inspection target.
The command terminal generates navigation control information based on the navigation information from the server device, generates shooting control information for controlling the camera, and generates shooting control information for controlling the camera, and the unmanned aerial vehicle based on the navigation control information and the shooting control information. Control navigation and shooting,
The unmanned aerial vehicle navigates and shoots under the control of the command terminal, and transmits the shot image to the command terminal.
The command terminal transfers the captured image to the server device, and the command terminal transfers the captured image to the server device.
The server device classifies the module region of the captured image, generates a threshold image of the module region, detects a region where the captured image has a larger RGB value than the threshold image, and detects the region as a difference image. By removing noise from the difference image, identifying the area to be classified, and excluding the area where the short film of this specially projected area is less than a predetermined pixel, the abnormal area in the captured image is specified, and the past. An inspection method that identifies the mode of abnormality while referring to the image analysis results of. It is a server device that inspects objects to be inspected by freely communicating with unmanned aerial vehicles and command terminals.
A communication unit that communicates with the command terminal,
The first storage unit that stores route information and
A second storage unit that stores image analysis results,
It is equipped with an image analysis unit that analyzes the captured image obtained by shooting with the unmanned aerial vehicle.
When the image analysis unit receives the captured image from the command terminal via the communication unit, the image analysis unit classifies the captured image into a module region, generates a threshold image of the module region, and the captured image is better. A region having a larger RGB value than the threshold image is detected as a difference image, noise is removed from the difference image, a region to be classified is specified, and a short film of this specially projected region is a region having less than a predetermined pixel. By excluding the above, a server device that identifies an abnormal region in the captured image and identifies an abnormal mode while referring to a past image analysis result stored in the second storage unit. A program that can communicate freely with unmanned aerial vehicles and command terminals and is executed by a server device that inspects objects to be inspected.
The server device
An image analysis unit that analyzes captured images obtained by shooting with the unmanned aerial vehicle,
To function as
The image analysis unit classifies the module region of the captured image, generates a threshold image of the module region, and detects a region where the captured image has a larger RGB value than the threshold image as a difference image. By removing noise from the difference image, identifying an area to be classified, and excluding an area where the short film of this specially projected area is less than a predetermined pixel, an abnormal area in the captured image is specified. A program that identifies the mode of abnormality while referring to past image analysis results.

Images