JP7757741B2 - Ground abnormality detection device, ground abnormality detection system, and ground abnormality detection method - Google Patents

Description

The present invention relates to a ground abnormality detection device and a ground abnormality detection method.

Substations often require site development or the addition of additional electrical equipment to improve and expand their operations. As a result, substation sites are generally large and numerous pieces of equipment are installed there. This makes it important to monitor the ground to detect any subsidence or tilt that may occur as a result of site development work or the installation of new electrical equipment.

Currently, the method of checking for ground abnormalities involves maintenance personnel visiting the site and visually inspecting it, and if necessary, measuring the tilt of equipment in various locations to determine whether there are any abnormal changes in the ground. However, as substations occupy large areas, as mentioned above, patrolling the entire substation to check for abnormalities is extremely time-consuming and requires specialized knowledge and experience. Therefore, methods are being sought to more efficiently perform this work by making full use of IT (Information Technology) and other techniques.

As an example of a method for detecting abnormalities using IT technology, Patent Document 1 discloses a plant premises abnormality sign detection system for detecting abnormal signs within a plant premises. The system includes a first detection system installed in a target process within the plant premises to detect conditions inside the process, a second detection system to detect conditions outside the target process within the plant premises, and a complementary and integrated processing unit that uses information collected from the first detection system and information collected from the second detection system to integrate information in accordance with causal relationships and empirical rules and detects abnormal signs in the target process within the plant premises.

Japanese Patent Application Laid-Open No. 2019-204192

However, there has been insufficient development of anomaly detection technology that targets a wide range of anomalies, such as ground subsidence at substations, rather than just the equipment itself.
The present invention has been made in view of the above background, and its object is to provide a ground abnormality detection device and a ground abnormality detection method for efficiently detecting ground abnormalities in electric power facilities.

One aspect of the present invention that achieves the above-mentioned objectives is a ground abnormality detection device that includes an image acquisition unit having a processor and memory that acquires photographed images that include images of multiple pieces of equipment at a power facility, and an abnormality detection unit that outputs information regarding abnormal displacement of the ground on which the equipment included in the acquired photographed images is installed by inputting the acquired photographed images into a trained model that receives input of images that include multiple pieces of equipment at a power facility and outputs information indicating abnormal displacement of the ground on which the equipment is installed.

As in the present invention, by inputting a photographed image containing images of multiple pieces of equipment at a power facility into a trained model that outputs information indicating abnormal ground displacement at the power facility, and outputting information regarding abnormal ground displacement on which the equipment contained in the photographed image is installed, power facility maintenance personnel and others can learn about abnormalities in the power facility's ground (subsidence or tilt) using only images (photographs, etc.) of the on-site equipment, without having to visit the power facility to check or survey the site. This also reduces the cost of surveying. Furthermore, by detecting abnormalities using the entire image containing multiple pieces of equipment rather than using images of individual pieces of equipment, the risk of missing a ground abnormality can be reduced.

Another aspect of the present invention for achieving the aforementioned objective is the ground abnormality detection device, in which the abnormality detection unit inputs an image containing multiple pieces of equipment at the power facility and inputs the acquired captured image into a trained model that outputs information on whether or not the ground on which the equipment is installed at the time the image was captured has been surveyed, thereby outputting information on whether or not the ground on which the equipment is installed that is included in the captured image needs to be surveyed.

As in the present invention, by using a trained model that outputs information on whether or not the ground on which the equipment is installed has been surveyed at the time the image was captured, and outputting information on whether or not the ground on which the equipment is installed contained in the captured image needs to be surveyed, maintenance personnel and others can know whether or not the ground related to the captured image currently needs to be surveyed and take practical action.

Another aspect of the present invention for achieving the aforementioned objective is the ground abnormality detection device, in which the abnormality detection unit inputs an image containing multiple pieces of equipment at the power facility and inputs the acquired captured image into a trained model that outputs information on whether or not the ground on which the equipment is installed at the time the image was captured was surveyed and information on the measuring device used for the survey, thereby outputting information on whether or not the ground on which the equipment included in the captured image is installed needs to be surveyed and information on the measuring device to be used for the survey.

As with the present invention, by using a trained model that outputs information on whether or not the ground on which the equipment is installed has been surveyed at the time the image was taken, as well as information on the measuring instrument used for the survey, maintenance personnel and others can know what measuring instrument to use if a survey is necessary, and can take appropriate action in response to ground abnormalities even if they do not have specialized knowledge of surveying.

Another aspect of the present invention for achieving the aforementioned objective is the ground abnormality detection device, wherein the abnormality detection unit inputs an image including multiple pieces of equipment at the power facility into a trained model that outputs the amount of displacement of the ground on which the multiple pieces of equipment are installed at the time the image was taken, relative to their positions at a predetermined reference time. By inputting the acquired captured image into a trained model that outputs the amount of displacement of the ground on which the equipment is installed at the time the image was taken, relative to their positions at the predetermined reference time, the abnormality detection unit outputs information on the amount of displacement of the ground on which the equipment is installed, included in the captured image, relative to their positions at the predetermined reference time.

As in the present invention, by using a trained model that outputs the amount of displacement of the ground on which the equipment is installed at the time the image was taken, relative to the position of the ground on which the equipment is installed at a specified reference time, maintenance personnel and others can obtain information on the current amount of displacement of the ground on which the equipment is installed (for example, the amount of subsidence and tilt of the ground from the time the site was developed to the present). This allows maintenance personnel and others to take necessary and sufficient measures in response to ground abnormalities.

Another aspect of the present invention for achieving the aforementioned objective is the ground abnormality detection device, in which the abnormality detection unit inputs an image including multiple pieces of equipment at the power facility into a trained model that outputs information indicating the amount of displacement of the ground on which the multiple pieces of equipment are installed a predetermined time after the image was taken, by inputting the acquired captured image into the trained model that outputs information indicating the amount of displacement of the ground on which the multiple pieces of equipment are installed a predetermined time after the image was taken.

As in the present invention, by using a trained model that outputs information indicating the amount of displacement of the ground on which multiple pieces of equipment are installed a predetermined time after the image was captured, maintenance personnel and others can know the amount of future displacement after the image was captured. This allows maintenance personnel and others to take appropriate preemptive measures against ground abnormalities that may occur in the future.

Another aspect of the present invention for achieving the aforementioned objectives is the ground abnormality detection device, wherein the image acquisition unit acquires photographed images, including images of equipment installed on the ground of the power facility formed by embankment, taken by a photographing device installed on the ground of the power facility formed by cutting, and the abnormality detection unit inputs the acquired photographed images into the trained model, thereby outputting information regarding whether or not a survey of the embankment on which the equipment included in the photographed images is installed is required.

As in this invention, a camera installed on the ground in the cut section of a power facility captures images of equipment on the ground in the embankment section of the power facility, and by using these captured images in a trained model, maintenance personnel can accurately identify abnormalities in the ground in the embankment section of a power facility, where subsidence is likely to occur, by using a camera point in the cut section where the ground is relatively solid. For example, this can prevent various accidents that occur when ground movement due to construction or expansion work on a power facility causes equipment to tilt.

In another aspect of the present invention for achieving the aforementioned objectives, the ground abnormality detection device includes a screen display unit that displays, on a specified screen, information about abnormal displacement of the ground on which the equipment is installed, which is included in the acquired photographic images.

In this way, by displaying information about abnormal ground displacement on the screen, maintenance personnel can take appropriate measures in response to ground subsidence, tilt, etc.

Another aspect of the present invention for achieving the above-mentioned objective is a ground abnormality detection method in which an information processing device performs an image acquisition process to acquire photographed images including images of multiple pieces of equipment at a power facility, and an abnormality detection process to output information regarding abnormal displacement of the ground on which the equipment is installed, by inputting the acquired photographed images into a trained model that receives input of images including multiple pieces of equipment at the power facility and outputs information indicating abnormal displacement of the ground on which the equipment is installed, the information being included in the acquired photographed images.

This invention makes it possible to efficiently detect ground abnormalities at power facilities.

1 is a diagram showing an example of the configuration of an electrical facility in which a ground abnormality detection device according to an embodiment of the present invention is installed. FIG. 2 is a diagram illustrating an example of functions provided in the ground abnormality detection device. FIG. 10 is a diagram illustrating an example of a captured image. FIG. 10 is a diagram showing another example of a captured image. FIG. 2 is a diagram illustrating an example of hardware of a ground abnormality detection device. FIG. 10 is a flow diagram illustrating an example of an abnormality detection support process. FIG. 10 is a flow diagram illustrating an example of a model creation process. FIG. 2 is a diagram illustrating an example of a teacher data DB. FIG. 10 is a flow diagram illustrating an example of an abnormality detection process. FIG. 10 is a diagram showing an example of a ground information screen.

A ground abnormality detection device according to one embodiment of the present invention will be described with reference to the drawings.
--System Configuration--
1 is a diagram showing an example of the configuration of an electric facility in which a ground abnormality detection device according to this embodiment is installed. This electric power facility is, for example, a substation, a power plant, or the like, in which a plurality of devices are installed on a site (ground) of a predetermined area. In this embodiment, this electric power facility is assumed to be a substation.

The substation 1 according to this embodiment has a site that extends from a cut section 3 (e.g., on the land side) to an embankment section 5 (e.g., on the sea side). That is, the substation 1 comprises the cut section 3, which is a relatively solid area formed by cutting the slope on the land side, and the embankment section 5, which is an area on the sea side formed by embankment. The substation 1 is equipped with numerous devices 10, such as disconnectors, circuit breakers, and transformers, which are connected by predetermined electrical wires 12 (busbars, etc.).

The substation 1 is equipped with a camera 20 that takes comprehensive photographs of equipment 10 in a section installed on the embankment 5 and acquires the images as needed. The camera 20 is fixed at a position on the embankment 5 a predetermined distance from the equipment 10 being photographed, and is adjusted to an angle that allows it to photograph the entire equipment 10 being photographed. The camera 20 is, for example, an ITV (Industrial Television) or a camera. Note that while only one camera 20 is shown in the figure, multiple camera devices 20 may be used to photograph images of the equipment 10 from multiple different directions.

Furthermore, the substation 1 is provided with a ground abnormality detection device 30, which is an information processing device communicatively connected to the imaging devices 20. The ground abnormality detection device 30 and each imaging device 20 are communicatively connected via a wired or wireless communication network 7, such as the Internet, a LAN (Local Area Network), a WAN (Wide Area Network), or a dedicated line.

The ground abnormality detection device 30 creates an abnormality detection model that can detect abnormal displacements in the ground on which the equipment 10 is installed (e.g., subsidence or tilt of the embankment 5) from images of the equipment 10 in the substation 1 previously captured by the imaging device 20 (hereinafter referred to as teacher images).

The anomaly detection model is a trained model generated by machine learning. In this embodiment, the anomaly detection model is constructed by machine learning the features of a teacher image using deep learning. That is, the anomaly detection model of this embodiment is a neural network having an input layer to which pixel information of the image is input, one or more intermediate layers (hidden layers) that extract image features from the pixel information, and an output layer that outputs information regarding abnormal displacement of each device 10 and the ground on which each device 10 is installed (hereinafter referred to as ground information) from the image features. Note that the anomaly detection model of this embodiment outputs multiple types of output values, but a separate trained model may be created for each type of output value.

Note that, although it is possible to apply neural networks such as CNN (Convolution Neural Network), SVM (Support Vector Machine), Bayesian network, or regression tree, this embodiment assumes the use of CNN, which is a method suitable for image recognition.

Thereafter, the ground abnormality detection device 30 inputs a new image captured by the photographing device 20 (for example, an image of the equipment 10 at the current substation 1) into the abnormality detection model, thereby detecting abnormalities in the displacement of the ground on which the equipment 10 contained in the image (hereinafter referred to as the judgment image) is installed, and displays information about the abnormality (ground information, etc.) on a specified screen.
Next, the ground abnormality detection device 30 will be described in detail.

--Ground abnormality detection device--
2 is a diagram showing an example of functions of the ground abnormality detection device 30. The ground abnormality detection device 30 includes an image acquisition unit 31, a model creation unit 33, an abnormality detection unit 35, and a screen display unit 37.

The image acquisition unit 31 acquires captured images including images of multiple devices 10 in the substation 1. In this embodiment, these captured images are teacher images or judgment images captured by the imaging device 20, but they may also be images captured by a maintenance worker or the like from the same position and direction as the imaging device 20.

Here, Figure 3 shows an example of a captured image. This captured image 100 was taken by the camera 20 from the cut section 3 in the direction of the embankment 5, and shows multiple mounting equipment 101 installed at predetermined intervals on the ground surface of the embankment 5 facing the sea (the embankment 5 side), and the busbar 103 bridged by these mounting equipment 101.

Figure 4 shows another example of a captured image. This captured image 120 is a side view of multiple potential transformers 121 (here, PDs (potential devices)) located at various locations from the cut section 3 toward the bank section 5, and a busbar 123 bridged by the potential transformers 121.

Next, as shown in FIG. 2, the model creation unit 33 creates a trained model (anomaly detection model) based on the training image. The trained model receives an image including multiple pieces of equipment 10 in the substation 1 and outputs information indicating abnormal displacement of the ground on which the equipment 10 is installed. The training data including the training image is stored in the training data DB 50, which is a database described below.

The anomaly detection unit 35 inputs the captured image (determination image) acquired by the image acquisition unit 31 into the anomaly detection model, and outputs information regarding abnormal displacement of the ground on which the equipment 10 is installed, which is included in the captured image.

Specifically, the anomaly detection unit 35 inputs the captured image acquired by the image acquisition unit 31 into the anomaly detection model, and outputs information on whether or not the ground on which the equipment 10 is installed needs to be surveyed, as well as information on the measuring instrument to be used for the survey, which is included in the captured image.

The anomaly detection unit 35 also uses an anomaly detection model to estimate the current amount of subsidence of the ground on which the equipment 10 is installed, which is included in the judgment image. That is, the anomaly detection unit 35 inputs the photographed image acquired by the image acquisition unit 31 into the anomaly detection model, and outputs information on the amount of displacement of the ground on which the equipment is installed, which is included in the photographed image, at the time the photographed image was taken, relative to its position at a predetermined reference time.

The anomaly detection unit 35 also uses an anomaly detection model to predict the future amount of subsidence of the ground on which the equipment 10 included in the judgment image is installed. That is, the anomaly detection unit 35 inputs the captured image acquired by the image acquisition unit 31 into the anomaly detection model, and outputs information indicating the amount of displacement of the ground on which the multiple pieces of equipment 10 included in the captured image are installed a predetermined time after the image was captured.

Next, the screen display unit 37 displays information about abnormal displacement of the ground on which the equipment 10 is installed, which is included in the captured image acquired by the image acquisition unit 31, on a specified screen.

Here, Figure 5 is a diagram showing an example of the hardware of the ground abnormality detection device 30. The ground abnormality detection device 30 includes a processor 21, a main memory device 22, an auxiliary memory device 23, an input device 24, an output device 25, and a communication device 26.

The processor 21 is configured using, for example, a CPU (Central Processing Unit), MPU (Micro Processing Unit), GPU (Graphics Processing Unit), FPGA (Field Programmable Gate Array), ASIC (Application Specific Integrated Circuit), AI (Artificial Intelligence) chip, etc.

The main memory device 22 is a device that stores programs and data, and is, for example, ROM (Read Only Memory), RAM (Random Access Memory), or non-volatile memory (NVRAM (Non Volatile RAM)).

The auxiliary storage device 23 may be, for example, an SSD (Solid State Drive), a hard disk drive, an optical storage device (CD (Compact Disc), DVD (Digital Versatile Disc), etc.), a storage system, an IC card, a read/write device for recording media such as an SD card or optical recording media, or a storage area of a cloud server. Programs and data can be read into the auxiliary storage device 23 via a recording media reader or communication device 26. Programs and data stored in the auxiliary storage device 23 are read into the main storage device 22 as needed.

The input device 24 is an interface that accepts input from outside, and may be, for example, a keyboard, mouse, touch panel, card reader, pen-input tablet, or voice input device.

The output device 25 is an interface that outputs various information such as processing progress and results. The output device 25 is, for example, a display device (liquid crystal monitor, LCD (Liquid Crystal Display), graphics card, etc.) that visualizes the various information described above, a device that converts the various information described above into audio (audio output device (speaker, etc.)), or a device that converts the various information described above into text (printer, etc.). Note that, for example, the ground abnormality detection device 30 may be configured to input and output information to and from other devices via the communication device 26.

The input device 24 and output device 25 constitute a user interface that accepts and presents information to the user.

The communication device 26 is a device that enables communication with the image capture device 20. The communication device 26 is a wired or wireless communication interface that enables communication with other devices via a communication network such as the Internet, and is, for example, a NIC (Network Interface Card), a wireless communication module, a USB module, etc.

The ground abnormality detection device 30 may be realized, in whole or in part, using virtual information processing resources provided using virtualization technology, process space separation technology, or the like, such as a virtual server provided by a cloud system. Also, all or part of the functions provided by the ground abnormality detection device 30 may be realized, for example, by a service provided by a cloud system via an API (Application Programming Interface), or the like. The ground abnormality detection device 30 may also be configured as a system including multiple information processing devices connected to each other so that they can communicate with each other.

The above-mentioned functions of the ground abnormality detection device 30 are realized by the processor 21 of the ground abnormality detection device 30 reading and executing a program stored in the main memory device 22, or by hardware (FPGA, ASIC, AI chip, etc.) that constitutes the ground abnormality detection device 30. The ground abnormality detection device 30 stores the above-mentioned various information (data), for example, as a database table or a file managed by a file system.
Next, the processing performed by the ground abnormality detection device 30 will be described.

--process--
FIG. 6 is a flow diagram illustrating an example of a ground abnormality detection process that is a process performed by the ground abnormality detection device 30.

First, the model creation unit 33 of the ground abnormality detection device 30 executes a model creation process to create an abnormality detection model based on images (teacher images) containing the equipment 10 of the substation 1 taken at various times in the past (s1).

Then, the anomaly detection unit 35 inputs an image (determination image) of the current devices 10 taken from the same position and direction as the teacher image into the created anomaly detection model, and causes the anomaly detection model to output information (ground information) indicating whether or not an abnormality such as subsidence or tilting has occurred in the ground on which these devices 10 are installed, and executes an anomaly detection process that displays the content of this information on the screen (s3).

Thereafter, maintenance personnel at the substation 1 refer to the displayed screen and take appropriate measures, such as surveying the ground, improving the ground of the embankment 5 at the substation 1, or changing the installation location of the equipment 10, as necessary.
The model creation process and the abnormality detection process will be described in detail below.

-Model creation process-
7 is a flow diagram illustrating an example of the model creation process. The model creation process is started, for example, when a predetermined input is made to the ground abnormality detection device 30 by a maintenance worker or at a predetermined timing (a predetermined time or a predetermined time interval).

First, the model creation unit 33 acquires a training image (s11). For example, the model creation unit 33 acquires an image previously captured by the imaging device 20.

Then, the model creation unit 33 sets label data (correct labels) for each acquired training image, thereby creating a training data DB 50 that records training data, which is a set of training images and label data (s13).

(Teacher data DB)
8 is a diagram showing an example of the teacher data DB 50. The teacher data DB 50 is created by data input by a maintenance worker or the like.

The teacher data DB 50 contains the following information: a shooting point 51, which indicates the installation location of the camera 20 that captured each teacher image; a shooting date 53 for each teacher image; image data 55, which is a data file recording each teacher image; and label data 57 associated with each teacher image.

The label data 57 includes: Abnormal Equipment 61, which is area information for each piece of equipment 10 (or multiple pieces of equipment) in the teacher image that has abnormal displacement of the ground on which it is installed; Survey Status 63, which is information indicating whether a survey was conducted on each piece of equipment 10 in the teacher image to confirm an abnormality as of the image capture date 53; Measuring Instrument 65, which is information on the measuring instrument used for the survey; Current Subsidence 67, which is the amount of displacement of the ground on which each piece of equipment 10 in the teacher image is installed as of the image capture date 53 (e.g., the amount of displacement relative to the substation 1's designed position); and Subsidence in Three Years 69, which is the amount of displacement of the ground on which each piece of equipment 10 included in the teacher image is installed a predetermined time later (three years in this embodiment) from the image capture date 53. Note that Survey Status 63 is set to "Yes" if a survey has been conducted, or "No" if a survey has not been conducted, which respectively correspond to whether or not a survey is required in the output value of the anomaly detection model.

Note that the types of label data 57 are not limited to those shown here, and any information related to abnormal displacement of the installation ground may be used. Furthermore, label data 57 may be only a subset of the types shown here. Label data 57 may be set, for example, by a maintenance worker or the like.

Note that the model creation unit 33 may create an image corresponding to the teacher image, in which there is no displacement of the ground on which each piece of equipment 10 is installed (e.g., when the substation 1 is designed), and add this as one piece of teacher data. For example, the model creation unit 33 creates three-dimensional structural data (e.g., three-dimensional CAD data) of the substation 1 including each piece of equipment 10 based on data representing the dimensions of each piece of equipment 10 installed in the substation 1 (e.g., two-dimensional design drawings such as plan views and side views, and data on the height of the ground surface). The model creation unit 33 then creates an image corresponding to the teacher image by performing perspective transformation on the created three-dimensional structural data using a predetermined algorithm such as central projection. Note that, for example, data on the viewpoint and vanishing point (calculated based on, for example, the position, focal length, and field of view of the imaging device 20) are used as parameters for the perspective transformation.

Next, as shown in s15 of Figure 7, the model creation unit 33 inputs each training image registered in the training data DB 50 into the anomaly detection model, thereby obtaining each output value or intermediate value corresponding to the training image. Note that when this process is executed for the first time, initial values are set in advance for the hyperparameters (described next) of the anomaly detection model.

The model creation unit 33 optimizes the hyperparameters in the anomaly detection model so that each acquired output value or intermediate value approaches each correct answer data (label data 57) registered in the training data DB 50 (s17). For example, the model creation unit 33 adjusts hyperparameters such as weights between units (neurons) or coefficients in activation functions using a learning method such as backpropagation. This completes the model creation process.

The model creation unit 33 may configure the anomaly detection model from two or more trained models. As an example, the model creation unit 33 first receives an actual image as input and generates a first trained model that outputs area information for abnormal equipment in the actual image. Next, the model creation unit 33 receives area information for abnormal equipment as input and generates a second trained model that outputs whether or not a survey was conducted, the measuring device, the current subsidence amount, and the subsidence amount in three years. In this way, by configuring the anomaly detection model from two or more trained models, it is possible to accurately recognize areas for abnormal equipment and output more accurate results.

- Abnormality detection processing -
9 is a flow diagram illustrating an example of an abnormality detection process. The abnormality detection process is started, for example, when a predetermined input is received from a maintenance technician after the model creation process is executed, or at a predetermined timing (a predetermined time or a predetermined time interval).

First, the abnormality detection unit 35 acquires an image (determination image) of multiple devices 10 in the substation 1 (s31). For example, the abnormality detection unit 35 acquires an image of the current device 10 captured by the imaging device 20 via the communication network 7.

The abnormality detection unit 35 inputs the acquired judgment image into the abnormality detection model to acquire an output value (ground information) (s33). For example, the abnormality detection unit 35 acquires information regarding abnormal equipment in the judgment image and whether or not surveying of the ground on which the equipment 10 is installed is required (information corresponding to abnormal equipment 61, surveying presence/absence 63, measuring device 65, current subsidence amount 67, and subsidence amount in three years 69).

Then, the screen display unit 37 displays the acquired ground information on the screen (s35).

10 is a diagram showing an example of a ground information screen 300 that displays ground information. The ground information screen 300 displays a photographing point 301 of the determination image, a determination image 303 showing the abnormal device, and information 307 about the installation ground of the abnormal device.
The information 307 on the ground where the abnormal equipment is installed includes information on whether or not surveying is required, as well as information 309 identifying the abnormal equipment (such as the name of the equipment 10 and its management number), a measuring instrument 311 to be used to survey the equipment 10 or the ground where the equipment 10 is installed, the current estimated amount of subsidence 313 of the ground where the equipment 10 is installed, and the predicted amount of subsidence 315 of the ground where the equipment 10 is installed three years from now.

In the example shown in the figure, the assessment image 303 highlights the mounting equipment 305 on the ground that has been determined to have abnormal displacement (ground subsidence), and the busbar 306 attached to that mounting equipment 305 that has the same positional abnormality (sagging, etc.) as the mounting equipment 305. In this case, a maintenance worker or the like may take action such as inserting an adapter or the like to raise the busbar 306.

As described above, the ground abnormality detection device 30 of this embodiment inputs a determination image including multiple pieces of equipment 10 at the substation 1 into an abnormality detection model that outputs information indicating abnormal displacement of the ground on which the equipment 10 is installed, and outputs information related to abnormal displacement of the ground on which the equipment 10 included in the determination image are installed. This allows substation maintenance personnel and others to learn about ground abnormalities (subsidence or tilt) at the substation 1 using only images (photographs, etc.) of the on-site equipment, without having to visit the substation to check or survey, etc. This also reduces the cost required for surveying. Furthermore, by detecting abnormalities using the entire image including multiple pieces of equipment rather than using images of individual pieces of equipment 10, the risk of missing a ground abnormality can be reduced.

In addition, the ground abnormality detection device 30 of this embodiment uses an abnormality detection model that outputs information on whether or not the ground on which the equipment is installed has been surveyed at the time the image was captured, and outputs information on whether or not the ground on which the equipment is installed, which is included in the captured image, needs to be surveyed. This allows maintenance personnel and others to know whether or not the ground related to the captured image currently needs to be surveyed, and to take practical measures to address any ground abnormalities.

In addition, the ground abnormality detection device 30 of this embodiment uses a trained model that outputs information on whether the ground on which the equipment is installed has been surveyed at the time the image was captured, as well as information on the measuring instrument used for the survey. This allows maintenance personnel and others to know what measuring instrument to use if a survey is necessary, and allows them to take appropriate action in response to ground abnormalities even if they do not have specialized knowledge of surveying.

In addition, the ground abnormality detection device 30 of this embodiment uses a trained model that outputs the amount of displacement of the ground on which the equipment is installed at the time the image was taken, relative to the position of the ground on which the equipment is installed at a specified reference time. This allows maintenance personnel and others to know information about the current amount of displacement of the ground on which the equipment is installed (for example, the amount of subsidence and tilt of the ground from the time the site was developed to the present). This allows maintenance personnel and others to take necessary and sufficient measures in response to ground abnormalities.

In addition, the ground abnormality detection device 30 of this embodiment uses a trained model that outputs information indicating the amount of displacement of the ground on which multiple pieces of equipment are installed a predetermined time after the image is captured, allowing maintenance personnel and others to know the amount of future displacement after the captured image. This allows maintenance personnel and others to take appropriate pre-emptive measures against ground abnormalities that may occur in the future.

In addition, the ground abnormality detection device 30 of this embodiment uses a camera installed on the ground in the cut earth portion of the substation to capture images of equipment on the ground in the embankment portion of the substation. By using these captured images in the trained model, maintenance personnel can accurately determine ground abnormalities in the embankment portion of the substation, where subsidence is likely to occur, by using a camera point in the cut earth portion, where the ground is relatively solid. For example, this can prevent various accidents that occur when the ground shifts due to construction or expansion work on the substation, causing the equipment 10 to tilt.

In addition, the ground abnormality detection device 30 of this embodiment displays information about abnormal ground displacement on a screen, allowing maintenance personnel and others to take appropriate measures in response to ground subsidence, tilt, etc.

The above description of the embodiments is intended to facilitate understanding of the present invention and is not intended to limit the present invention. The present invention may be modified or improved without departing from its spirit, and equivalents thereof are also included within the scope of the present invention.

For example, in this embodiment, the ground abnormality detection device 30 is installed in a substation, but it can also be applied to other types of power facilities equipped with various devices, such as power plants.

In addition, in this embodiment, the imaging device 20 is installed in the cutting section 3 and is installed in a direction that will capture images of the equipment 10 in the embankment section 5, but the installation position and imaging direction of the imaging device 20 are not particularly limited. Furthermore, the equipment 10 may also be installed in the cutting section 3.

Furthermore, in this embodiment, the ground abnormality detection device 30 is installed within the premises of the substation 1, but it may also be installed outside the premises of the substation 1.

Furthermore, although the anomaly detection model in this embodiment is configured as a single model (neural network), it may also be configured as multiple models. For example, the anomaly detection model may be configured from a first model that outputs the area of each device 10 on the image, and a second model that outputs information on abnormal displacement of the installation ground from the output value of the first model.

Furthermore, the anomaly detection model of this embodiment is based on the premise that a specific value is output as each output value, but it may also output candidate output values along with their accuracy.

Furthermore, in this embodiment, the ground abnormality detection device 30 creates the abnormality detection model, but another information processing device may create the abnormality detection model and the ground abnormality detection device 30 may acquire it.

In addition, in this embodiment, the ground abnormality detection device 30 displays the ground information screen, but the ground information screen may also be displayed on a mobile device carried by a maintenance worker or the like.

In addition, in this embodiment, the anomaly detection model outputs whether or not a survey is required based on the survey status 63 learned from the teacher data DB 50, but other methods may also be used. For example, if the current estimated subsidence output by the anomaly detection model is equal to or greater than a predetermined threshold, it may be determined that a survey is required for the corresponding equipment 10.

In this embodiment, the anomaly detection model predicts ground displacement three years from now, but it may also predict ground displacement at multiple future points in time. In this case, measurement data of ground displacement at multiple points in time after the teacher image was captured is set in the teacher data DB 50, for example. Furthermore, by using this data at multiple points in time, a regression equation relating to the amount of ground displacement at any future point in time may be found, and the ground displacement at any future point in time may be found and displayed on the screen.

In addition, when creating an anomaly detection model, equipment fixed to the premises outside the substation 1 (e.g., high-voltage pylons) may be included in the teacher image and judgment image without fail. Then, before inputting the teacher image and judgment image into the anomaly detection model, a correction process (projective transformation, etc.) may be performed to make the positions of the areas on the image of the equipment in the teacher image and judgment image the same. This makes it possible to prevent a decrease in the accuracy of detecting abnormal ground displacement, even if, for example, the position or angle of the imaging device 20 changes for some reason.

1 Substation, 3 Cutting section, 5 Embankment section, 10 Equipment, 12 Power line, 20 Photography device, 30 Ground abnormality detection device, 7 Communication network, 31 Image acquisition unit, 33 Model creation unit, 35 Abnormality detection unit, 37 Screen display unit, 100 Photographed image, 101 Mounting equipment, 103 Bus bar, 120 Photographed image, 121 Potential transformer, 123 Bus bar, 21 Processor, 22 Main memory device, 23 Auxiliary memory device, 24 Input device, 25 Output device, 26 Communication device, 50 Teacher data DB, 51 Photography point, 53 Photography date, 55 Image data, 57 Label data, 61 Abnormal equipment, 63 Surveying presence/absence, 65 Measuring device, 67 Current subsidence, 69 Subsidence amount after three years, 300 Ground information screen, 301 Photography point, 303 Determination image, 305 Mounting equipment, 306 Bus bar, 307 Information on the installation ground of the abnormal equipment, 309 Information to identify the abnormal equipment, 311 Measuring device to be used, 313 Current estimated subsidence amount, 315 Predicted subsidence amount three years from now

Claims (6)

a processor and a memory;
an image acquisition unit that acquires a photographed image including an image of a plurality of devices of the power facility and an image of the ground on which the plurality of devices are installed ;
a trained model that receives as input an image including a plurality of pieces of equipment at an electric power facility and the ground on which the plurality of pieces of equipment are installed, and outputs information indicating abnormal displacement of the ground on which the plurality of pieces of equipment are installed, information on whether a survey of the ground on which the plurality of pieces of equipment is installed at the time the image was taken, and information on the measuring instrument used for the survey , and The anomaly detection unit receives an image including a plurality of devices of the power facility and the ground on which the plurality of devices are installed , and outputs information about the amount of displacement of the ground on which the plurality of devices are installed at the time the image was taken, relative to the position of the ground on which the plurality of devices are installed at a predetermined reference time, by inputting the acquired photographed image into a trained model that outputs information about the amount of displacement of the ground on which the devices are installed, included in the photographed image, relative to the position at the predetermined reference time,
The ground abnormality detection device according to claim 1. The anomaly detection unit inputs an image including a plurality of devices of the power facility and the ground on which the plurality of devices are installed , and outputs information indicating the amount of displacement of the ground on which the plurality of devices are installed after a predetermined time from the time the image is taken, by inputting the acquired photographed image into a trained model that outputs information indicating the amount of displacement of the ground on which the plurality of devices are installed after a predetermined time from the time the image is taken.
The ground abnormality detection device according to claim 1. The ground abnormality detection device according to claim 1;
an imaging device installed on the ground of the power facility formed by cutting earth;
and a plurality of devices installed on the ground of the power facility formed by embankment,
an image acquisition unit of the ground abnormality detection device acquires a photographed image including the plurality of devices and the ground on which the plurality of devices are installed, the photographed image being captured by the photographing device ;
The anomaly detection unit of the ground anomaly detection device inputs the acquired photographed image into the trained model, and outputs information regarding whether or not a survey of the embankment on which the equipment is installed, which is included in the photographed image, is necessary, and information regarding the measuring instrument to be used for the survey .
Ground anomaly detection system. a screen display unit that displays information about abnormal displacement of the ground on which the equipment is installed, included in the acquired photographed image, on a predetermined screen;
The ground abnormality detection device according to claim 1. The information processing device
an image acquisition process for acquiring a photographed image including an image of a plurality of devices of the power facility and an image of the ground on which the plurality of devices are installed ;
a trained model that receives as input an image including a plurality of pieces of equipment at a power facility and the ground on which the plurality of pieces of equipment are installed, and outputs information indicating abnormal displacement of the ground on which the plurality of pieces of equipment are installed, information on whether a survey of the ground on which the plurality of pieces of equipment is installed at the time the image was taken, and information on the measuring instrument used for the survey; and