Computer program, deterioration level determination device, and deterioration level determination method

Abstract

To provide a computer program, a deterioration level determination device and a deterioration level determination method which can improve deterioration diagnostic accuracy of an accessory instrument of a ground wire or electric wire.SOLUTION: A computer program causes a computer to execute processing of acquiring an image obtained by imaging an accessory instrument of a ground wire or electric wire, acquiring the likelihood of a deterioration level of the accessory instrument captured in the image by inputting the acquired image to a learning model that outputs the likelihood of the deterioration level of the accessory instrument when inputting the image obtained by imaging the accessory instrument of the ground wire or electric wire, and determining whether or not the deterioration level of the accessory instrument is the specific deterioration level on the basis of the acquired likelihood and a prescribed threshold.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a computer program, a deterioration level determination device, and a deterioration level determination method. [Background technology]

[0002] When wind blows against overhead electric wires or overhead ground wires, Karman vortices, which are alternating vortex currents, are generated on the downwind side of the electric wires or ground wires, causing vibrations with amplitudes of one to several tens of millimeters. These vibrations repeatedly apply stress to the electric wires or ground wires at vibration reflection points such as suspension support points and tension clamps, which are the fixed points of the overhead electric wires or overhead ground wires, causing metal fatigue deterioration and breakage of the overhead electric wires and overhead ground wires over time. To prevent this, auxiliary devices such as double torsional dampers are installed near the support points of the overhead electric wires and overhead ground wires.

[0003] Patent Document 1 discloses a double torsional damper that includes a restraining wire, a weight fixed to both ends of the restraining wire, and a wire gripping clamp fixed to the center of the restraining wire, and the restraining wire has a mechanism that prevents the weight from falling. It also devised a method of acquiring an image of the double torsional damper to prevent the weight from falling, inputting the acquired image into a learning model generated by machine learning, and determining the deterioration level of the double torsional damper. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Laid-Open No. 2012-5203 Summary of the Invention [Problem to be solved by the invention]

[0005] Such a learning model uses a function (e.g., a softmax function) that converts the output values ​​of each deterioration level so that the sum of the values ​​is 1.0 (100%). When the deterioration level is finally determined, the deterioration level belonging to the class with the largest output value of the function is diagnosed as the deterioration level of the double torsional damper. For this reason, even if the output value of the deterioration level diagnosed as abnormal is not negligible, if the output value of the deterioration level diagnosed as normal is larger, it may ultimately be diagnosed as normal, and something that is actually abnormal may be diagnosed as normal.

[0006] The present invention has been made in view of the above circumstances, and aims to provide a computer program, a deterioration level determination device, and a deterioration level determination method that can improve the accuracy of deterioration diagnosis of a ground wire or an attachment device of an electric wire. [Means for solving the problem]

[0007] The present application includes multiple means for solving the above-described problems. As one example, a computer program causes a computer to execute a process of acquiring an image of an accessory to a ground wire or electric cable, inputting the acquired image into a learning model that outputs the accuracy of the deterioration level of the accessory when the image of the accessory to a ground wire or electric cable is input, acquiring the accuracy of the deterioration level of the accessory captured in the image, and determining whether the deterioration level of the accessory is a specific deterioration level based on the acquired accuracy and a predetermined threshold value. [Effects of the Invention]

[0008] According to the present invention, it is possible to improve the accuracy of diagnosing deterioration of a ground wire or an attachment device of an electric wire. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 illustrates an example of the configuration of a deterioration level determination system. [Figure 2] FIG. 10 is a diagram illustrating an example of an image of a damper. [Figure 3]FIG. 10 is a diagram showing an example of a deterioration level and a determination criterion of a damper. [Figure 4] FIG. 10 is a diagram illustrating an example of deterioration determination using a learning model. [Figure 5] FIG. 10 is a diagram illustrating an example of a deterioration diagnosis of a damper in a comparative example. [Figure 6] 10A and 10B are diagrams illustrating an example of deterioration determination by a deterioration level determining unit. [Figure 7] FIG. 2 is a diagram illustrating a first example of degradation determination by the degradation level determination device. [Figure 8] FIG. 10 is a diagram showing a second example of degradation determination by the degradation level determination device. [Figure 9] FIG. 10 is a diagram showing a third example of degradation determination by the degradation level determination device. [Figure 10] FIG. 10 is a diagram showing a first example of an evaluation result of deterioration determination by the deterioration level determination device. [Figure 11] FIG. 10 is a diagram showing a second example of the evaluation result of the deterioration determination by the deterioration level determination device. [Figure 12] FIG. 10 is a diagram showing an example of a diagnosis result screen. [Figure 13] FIG. 10 is a diagram illustrating an example of a processing procedure of a deterioration level determination device. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a diagram showing an example of the configuration of a deterioration level determination system. The deterioration level determination system includes a deterioration level determination device 50. A data server 10 and a terminal device 20 are connected to the deterioration level determination device 50 via a communication network 1. The data server 10 collects and records images (image data) of attached devices of ground wires or electric wires. The attached devices are, for example, dampers (torsion dampers, double torsion dampers). Hereinafter, the attached devices will also be referred to as dampers. Images of the dampers are captured, for example, during inspection, and inspection and image capture are performed by a drone, human hands, a self-propelled robot, or a helicopter.

[0011] FIG. 2 is a diagram showing an example of an image of a damper. FIG. 2 is a schematic diagram of a captured image. The damper has a restraining wire 101, a weight 102, and a wire-holding clamp 103. The weight 102 is fixed to both ends of the restraining wire 101. The wire-holding clamp 103 holds an overhead electric wire (overhead ground wire) 104. If the restraining wire 101 deteriorates due to rust or the like, there is a risk that the restraining wire 101 will break and the weight 102 will fall. Therefore, in diagnosing the deterioration of a damper, it is particularly important to determine the deterioration level of the restraining wire 101. The captured image used by the deterioration level determination device 50 to determine the deterioration level may be an image of a predetermined size that shows the entire damper from an image of the entire damper, or an image that shows the entire restraining wire may be cut out from an image of the entire damper.

[0012] The data server 10 records the captured image as shown in Fig. 2 in association with information such as the damper model, manufacturing date, installation location, and photographing date. For dampers for which a deterioration diagnosis has been performed, the data server 10 also records the diagnosis date and the judgment result in association with the captured image.

[0013] The terminal device 20 can be configured, for example, as a personal computer, a tablet terminal, or the like, and is used by a person in charge of diagnosing deterioration of a damper. The person in charge can use the terminal device 20 to retrieve captured images of the object to be diagnosed with deterioration from the data server 10 and upload the retrieved captured images to the deterioration level determination device 50, thereby obtaining and displaying the diagnosis results from the deterioration level determination device 50. Note that the captured images may be sent directly from the data server 10 to the deterioration level determination device 50 by running a predetermined program or the like, without going through the terminal device 20.

[0014] The deterioration level determination device 50 includes a control unit 51 that controls the entire device, a communication unit 52, a memory 53, a deterioration level determination unit 54, an output unit 55, and a storage unit 56. The storage unit 56 stores a computer program 57 and a learning model 58.

[0015] The control unit 51 can be configured with a CPU (Central Processing Unit), an MPU (Micro-Processing Unit), a GPU (Graphics Processing Unit), etc. The control unit 51 can execute processing defined by a computer program 57. In other words, the processing by the control unit 51 is also processing by the computer program 57.

[0016] The communication unit 52 includes, for example, a communication module, and has a function of communicating with the data server 10 and the terminal device 20 via the communication network 1. The communication unit 52 can acquire (receive) an image (captured image) of the degradation diagnosis target from the terminal device 20 or the data server 10.

[0017] The memory 53 can be configured with semiconductor memory such as SRAM (Static Random Access Memory), DRAM (Dynamic Random Access Memory), flash memory, etc. A computer program 57 can be loaded into the memory 53, and the control unit 51 can execute the computer program 57.

[0018] The storage unit 56 can be configured, for example, with a hard disk or semiconductor memory, and may store necessary information in addition to the computer program 57 and learning model 58.

[0019] The output unit 55 outputs the deterioration diagnosis result of the damper obtained by the deterioration level determination device 50 to the terminal device 20.

[0020] The learning model 58 can be configured, for example, by a convolutional neural network (CNN), and is equipped with a softmax function in the final layer, and is generated so as to output the probability (accuracy) of the deterioration level of the damper when an image of the damper is input. The learning model 58 may also be AlexNet, ZFNet, GoogLeNet, VGGNet, ResNet, SENet, etc.

[0021] Fig. 3 is a diagram showing an example of the deterioration levels of a damper and the criteria for judgment. As shown in Fig. 3, the deterioration levels of a damper can be classified into five stages: Rank I, Rank II, Rank III, Rank IV, and Rank V, with the higher the rank, the greater the degree of deterioration. Ranks IV and V indicate that the deterioration diagnosis of the damper is abnormal, and ranks I to III indicate that the deterioration diagnosis of the damper is normal (minor). Note that the classification of the ranks is not limited to the example shown in Fig. 3.

[0022] For example, Rank I indicates that there is no red rust on the deterrent wire, and the color characteristic of zinc plating remains. Rank II indicates that there is no red rust on the deterrent wire, but the surface is discolored to a brown or black color. Rank III indicates that there is a small amount of red rust on the deterrent wire, or that red rust has occurred on only part of the deterrent wire. Rank IV indicates that red rust has occurred on almost the entire front surface of the deterrent wire, or that red rust has occurred on the front surface of the deterrent wire and has not yet reached severe pitting corrosion. Rank V indicates that there is severe red rust on the deterrent wire accompanied by pitting corrosion.

[0023] Fig. 4 is a diagram showing an example of deterioration determination using the learning model 58. The control unit 51 has functions as an image acquisition unit and an accuracy acquisition unit, and when it acquires an image (captured image) of the damper, it inputs the acquired image into the learning model 58 and acquires output data from the learning model 58. The output data is the probability of each deterioration level (rank I to V). In the example of Fig. 4, for convenience, the probabilities of ranks I to V are represented by P1, P2, P3, P4, and P5, respectively.

[0024] Here, a case where deterioration diagnosis of a damper is performed only by the deterioration determination using the learning model 58 will be described as a comparative example.

[0025] FIG. 5 is a diagram illustrating an example of a damper deterioration diagnosis in a comparative example. As shown in FIG. 5, the output of the learning model 58 is, for example, set to have probabilities of 0 (%), 5 (%), 50 (%), 40 (%), and 5 (%) for ranks I to V, respectively. In this case, the maximum probability is 50% for rank III, so the deterioration diagnosis of the damper is determined to be deterioration level III. However, although the probability of rank IV is 40%, this is not the maximum probability, and therefore is not taken into consideration at all in the deterioration diagnosis. Considering that the probability of rank IV is 40%, it is considered that there is a risk in diagnosing the damper as having a deterioration level III. Therefore, in this embodiment, the deterioration diagnosis of the damper is performed not only using the learning model 58 but also using the deterioration level determination unit 54. The deterioration determination by the deterioration level determination unit 54 will be described below.

[0026] The deterioration level determination unit 54 functions as a determination unit and determines whether the damper deterioration level is a specific deterioration level based on the probability (accuracy) of the deterioration level output by the learning model 58 and a predetermined threshold. The specific deterioration levels are deterioration levels at which the damper is diagnosed as abnormal, and in this embodiment, these are rank IV and rank V. Specifically, the deterioration level determination unit 54 performs deterioration diagnosis of the damper in the following order of priority:

[0027] 6 is a diagram showing an example of deterioration determination by the deterioration level determination unit 54. As priority 1, the deterioration level determination unit 54 determines whether the probability of rank V is equal to or greater than a first threshold, and if the probability of rank V is equal to or greater than the first threshold, determines the deterioration level of the damper to be rank V. The first threshold can be set as appropriate, for example, to 10%, 5%, etc. If the probability of rank V is not equal to or greater than the first threshold, the deterioration level determination unit 54 moves on to determination of priority 2.

[0028] For priority 2, the deterioration level determination unit 54 determines whether the probability of rank IV is equal to or greater than a second threshold, and if the probability of rank IV is equal to or greater than the second threshold, determines the deterioration level of the damper to be rank IV. The second threshold can be set appropriately, for example, to 15%, 10%, or the like. The second threshold can be set to a value greater than the first threshold, but is not limited to this. If the probability of rank IV is not equal to or greater than the second threshold, the deterioration level determination unit 54 moves on to determining priority 3. In this way, for ranks V and IV, where the deterioration diagnosis is determined to be abnormal, the deterioration level of the damper is determined using the probability of the deterioration level output by the learning model 58 and a predetermined threshold.

[0029] As described above, the predetermined threshold value can be set according to a specific deterioration level (rank V, IV). The deterioration level determination unit 54 determines the deterioration level of the damper according to the priority of the specific deterioration level. The priority corresponds to the degree of abnormality of the specific deterioration level (rank V is priority 1, and rank IV is priority 2).

[0030] As the third priority, the deterioration level determination unit 54 determines which rank among ranks I to III has the highest probability, and determines the rank with the highest probability as the deterioration level of the damper. That is, if the deterioration level of the damper is not a specific deterioration level, the deterioration level determination unit 54 determines the deterioration level of the damper based on the probability output by the learning model 58. A specific example will be described below.

[0031] FIG. 7 is a diagram showing a first example of deterioration determination by the deterioration level determination device 50. As shown in FIG. 7, the output of the learning model 58 has probabilities of ranks I to V of 15(%), 50(%), 10(%), 10(%), and 15(%), for example. In this case, the maximum probability is 50% for rank II. First, deterioration determination is performed for rank V. The probability of rank V is 15%, which is greater than or equal to a first threshold value (for example, 10%), so the deterioration level of the damper is determined to be rank V.

[0032] FIG. 8 is a diagram showing a second example of deterioration determination by the deterioration level determination device 50. As shown in FIG. 8, the output of the learning model 58 is, for example, 5% (%), 20% (%), 40% (%), 30% (%), and 5% (%) for ranks I to V, respectively. In this case, the maximum probability is 40% for rank III. First, deterioration determination is performed for rank V. Since the probability of rank V is 5% and is not greater than or equal to a first threshold value (for example, 10%), deterioration determination is performed for rank IV in order of priority. Since the probability of rank IV is 30% and is greater than or equal to a second threshold value (for example, 15%), the deterioration level of the damper is determined to be rank IV.

[0033] FIG. 9 is a diagram showing a third example of deterioration determination by the deterioration level determination device 50. As shown in FIG. 9, the output of the learning model 58 is, for example, 5% (%), 50% (%), 30% (%), 10% (%), and 5% (%) for ranks I to V, respectively. In this case, the maximum probability is 50% for rank II. First, deterioration determination is performed for rank V. Since the probability of rank V is 5% and is not equal to or greater than a first threshold value (e.g., 10%), deterioration determination is performed for rank IV in order of priority. Since the probability of rank IV is 10% and is not equal to or greater than a second threshold value (e.g., 15%), deterioration determination is performed for ranks I to III in order of priority. Since the maximum probability among ranks I to III is 50% for rank II, the deterioration level of the damper is determined to be rank II.

[0034] Next, the evaluation results of the deterioration determination by the deterioration level determining device 50 will be described.

[0035] FIG. 10 is a diagram showing a first example of the evaluation results of the deterioration determination by the deterioration level determination device 50. The indicators of "failure rate" and "false alarm rate" are used to evaluate the deterioration determination. The failure rate is the rate at which abnormal dampers are not diagnosed as abnormal, and includes, for example, the rate at which rank V is not diagnosed as rank V and the rate at which rank IV is not diagnosed as rank IV. Diagnosing something that is actually abnormal as normal should be minimized in deterioration diagnosis, and the rate at which this oversight is called the failure rate. A system that performs deterioration diagnosis in actual operation is required to have a low failure rate. The false alarm rate is the rate at which dampers that are not abnormal are diagnosed as abnormal, and includes, for example, the rate at which ranks I to III are diagnosed as rank IV and the rate at which ranks I to III are diagnosed as rank V.

[0036] For dampers that have undergone deterioration diagnosis within a predetermined period such as one year, six months, or three months, the target is a failure rate of 5% or less and a false alarm rate of 20% or less for both ranks IV and V. According to the deterioration level determination device 50 of this embodiment, for rank IV, the failure rate was 5% and the false alarm rate was 4%, both of which achieved the target. Also, for rank V, the failure rate was 3% and the false alarm rate was 8%, both of which achieved the target. Furthermore, although not shown, the false alarm rates for ranks I to III were 10% or less.

[0037] FIG. 11 is a diagram showing a second example of the evaluation results of deterioration determination by the deterioration level determination device 50. FIG. 11 illustrates the "accuracy" and "detection rate" for each of ranks IV and V. As with the comparative example in FIG. 5, the accuracy is the accuracy when the deterioration diagnosis of the damper is performed using only the deterioration determination by the learning model 58. The accuracy is the proportion of those determined to be abnormal that were actually abnormal. The accuracy for rank IV was approximately 75%, and the accuracy for rank V was approximately 85%.

[0038] The detection rate is the detection rate when the deterioration level determination unit 54 is also used to perform the deterioration diagnosis of the damper. The detection rate is the rate at which abnormal dampers are determined to be abnormal, and is calculated as {100(%) - detection rate (%)} = false alarm rate (%). As shown in FIG. 11, by also using the deterioration level determination unit 54, the detection rate for rank IV was 95%, and the detection rate for rank V was 97%. That is, the false alarm rate for rank IV was 5%, and the false alarm rate for rank V was 3%.

[0039] FIG. 12 is a diagram showing an example of a diagnosis result screen. The diagnosis result screen is displayed on, for example, the terminal device 20. The diagnosis result screen illustrates the device ID of the damper being diagnosed with deterioration, a deterioration level column, a maximum probability column, and the probability for each rank. The deterioration level column displays the deterioration diagnosis result for the damper. The deterioration level column indicates the final deterioration diagnosis of the damper using the deterioration level determination unit 54 as well (rank IV in the example shown). The maximum probability indicates the rank with the highest probability among the probabilities for each deterioration level output by the learning model 58 (rank III in the example shown). Although the probability for each rank is illustrated as being ranks I to V, it is also possible to display a comparison between the probability of rank IV, which is a specific deterioration level, and the probability of rank III, which is the maximum probability.

[0040] As described above, when the deterioration level of the damper is a specific deterioration level (rank IV in the example shown in the figure), the control unit 51 may display the maximum probability (maximum value of probability) of the deterioration level output by the learning model 58 in comparison with the probability of the specific deterioration level output by the learning model 58.

[0041] 13 is a diagram showing an example of a processing procedure of the deterioration level determination device 50. For convenience, the following description will be given assuming that the processing is performed by the control unit 51. The control unit 51 acquires an image of the accessory device (damper) (S11), inputs the acquired image into the learning model 58, and acquires the probability of the deterioration level of the damper (S12). The control unit 51 determines whether the probability of rank V is equal to or greater than a first threshold (S13).

[0042] If the probability of rank V is equal to or greater than the first threshold (YES in S13), the control unit 51 sets the deterioration level of the damper to rank V (S14) and ends the process. If the probability of rank V is not equal to or greater than the first threshold (NO in S13), the control unit 51 determines whether the probability of rank IV is equal to or greater than the second threshold (S15).

[0043] If the probability of rank IV is equal to or greater than the second threshold (YES in S15), the control unit 51 sets the deterioration level of the damper to rank IV (S16) and ends the process. If the probability of rank IV is not equal to or greater than the second threshold (NO in S15), the control unit 51 sets the rank with the highest probability among the probabilities of ranks I to III as the deterioration level of the damper (S17) and ends the process.

[0044] As described above, according to this embodiment, the failure rate for rank IV is 5%, the failure rate for rank V is 3%, and the false alarm rate for ranks I to III is 10% or less, which is sufficient for practical use.

[0045] The computer program of this embodiment causes a computer to execute processing to acquire an image of an accessory device attached to a ground wire or electric cable, input the acquired image into a learning model that outputs the accuracy of the deterioration level of the accessory device when the image of the accessory device attached to the ground wire or electric cable is input, acquire the accuracy of the deterioration level of the accessory device captured in the image, and determine whether the deterioration level of the accessory device is a specific deterioration level based on the acquired accuracy and a predetermined threshold value.

[0046] In the computer program of this embodiment, the predetermined threshold can be set according to the specific degradation level.

[0047] The computer program of this embodiment causes the computer to execute a process of determining the deterioration level of the auxiliary device according to the priority of the specific deterioration level.

[0048] In the computer program of this embodiment, the priority corresponds to the degree of abnormality of the specific degradation level.

[0049] The computer program of this embodiment causes a computer to determine the deterioration level of the accessory device based on the probability output by the learning model if the deterioration level of the accessory device is not the specific deterioration level.

[0050] The computer program of this embodiment causes the computer to execute a process of, when the deterioration level of the accessory equipment is a specific deterioration level, displaying a comparison between the maximum value of the accuracy of the deterioration level output by the learning model and the accuracy of the specific deterioration level output by the learning model.

[0051] The deterioration level determination device of this embodiment includes an image acquisition unit that acquires an image of an accessory device of a ground wire or electric cable, an accuracy acquisition unit that inputs the acquired image into a learning model that outputs the accuracy of the deterioration level of the accessory device when an image of an accessory device of a ground wire or electric cable is input, and acquires the accuracy of the deterioration level of the accessory device captured in the image, and a determination unit that determines whether the deterioration level of the accessory device is a specific deterioration level based on the acquired accuracy and a predetermined threshold value.

[0052] The degradation level determination method of this embodiment acquires an image of an accessory device for a ground wire or electric cable, inputs the acquired image into a learning model that outputs the accuracy of the degradation level of the accessory device when the image of the accessory device for a ground wire or electric cable is input, acquires the accuracy of the degradation level of the accessory device captured in the image, and determines whether the degradation level of the accessory device is a specific degradation level based on the acquired accuracy and a predetermined threshold value. [Explanation of symbols]

[0053] 1. Communication Network 10 Data Server 20 Terminal equipment 50 Deterioration level determination device 51 Control section 52 Communications Department 53 Memory 54 Deterioration level determination unit 55 Output section 56 Memory section 57 Computer Programs 58 Learning Model

Claims

1. On the computer, Acquire an image of the ground wire or the attached equipment of the electric wire; inputting the acquired image into a learning model that outputs the accuracy of the deterioration level of the accessory when an image of an accessory of a ground wire or an electric wire is input, and acquiring the accuracy of the deterioration level of the accessory captured in the image; determining whether the deterioration level of the accessory device is a specific deterioration level based on the acquired accuracy and thresholds associated with each of a plurality of deterioration levels of the accessory device, in accordance with a priority order set based on the plurality of deterioration levels; A computer program that executes a process.

2. The threshold value can be set according to the particular level of degradation.

2. The computer program of claim 1.

3. The priority corresponds to the degree of abnormality of the particular deterioration level.

3. A computer program according to claim 1 or claim 2.

4. On the computer, If the deterioration level of the accessory device is not the specific deterioration level, the deterioration level of the accessory device is determined based on the probability output by the learning model. The computer program according to any one of claims 1 to 3, which causes a process to be executed.

5. On the computer, When the deterioration level of the accessory device is a specific deterioration level, a maximum value of the accuracy of the deterioration level output by the learning model is displayed in comparison with the accuracy of the specific deterioration level output by the learning model.

5. A computer program product according to claim 1, which causes a process to be executed.

6. an image acquisition unit that acquires an image of the ground wire or an attachment device of the electric wire; an accuracy acquisition unit that inputs an acquired image into a learning model that outputs an accuracy of a deterioration level of an accessory device when an image of an accessory device of a ground wire or an electric wire is input, and acquires an accuracy of a deterioration level of the accessory device captured in the image; a determination unit that determines whether the deterioration level of the accessory device is a specific deterioration level in accordance with a priority order set based on the plurality of deterioration levels, based on the acquired accuracy and thresholds associated with each of the plurality of deterioration levels of the accessory device; Equipped with Deterioration level determination device.

7. Acquire an image of the ground wire or the attached equipment of the electric wire; inputting the acquired image into a learning model that outputs the accuracy of the deterioration level of the accessory when an image of an accessory of a ground wire or an electric wire is input, and acquiring the accuracy of the deterioration level of the accessory captured in the image; determining whether the deterioration level of the accessory device is a specific deterioration level based on the acquired accuracy and thresholds associated with each of a plurality of deterioration levels of the accessory device, in accordance with a priority order set based on the plurality of deterioration levels; Deterioration level determination method.

Images