Electric wire sag control device and electric wire sag control method

Abstract

To provide a wire slackness control device and a wire slackness control method capable of easily and highly accurately setting the slackness of a wire within the range of designed slackness and reducing the burden at the time of introduction.SOLUTION: A wire slackness control device 1 that controls the amount of incoming and outgoing wires held by holding means such that the slackness of the wires supported by a transmission tower is within a specified range includes an acquisition unit 2 that acquires slackness-related data related to slackness and feature data consisting of an in-and-out amount corresponding to the slackness-related data, a learning unit 3 that constructs a learning model for estimating a new in-and-out amount corresponding to new slackness-related data from the feature data, an estimation unit 4 that estimates a new in-and-out amount corresponding to the new slackness-related data on the basis of the learning model constructed in the learning unit 3, and an execution unit 5 that executes driving of the holding means 52 on the basis of the new in-and-out amount estimated by the estimation unit 4.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a wire sag control device and a wire sag control method for controlling the sag of wires supported by a plurality of transmission towers, and in particular to a wire sag control device and a wire sag control method that can easily and accurately keep the sag within the design sag range by automatically controlling the amount of wire that enters and exits from a wire held by a holding means. [Background technology]

[0002] Conventionally, when extending electric wires on a transmission tower, slack is provided in the electric wires between multiple transmission towers, taking into consideration the height of the electric wires and the strength of the supports that support the electric wires. For example, the work procedure for this would involve a worker climbing up one transmission tower and placing a pocket compass at a position that is a distance below the wire support point on that tower that is equal to the predetermined sag. Similarly, a vertex would be attached to the other transmission tower at a position that is a distance below the wire support point on that tower that is equal to the predetermined sag. The worker would then measure the lowest position of the wire while looking through the telescope of the pocket compass, and adjust the amount of winding on the wire drawer around which the leading end of the wire is wound, and the amount of winding on the wire drum around which the base end of the wire is wound, so that this lowest position coincides with the vertex. However, the above procedure takes time to complete. Furthermore, the worker is required to measure the lowest position of the wire while maintaining the same posture throughout the wire stringing process, which places a heavy burden on the worker, especially during hot summers and cold winters. Furthermore, the vertex may be difficult to see due to obstacles, which can lead to errors in the slack of the wire. In order to solve this problem, techniques have been developed for easily and accurately adjusting the slack within a specified range, and several inventions relating to this have already been disclosed.

[0003] Patent Document 1, entitled "Method for measuring sag of power transmission line, method for fastening wire, and tool," discloses an invention that can measure the sag of a power transmission line with high accuracy and fasten the wire so that it has the required sag. The invention disclosed in Patent Document 1 is characterized in that a receiving antenna is installed on the electric wire whose sag is to be measured, the receiving antenna receives a signal transmitted from a satellite, and the three-dimensional position of each antenna is determined by calculating and processing the phase difference of the received wave at the installation position of each antenna, thereby actually measuring the sag of the electric wire to which the antenna is attached, and if there is an excess or deficiency between this measured sag and the desired design sag, an excess or deficiency signal is input to the winding control device of the tensioning winch. In an invention having such characteristics, as long as the receiving antenna is installed on the electric wire, sag measurement can be performed regardless of day or night or weather conditions. Moreover, installation of the receiving antenna on the electric wire can be made very easy by using a tool that allows the base on which the receiving antenna is placed to be moved along the electric wire. Therefore, according to the invention disclosed in Patent Document 1, it is possible to measure sag with extremely high accuracy by using signals transmitted from an artificial satellite. Furthermore, it is possible to reduce the burden on workers and significantly shorten the time required for measurement compared to conventional methods.

[0004] Patent Document 2 discloses an invention entitled "Protection Wire Sag Adjustment Device" that is capable of adjusting the sag of a protection wire so that the sag of the protection wire approaches that of a power transmission line. The invention disclosed in Patent Document 2 is characterized by having a data collection means for collecting data that are factors that change the sag of an installed transmission line; an adjustment means attached to a protection line installed at a predetermined interval below the transmission line and for adjusting the length of the installed protection line; and a sag adjustment control means for estimating the sag of the transmission line based on the collected data, controlling the drive of the adjustment means so that the sag of the protection line approaches the estimated sag, and adjusting the length of the protection line. In the invention having such characteristics, the sag of the power line is estimated by correlating the sag of the power line with the collected data, and the sag of the protection line is adjusted in accordance with the estimated sag of the power line, thereby maintaining the clearance between the power line and the protection line. Therefore, by modifying the design so that the adjustment means is driven and controlled so as to adjust the sag based on the estimated sag of the power line, the sag of the power line can be easily and automatically controlled. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 2-306102 [Patent Document 2] Japanese Patent Application Laid-Open No. 2011-151939 Summary of the Invention [Problem to be solved by the invention]

[0006] However, the invention disclosed in Patent Document 1 requires a receiving antenna to receive the signal transmitted from the satellite and a new tool to move the receiving antenna along the electric wire. This may result in high initial costs and difficulty in implementation. Furthermore, poor weather conditions may cause noise to be mixed into the signal transmitted from the satellite, which may result in reduced accuracy of sag measurement.

[0007] Furthermore, in the invention disclosed in Patent Document 2, it is necessary to obtain the relationship between the conductor temperature and the sag of the transmission line in advance by actual measurement. However, this relationship differs depending on the structure of the transmission line, and therefore it is extremely time-consuming to measure the relationship for each of the transmission lines with different structures.

[0008] The present invention has been made in response to the above-mentioned conventional circumstances, and aims to provide an electric wire sag control device and an electric wire sag control method that can easily and accurately keep the sag within the design sag range without the need to take the time to actually measure the numerical data required for measuring the sag, and that can reduce the burden at the time of installation and operation. [Means for solving the problem]

[0009] In order to achieve the above-mentioned object, the first invention is an electric wire sag control device that controls the amount of inflow and outflow of an electric wire held by a holding means so that the sag of the electric wire supported on a plurality of transmission towers is within a specified range, and is equipped with an acquisition unit that acquires sag-related data related to the sag and feature data consisting of the amount of inflow and outflow corresponding to the sag-related data, and a learning unit that constructs a learning model from the feature data to estimate a new amount of inflow and outflow corresponding to the new sag-related data, and is characterized in that the sag-related data includes at least one of image data of the electric wire, design sag representing a specified range, load per unit length of the electric wire, maximum tension of the electric wire, actual length of the electric wire, span length of a plurality of transmission towers, vertical height of each support point of a plurality of transmission towers, elevation difference between each support point of a plurality of transmission towers, and outside air temperature around the electric wire.

[0010] In the first invention configured as described above, the sag of the electric wire refers to the maximum vertical distance between the straight line connecting the support points of the multiple transmission towers and the curve formed by the electric wire. The straight line connecting the support points may be parallel to the horizontal direction or may be inclined relative to the horizontal direction. The span length refers to the length of the straight line connecting the support points of the multiple transmission towers when projected horizontally. The holding means specifically refers to a wire drum around which the base end of the wire is wound, or a wire drawing machine around which the tip end of the wire is wound, which are used when extending the wire to a transmission tower, for example. Therefore, the amount of wire drawn in and out is the length of the wire drawn out from or wound up on the wire drum or wire drawing machine. Furthermore, sag-related data may include, for example, image data of the electric wire, the load per unit length of the electric wire, each design value or each measured value of the height difference of each support point of the transmission tower, and each measured value of the outside air temperature.

[0011] In the present invention, at least one of the sag-related data or the amount of electrical wires corresponding to the sag-related data is considered as an input, and the amount of electrical wires is considered as an output. However, the sag-related data may include the amount of electrical wires. Note that "at least one" in the sag-related data refers to the number of types of physical quantities included in the sag-related data. In the first invention having the above configuration, a learning model is constructed in the learning unit using feature data consisting of sag-related data and inflow / outflow amounts acquired by the acquisition unit. That is, a learning model is constructed that can estimate new inflow / outflow amounts of electric wires corresponding to, for example, design values ​​such as the span length of a transmission tower, actual measured values ​​of outside air temperature, new image data of the electric wires, etc. Such a learning model is constructed by machine learning, and as this machine learning, for example, supervised learning, unsupervised learning, or reinforcement learning is used.

[0012] Next, the second invention is characterized in that, in the first invention, the slackness-related data includes at least image data, and the learning unit is equipped with an association unit that associates the slackness-related data with input and output amounts to generate association data, and constructs a learning model that estimates new input and output amounts corresponding to the new slackness-related data based on the association data. In the second invention having such a configuration, the learning model is constructed by supervised learning using deep learning technology. That is, the learning model is constructed when input data including at least image data and output data corresponding to the input data are given in advance.

[0013] In the second invention having such a configuration, the associated data is, for example, a data group consisting of a plurality of image data and a plurality of inflow / outflow amounts corresponding to the plurality of image data, or a data group consisting of a plurality of sag-related data with various selected types and values, a plurality of image data corresponding to the plurality of sag-related data, and a plurality of inflow / outflow amounts of electric wires corresponding to the plurality of image data. In the second invention having the above configuration, in addition to the effects of the first invention, information useful for determining the amount of input and output of the electric wire is extracted from the association data, and a learning model is constructed based on this information that can estimate new input and output amounts corresponding to the new sag-related data.

[0014] Furthermore, the third invention is characterized in that in the first invention, the learning unit includes a reward unit that determines a reward for the result of selecting the input / output amount, and a function update unit that updates a function that represents the value of selecting the input / output amount based on the reward, and by repeatedly updating the function in the function update unit, a learning model is constructed that estimates new input / output amounts so that the reward determined by the reward unit is maximized. In the third invention having such a configuration, the learning model is constructed by reinforcement learning. That is, the learning model is for a case where the answer of the output data, which is the amount of input and output of the electric wire, is not given in advance. In addition, the reward determined by the reward unit is the quantity obtained when the subject of action performs a specific action in a certain state, and can be set to a large value when the accuracy of the desired output data is high, and a small value when the accuracy of the output data is low, for example. Furthermore, the function updated in the function update unit is a function called an action value function, which indicates the value of selecting an action under a certain condition using a reward. Therefore, in the third invention, the function is repeatedly updated by repeating trial and error to select various actions under any given state, and the action that maximizes the value of this function is selected to be the optimal solution for the output data.

[0015] In the third invention having the above configuration, in addition to the function of the first invention, the learning unit constructs a learning model as described above. For example, if the action is to select the amount of input / output of an electric wire, the most appropriate amount of input / output is finally determined after trial and error. Therefore, the determined most appropriate amount of input / output is used as the new amount of input / output. Note that an arbitrary value is used as the initial amount of input / output before trial and error.

[0016] Next, the fourth invention is characterized in that in the third invention, the sag-related data includes at least a combination of load and maximum tension and / or design sag, span length, vertical height and / or elevation difference, and actual length, and the reward is set so that it increases as the sag corresponding to the selected inflow / outflow amount approaches the design sag. In the fourth invention having such a configuration, in addition to the effect of the third invention, as the function is updated, the direction in which the reward increases, i.e., the sag is selected so that it increases as it approaches the design sag. Therefore, when the most appropriate amount of wire in and out is finally determined, it can be assumed that the wire sag is within the specified range.

[0017] Furthermore, the fifth invention is characterized in that, in any of the first to fourth inventions, it comprises an estimation unit that estimates new inflow and outflow amounts corresponding to new slack-related data based on a learning model constructed by the learning unit, and an execution unit that drives the holding means based on the new inflow and outflow amounts estimated by the estimation unit, and the execution unit comprises a signal generating unit that generates a drive signal that reflects the new inflow and outflow amounts estimated by the estimation unit, a drive unit that drives the holding means, and a transmission unit that transmits the drive signal to the drive unit. In the fifth invention having such a configuration, in addition to the action of any one of the first to fourth inventions, the estimation unit estimates a new inflow / outflow amount corresponding to the new sag-related data, i.e., the inflow / outflow amount of the electric wire to be determined. Next, the drive signal generated by the signal generating unit of the execution unit is transmitted to the drive unit by the transmitting unit, thereby driving the holding means. This drive signal reflects the new inflow / outflow amount estimated by the estimating unit, so the drive of the holding means realizes the new inflow / outflow amount.

[0018] The sixth invention is a wire sag control method for controlling the amount of inflow and outflow of a wire held by a holding means so that the sag of the wire supported on multiple transmission towers is within a specified range, and includes an acquisition process for acquiring sag-related data related to the sag and feature data consisting of the amount of inflow and outflow corresponding to the sag-related data, and a learning process for constructing a learning model from the feature data for estimating a new amount of inflow and outflow corresponding to the new sag-related data, wherein the sag-related data includes at least one of image data of the wire, design sag representing the specified range, load per unit length of the wire, maximum tension of the wire, actual length of the wire, span length of multiple transmission towers, vertical height of each support point of multiple transmission towers, elevation difference between each support point of multiple transmission towers, and outside air temperature around the wire.

[0019] In the fifth aspect of the invention having such a configuration, the acquisition step and the learning step are respectively executed by the acquisition section and the learning section constituting the first aspect of the invention, and the same effects as those of the first aspect of the invention are exerted. [Effects of the Invention]

[0020] According to the first invention, the new amount of incoming and outgoing electric wires can be obtained without climbing up the transmission tower and without the hassle of obtaining the necessary numerical data, so that the sag can be easily kept within the specified range and the burden at the time of installation and operation can be reduced. Furthermore, the learning unit constructs a learning model that can estimate the new wire inflow / outflow amount corresponding to the new image data of the wire from, for example, the image data of the wire and the corresponding wire inflow / outflow amount, and by using this learning model, the new wire inflow / outflow amount can be accurately determined. Therefore, the wire sag can be kept within the specified range with high accuracy.

[0021] According to the second invention, in addition to the effects of the first invention, a learning model can be constructed that can estimate new wire inflow / outflow amounts corresponding to new sag-related data from information useful for determining wire inflow / outflow amounts, which is automatically extracted from the association data. Therefore, by using this learning model, for example, it is no longer necessary for an operator to operate the holding means while observing the wire when extending it, and the sag can be accurately kept within the designed sag range. Therefore, work efficiency can be significantly improved compared to the prior art.

[0022] According to the third invention, in addition to the effects of the first invention, the most appropriate amount of wire inlet and outlet can be finally determined based on the initial arbitrary amount of wire inlet and outlet, and therefore, similar to the second invention, work efficiency can be significantly improved compared to conventional techniques. Furthermore, since the most appropriate wire input / output amount obtained by the learned learning model is used as the new input / output amount, the reliability of the obtained new input / output amount can be increased.

[0023] According to the fourth invention, in addition to the effects of the third invention, when the most appropriate amount of wire in and out is finally determined, it is estimated that the slack of the wire is within the specified range, so that the designed slack can be achieved without actually measuring the slack.

[0024] According to the fifth invention, in addition to the effects of any one of the first to fourth inventions, the new amount of wire inflow and outflow estimated by the estimation unit is realized in the holding means by the execution unit, so that the inflow and outflow of the wire can be freely controlled and the slack can be kept within a specified range.

[0025] According to the sixth aspect of the invention, the same effects as those of the first aspect of the invention are achieved. [Brief explanation of the drawings]

[0026] [Figure 1] (a) is an explanatory diagram for explaining the outline of the work of extending an electric wire, and is an external view seen from a horizontal direction approximately perpendicular to the running direction of the electric wire, and (b) is an enlarged view of a part of (a). [Figure 2]1 is a configuration diagram of a wire slack control device according to a first embodiment of the present invention. [Figure 3] FIG. 2 is a flowchart showing a procedure for estimating the amount of wire inflow and outflow by the wire slack control device according to the first embodiment of the present invention. [Figure 4] FIG. 6 is a configuration diagram of a wire slack control device according to a second embodiment of the present invention. [Figure 5] FIG. 10 is a flowchart showing a procedure for estimating the amount of wire inflow and outflow by the wire slack control device according to the second embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION [Example]

[0027] First, an overview of the work of extending an electric wire will be explained using Fig. 1. Fig. 1(a) is an explanatory diagram for explaining the overview of the work of extending an electric wire, and is an external view when viewed from a horizontal direction approximately perpendicular to the running direction of the electric wire, and Fig. 1(b) is an enlarged view of a part of Fig. 1(a). 1(a) and 1(b), when an electric wire 51 is stretched on transmission towers 50A to 50C, the electric wire 51, whose base end is wound around one holding means 52, is stretched at support points 50a to 50c of the transmission towers 50A to 50C, and the tip of the wound electric wire 51 is wound around the other holding means 52. One holding means 52 is an electric wire drum 53, and the other holding means 52 is a wire stretching machine 54. When the wire drawing machine 54 is loaded onto, for example, a vehicle (not shown) and driven in the direction of the outline arrow, the electric wire 51 is supported with a certain degree of slack between support points 50a and 50b and between support points 50b and 50c of the transmission towers 50A to 50C. Note that the support points 50a to 50c are set at arbitrary positions (not shown) based on the support arms of the transmission towers 50A to 50C to which the insulators are attached. Next, when attention is paid to the electric wire 51 between the support points 50a and 50b, as shown in FIG. 1(b), the slack d (m) representing the slack of the electric wire 51 is calculated by dividing the slack d (m) by the straight line L connecting the support points 50a and 50b. 50The span length S (m) is the maximum value of the vertical distance (distance along the vertical direction V) between the wire 51 and the straight line L 50 is the length of the straight line formed when projecting onto the horizontal direction H.

[0028] In addition, the outside temperature T e The actual temperature measured by the thermometer 2c (see Figure 2) can be used as the temperature (°C). e Since it expands and contracts depending on the change in the outside temperature T e This is to reflect the change in slack caused by the change in the new inflow and outflow amounts.

[0029] Furthermore, if the vertical height from ground surface G to support point 50a of transmission tower 50A is h1 (m), and the vertical height from ground surface G to support point 50b of transmission tower 50B is h2 (m), then the height difference h (m) is the difference between h1 and h2. Because both vertical heights h1 and h2 are known, the height difference h can be immediately calculated. The load W (N / m) per unit length of the electric wire 51 and the maximum tension T of the electric wire 51 are max (N) is a design value for each type of electric wire 51, such as the structure and material of electric wire 51.

[0030] Next, the configuration of the electric wire slack control device according to the first embodiment of the present invention will be described with reference to Fig. 2. Fig. 2 is a configuration diagram of the electric wire slack control device according to Example 1. Note that the components shown in Fig. 1 are given the same reference numerals in Fig. 2, and their description will be omitted. As shown in Figure 2, the electric wire sag control device 1 of Example 1 is an electric wire sag control device that controls the inflow / outflow amount ΔL (m) of the electric wire 51 held by the holding means 52 so that the sag d of the electric wire 51 supported by the transmission towers 50A to 50C is within a specified range, and is equipped with an acquisition unit 2, a learning unit 3, an estimation unit 4, and an execution unit 5.

[0031] Among these, the acquisition unit 2 acquires slackness-related data D related to the slackness d. R and slack-related data D RThe acquisition unit 2 acquires the characteristic data D consisting of the amount of inflow / outflow ΔL of the electric wire 51 corresponding to the sag. R an input unit 2a including a storage medium, a keyboard, or an internet site into which image data D I and an image capturing means 2b capturing an image of the outside temperature T e The input unit 2a, the imaging means 2b, and the thermometer 2c are communicably connected to one another.

[0032] Here, the sag-related data D R is the image data D of the electric wire 51. I and a given outside temperature T e Design sag d0, load W, and maximum tension T max , the span length S, the actual length L of the electric wire 51, the vertical heights h1 and h2, the height difference h, and the outside temperature T e It can include at least one of: In addition, slack related data D R Among them, image data D I It is desirable that the image be taken from one direction substantially perpendicular to the running direction of the electric wire 51 so that the overall shape of the electric wire 51 can be grasped in each section of the support points 50a to 50c. However, since the vertical heights h1 and h2 of the support points 50a and 50b are both known, the bases of the transmission towers 50A and 50B are taken from the image data D I The image does not have to be captured within the image. However, depending on the installation positions of the transmission towers 50A to 50C, the vertical heights h1 and h2 of the diameter, or the span length S, it may be difficult to grasp the overall shape of the electric wire 51 by imaging from one direction. In this case, the image data D I A plurality of images taken from different viewpoints may be used as the plurality of images. The plurality of images may be, for example, still images or videos captured from any oblique direction relative to the running direction of the electric wire 51 using the imaging means 2b installed on the ground or a drone equipped with the imaging means 2b. The shape of the electric wire 51 is restored as three-dimensional point cloud data from the plurality of images using, for example, publicly available three-dimensional restoration software.

[0033] The actual length L (m) of the electric wire 51 is the length of the electric wire 51 in each section between the support points 50a to 50c. Furthermore, the in / out amount ΔL (m) of the electric wire 51 is the length of the electric wire 51 that is drawn out or wound on the electric wire drum 53 or the wire drawing machine 54, and is a value that allows the slack d of the drawn electric wire 51 to fall within a specified range.

[0034] In Example 1, the slack-related data D R is the image data D of the electric wire 51. I and image data D I Other slack related data D R This image data D I Other slack related data D R is known, and the image data D I Image data D I Other slack related data D R The image was captured in response to changes in However, the actual length L of the electric wire 51 is difficult to measure and the image data D I The sag-related data D R Furthermore, the in / out amount ΔL of the electric wire 51 can be measured in advance in the electric wire drum 53 or the wire drawing machine 54 using, for example, an encoder that measures the amount of the electric wire 51 that is drawn out or wound up.

[0035] Next, the learning unit 3 includes an associating unit 6, a constructing unit 7, and a storage unit 8, and generates new slack-related data D from the feature data D acquired by the acquiring unit 2. RN New input / output volume ΔL corresponding to N We build a learning model to estimate In the learning unit 3, the associating unit 6 receives the slack-related data D R (i.e., image data D I and image data D I Other slack related data D R and the amount of input / output ΔL of the electric wire 51 are associated with each other to generate association data D A Generate. Specifically, the slack-related data D processed by the association unit 6 R is image data D I Other slack related data D R The image data D I is a set of multiple slack-related data D R Therefore, the association data D A is a set of multiple image data D I The data group is made up of multiple pieces of data corresponding to the above. The construction unit 7 also constructs the association data D A The learning model is constructed using a known deep learning technique based on the above. A The learning model constructed by the construction unit 7 is also stored.

[0036] Furthermore, the estimation unit 4 generates new slack-related data D based on the learning model constructed by the construction unit 7 of the learning unit 3. RN New input / output volume ΔL corresponding to N Estimate. Next, the execution unit 5 calculates the new input / output amount ΔL estimated by the estimation unit 4. N The holding means 52 is driven based on the new inflow / outflow amount ΔL estimated by the estimation unit 4. N a signal generating unit 5a that generates a drive signal that reflects the force of the electric wire, a drive unit 5b that drives the holding means 52, and a transmitter 5c that transmits the drive signal generated by the signal generating unit 5a to the drive unit 5b. The drive unit 5b is configured to rotate the rotation shaft of a drive motor (not shown) of at least one of the electric wire drum 53 and the wire drawing machine 54 that are the holding means 52.

[0037] Furthermore, the operation of the electric wire slack control device 1 will be described with reference to Fig. 3. Fig. 3 is a flow chart showing the procedure for estimating the amount of incoming and outgoing electric wire by the electric wire slack control device according to the first embodiment of the present invention. Note that the same reference numerals are used in Fig. 3 to denote the components shown in Figs. 1 and 2, and the description thereof will be omitted. As shown in Figure 3, in the electric wire slack control device 1, an electric wire slack control method 9 is implemented, which controls the inflow / outflow amount ΔL of the electric wire 51 held by the electric wire drum 53 and the wire pulling machine 54 so that the slack d of the electric wire 51 supported by the transmission towers 50A to 50C is within a specified range. This electric wire sag control method 9 includes an acquisition step of step S1, a learning step of step S2, an estimation step of step S3, and an execution step of step S4. Of these, the acquisition step of step S1 includes a new sag-related data acquisition step of step S1-1 and a sag-related data, etc. acquisition step of step S1-2. Furthermore, the learning step of step S2 includes an association step of step S2-1, a construction step of step S2-2, and a storage step of step S2-3.

[0038] First, in the acquisition step of step S1, the acquisition unit 2 acquires new slack-related data D RN Then, a new sag-related data acquisition step S1-1 is executed to acquire the sag-related data. In the new sag-related data acquisition process of step S1-1, new image data D of the electric wire 51 captured immediately before the sag d is adjusted is acquired. IN and new image data D IN New slack-related data other than D RN At least one of the new slack-related data D RN is the slack-related data D R Similarly, given the outside temperature T e Design sag d0, load W, and maximum tension T max , the span length S, the actual length L of the electric wire 51, the vertical heights h1 and h2, the height difference h, and the outside temperature T e It can include at least one of: In addition, the outside temperature T e New slack-related data other than D RN The temperature is acquired by inputting it into the input unit 2a by the worker and then transmitting it to the acquisition unit 2. e Alternatively, the measured value by the thermometer 2c may be automatically transmitted to the acquisition unit 2 at a predetermined timing.

[0039] Furthermore, immediately after the new sag-related data acquisition step in step S1-1, the user selects whether to proceed directly to the sag-related data acquisition step in step S1-2 (Yes) or to the estimation step in step S3 (No). This is because, for example, if the learning step in step S2 has already been executed once between the support points 50a and 50c to construct a learning model, it is necessary to select whether to acquire new sag-related data D again between the same support points 50a and 50c. RN New input / output volume ΔL corresponding to N This is because when calculating the feature data D required to construct a learning model, it is not necessary to obtain the feature data D. That is, when proceeding to the step of acquiring sag-related data, etc., of S1-2, the learning step of step S2, the estimation step of step S3, and the execution step of step S4 are executed. On the other hand, when proceeding to the estimation step of step S3, the step of acquiring sag-related data, etc., of S1-2 and the learning step of step S2 are omitted, and then the estimation step of step S3 and the execution step of step S4 are executed. Note that the above selection is input by the worker to the input unit 2a and then transmitted to the acquisition unit 2.

[0040] When proceeding to the step of acquiring slack-related data etc. in S1-2, the acquisition unit 2 acquires image data D I and image data D I Other slack related data D R and at least one of the slack-related data D R As described above, this feature data D is data necessary for constructing a learning model, and is called teacher data in supervised learning. In addition, the outside temperature T e Other slack related data D R The acquisition of the outside temperature T e New slack-related data other than D RN The same applies to the acquisition of Furthermore, the predetermined outside temperature T e The design sag d0 in the wire 51 is calculated by the load W per unit length of the wire 51 and the maximum tension T of the wire 51 as shown in the following formula (1): MAXis set to a constant value calculated from the span length S. However, the design sag d0 may be a value having a range determined by practical requirements.

[0041]

number

[0042] However, in Example 1, the design sag d0 is not a physical quantity estimated by a learning model or used to determine the accuracy of the wire inlet / outlet amount ΔL, or is calculated from the load W and the maximum tension T MAX Therefore, the design sag d0 is not necessarily determined by the sag-related data D R It does not have to be included in Also, the load W and the maximum tension T MAX are the design values ​​for each type of electric wire 51 for which the sag d is to be adjusted. Furthermore, the span length S, the vertical heights h1 and h2 of the support points 50a and 50b, and the height difference h remain unchanged unless the structure or arrangement of the transmission towers 50A and 50B supporting the electric wire 51 is changed. Note that the height difference h is the difference between the vertical height h1 and the vertical height h2, and therefore the vertical heights h1 and h2 are used as the sag-related data D R If it is included in the slack related data D R It does not have to be included in

[0043] Next, in the learning step of step S2, first, the associating section 6 of the learning section 3 executes the associating step of step S2-1. In the association process of step S2-1, image data D I and this image data D I Other slack related data D R Associate one of them with the known inflow / outflow amount ΔL to obtain the association data D A As mentioned above, the association data D A means multiple image data D I The data group is made up of multiple pieces of data corresponding to the above.

[0044] Then, in the construction step of step S2-2, the construction unit 7 constructs a plurality of association data D A Based on this, the new input / output volume ΔL N Then, the construction unit 7 extracts information useful for determining the sag related data D based on the extracted information. RN New input / output volume ΔL corresponding to N We build a learning model to estimate Furthermore, in the storage step of step S2-3, the storage unit 8 stores the plurality of association data D generated in the association step of step S2-1. A The learning model constructed in the construction process of step S2-2 is stored.

[0045] Next, in the estimation step of step S3, the estimation unit 4 calculates the new slack-related data D acquired in the new slack-related data acquisition step of step S1-2 based on the learning model constructed in the construction step of step S2-2. RN New input / output volume ΔL corresponding to N Estimate. As a result, for example, the outside air temperature T measured by the thermometer 2c is measured for the electric wire 51 newly laid between the support points 50a and 50b or between the support points 50b and 50c. e The amount of wire drawn out and the amount of wire taken up by the electric wire drum 53 and the wire drawing machine 54 are calculated.

[0046] Furthermore, in the execution process of step S4, a new inflow / outflow amount ΔL N A drive signal reflecting the new inflow / outflow amount ΔL is generated by the signal generating unit 5a and transmitted to the driving unit 5b via the transmitting unit 5c. N The outside temperature T measured by the thermometer 2c e The slack d of the electric wire 51 at this point is set within a specified range.

[0047] As described above, according to the electric wire sag control device 1 of the first embodiment, it is not necessary to measure the sag d when extending the electric wire 51, so that the work of climbing up the transmission towers 50A to 50C can be eliminated.R Furthermore, the new inflow / outflow amount ΔL of the electric wire 51 can be obtained without much effort. N By estimating based on the learning model, it is possible to easily keep the sag d within the specified range. Therefore, the burden on installing and operating the wire sag control device 1 can be reduced.

[0048] In detail, according to the electric wire slack control device 1, a plurality of association data D A New slack-related data D is generated from the information automatically extracted from RN New input / output volume ΔL corresponding to N Therefore, by using this learning model, for example, when extending a wire, it is not necessary for an operator to operate the holding means 52 while observing the wire 51, and the sag d can be accurately kept within the specified range. Therefore, it is possible to significantly improve workability compared to conventional techniques. Furthermore, it is possible to eliminate the influence of sag reading errors and variations between operators that can occur when a person is involved in sag measurement. In addition, according to the electric wire sag control device 1, the new inflow / outflow amount ΔL of the electric wire 51 estimated by the estimation unit 4 is N However, since this is realized by the execution unit 5 in the holding means 52, the entry and exit of the electric wire 51 can be freely controlled, and the slack d can be kept within a specified range. [Example]

[0049] The configuration of a wire slack control device according to a second embodiment of the present invention will be described with reference to Fig. 4. Fig. 4 is a configuration diagram of a wire slack control device according to Example 2. Note that the components shown in Fig. 1 are given the same reference numerals in Fig. 2, and their description will be omitted. 2, the wire slack control device 10 according to the second embodiment includes a learning unit 11 instead of the learning unit 3 constituting the wire slack control device 1 according to the first embodiment. Other configurations of the wire slack control device 10 are the same as those of the wire slack control device 1. The learning unit 11 includes a reward unit 12, a function update unit 13, and a storage unit 14. Among these, the reward unit 12 determines a reward for the result of selecting the input / output amount ΔL. In addition, the storage unit 14 stores the learning model learned by the function update unit 13. Furthermore, the function update unit 13 updates the function that represents the value of selecting the input / output amount ΔL based on the reward. This function is generally an action value function Q expressed by the following equation (2).

[0050]

number

[0051] The action value function Q expressed by equation (2) is used in Q-learning to indicate the value of an action when an agent selects action a in state s. Therefore, by repeating trial and error to select various actions a under an arbitrary state s according to equation (2), the action value function Q is repeatedly updated, and selecting the action that maximizes the value of this action value function Q becomes the output data, that is, the optimal solution for the input / output volume ΔL in this application. N The learning coefficient α and discount rate γ are usually set to 0.1 and 0.9, respectively. Also, reward r t+1 is preset so that the sag d corresponding to the inflow / outflow amount ΔL increases as it approaches the design sag d0, which indicates that it is within the specified range.

[0052] Next, the operation of the electric wire slack control device 10 will be described with reference to Fig. 5. Fig. 5 is a flow chart showing the procedure for estimating the amount of incoming and outgoing electric wire by the electric wire slack control device according to the second embodiment of the present invention. Note that the same reference numerals are used in Fig. 5 to denote the components shown in Figs. 1 to 4, and the description thereof will be omitted. As shown in Figure 5, in the wire slack control device 10, a wire slack control method 15 is implemented to control the inflow / outflow amount ΔL of the wire 51 held by the wire drum 53 and the wire drawing machine 54, similar to the wire slack control method 9 implemented in Example 1. The wire sag control method 15 includes a learning step of step S5 instead of the learning step of step S2 of the wire sag control method 9. The characteristic data D acquired in the acquisition process of step S1 of the electric wire slack control method 15 includes the load W and the maximum tension T max combination of and / or a given outside temperature T e The design sag d0, the span length S, the vertical height h1, h2 and / or the height difference h, the actual length L of the electric wire 51, and the measured outside air temperature T e Sag-related data D including R and the inflow / outflow amount ΔL of the electric wire 51. However, in the acquisition step of step S1 in the electric wire sag control method 15 of the second embodiment, the new sag-related data acquisition step of step S1-1 in the electric wire sag control method 9 of the first embodiment is omitted. Also, immediately after the acquisition step of step S1, the process proceeds directly to the learning step of step S5.

[0053] Of these, the specified outside temperature T e the outside temperature T e0 Then, the load W and the maximum tension T max The combination of and the outside temperature T e0 The design sag d0 in at least one of the sag related data D R This means that the load W and the maximum tension T max If the combination of these is known, the design sag d0 can be obtained by equation (1). Also, if at least one of the vertical heights h1, h2 and the height difference h is included in the sag-related data D R The actual length L and the amount of extension / retraction ΔL of the electric wire 51 are arbitrary values ​​for setting a predetermined range, which will be described later. Other steps constituting the wire sag control method 15 are similar to the steps constituting the wire sag control device 1.

[0054] The learning process in step S5 includes a reward process in step S5-1 and a function update process in step S5-2. Among these, the reward process of step S5-1 is to calculate the reward r for the result of selecting the actual length L of the electric wire 51 and the input / output amount ΔL. t+1 Determine. Specifically, in the compensation process of step S5-1, an actual length L within a predetermined range and, for example, the minimum value of the inflow / outflow amount ΔL within the predetermined range are selected. After that, any minute amount is sequentially added to the selected minimum value of the actual length L to increase the actual length L. Similarly, for the inflow / outflow amount ΔL within the predetermined range, any minute amount is sequentially added to the selected minimum value of the inflow / outflow amount ΔL to increase the inflow / outflow amount ΔL. Then, for each combination of the increased actual length L and the increased amount of input / output ΔL, the reward r t+1 Specifically, for example, the closer the slack d is to the design slack d0, the higher the reward r t+1 is increased as a larger positive value, and the further the slack d is from the design slack d0, the higher the reward r t+1 can be set to decrease as a larger negative value.

[0055] Here, the relationship between the actual length L of the electric wire 51, the extension / retraction amount ΔL, the sag d before adjustment, the span length S, and the difference in elevation h of the support points is expressed by equation (3). Among these, the actual length L is a value when the electric wire 51 has the design sag d0.

[0056]

number

[0057] Therefore, from equation (3), when the actual length L, the inward / outward displacement ΔL, the span length S, and the difference in elevation h of the support points are given, the sag d (d>0) can be calculated. As mentioned above, the actual length L and the inward / outward displacement ΔL are given values ​​that increase sequentially, and the span length S and the difference in elevation h are both known, so it is possible to determine whether the sag d calculated from equation (3) is close to the design sag d0. Then, depending on the result of this determination, the reward r t+1 The increase or decrease is determined. reward r t+1 After the increase or decrease of is determined, in the function update process of step S5-2, the action value function Q is repeatedly updated according to equation (2), and the optimal solution for the input / output volume ΔL is obtained by selecting the action that maximizes the value of this action value function Q. Therefore, in the estimation process of step S3, the optimal solution for the input / output volume ΔL is calculated as the new input / output volume ΔL.N The actual length L corresponding to the optimal solution of the inflow / outflow amount ΔL is estimated as the new sag-related data D RN Let's say.

[0058] However, the outside temperature T measured by the thermometer 2c e the outside temperature T e1 Then, this outside temperature T e1 However, the design sag d0 is set at a predetermined outside temperature T e0 If different from the above, the design sag d0 to be compared with the sag d is set to the outside temperature T e1 and the outside temperature T e0 It is desirable to make corrections taking into account the difference between Here, the given outside temperature T e0 (℃), the actual length of the electric wire 51 at the design sag d0 (m) is L0 (m), and the measured outside temperature T e1 (℃), the actual length of the electric wire 51 at the design sag d1 (m) is L1 (m), and the outside temperature T e0 is the outside temperature T e1 The temperature change when e (°C) and the linear expansion coefficient of the electric wire 51 is β(1 / °C), the following formulas (4) to (6) are established (S: span length, h: height difference). Therefore, from these formulas (4) to (6), the actually measured outside air temperature T e1 Design sag d1 2 is expressed by equation (7).

[0059]

number

[0060]

number

[0061]

number

[0062]

number

[0063] Therefore, by comparing the sag d on the right side of equation (3) with d1 (d1>0) obtained from equation (7) instead of the design sag d0, the measured outside air temperature T e1 New input / output volume ΔL N is estimated.

[0064] As described above, according to the wire slack control device 10 of the second embodiment, a new inflow / outflow amount ΔL is calculated based on the initial arbitrary actual length L and the inflow / outflow amount ΔL. N In other words, a new inflow / outflow amount ΔL can be calculated without measuring the actual length L or inflow / outflow amount ΔL. N Since it is possible to estimate the slack in the wire, the working efficiency can be significantly improved compared to the prior art, similarly to the wire slack control device 1 of the first embodiment. In addition, the sag d at an arbitrary actual length L and an arbitrary inflow / outflow amount ΔL is determined by determining whether it is close to the design sag d0. N Since the new input / output volume ΔL N The reliability of the system can be improved. Furthermore, the design sag d0 to be compared with the sag d is set at the outside temperature T e1 and the outside temperature T e0 The temperature change ΔT e Since correction can be made taking into account the new inflow / outflow amount ΔL N It is possible to further improve the reliability of the system.

[0065] The electric wire sag control device according to the present invention is not limited to those shown in the examples. For example, in the electric wire sag control device 1 according to the first embodiment, the sag related data D acquired by the acquisition unit 2 is R is image data D I In this case, for example, image data D I The upper electric wire 51 is traced and one image data D I The actual length of the traced electric wire 51 is detected by comparing it with known dimensions of the object captured in the image, such as the span length S or the height of a part of the transmission tower, and the detected actual length is associated with the amount of input and output to generate associated data.I As the image, a three-dimensional image may be used in addition to a two-dimensional image, and a moving image may be used in addition to a still image. Furthermore, the new slack-related data acquisition step of step S1-1 in the acquisition step of step S1 may be executed between the learning step of step S2 and the estimation step of step S3, instead of being executed immediately before the slack-related data, etc. acquisition step of step S1-2. In this case, the system may be configured so that a selection of whether to repeat the new slack-related data acquisition step of step S1-1 or to return to the slack-related data, etc. acquisition step of step S1-2 can be made via the input unit 2a after the estimation step of step S3. Furthermore, in the electric wire sag control device 10 according to the second embodiment, when the sag d exceeds the design sag d0 and approaches the design sag d0, the reward r t+1 is increased to a larger positive value, and the reward r is calculated when the slack d is smaller than the design slack d0. t+1 may be set to decrease as a constant negative value or zero. [Industrial Applicability]

[0066] INDUSTRIAL APPLICABILITY The present invention can be used as an electric wire sag control device and an electric wire sag control method for controlling the sag of an electric wire supported by a plurality of transmission towers. [Explanation of symbols]

[0067] 1...Electric wire sag control device 2...Acquisition unit 2a...Input unit 2b...Imaging means 2c...Thermometer 3...Learning unit 4...Estimation unit 5...Execution unit 5a...Signal generation unit 5b...Drive unit 5c...Transmission unit 6...Association unit 7...Construction unit 8...Memory unit 9...Electric wire sag control method 10...Electric wire sag control device 11...Learning unit 12...Reward unit 13...Function update unit 14...Memory unit 15...Electric wire sag control method 50A to 50C...Transmission tower 50a to 50c...Support point 51...Electric wire 52...Holding means 53...Electric wire drum 54...Wire pulling machine S1...Acquisition process S1-1...New sag-related data acquisition process S1-2...Sag-related data acquisition process S2, S5...Learning process S2-1...Association process S2-2...Construction process S2-3...Memory process S3...Estimation process S4...Execution process S5-1...Reward process S5-2...Function update process

Claims

1. A wire slack control device that controls the amount of wires held by a holding means so that the slack of the wires supported by a plurality of transmission towers is within a specified range, an acquisition unit that acquires slackness-related data related to the slackness and feature data consisting of the inflow and outflow amounts corresponding to the slackness-related data; a learning unit that constructs a learning model for estimating new inflows and outflows corresponding to new slack-related data from the feature data, An electric wire sag control device characterized in that the sag-related data includes at least one of the combination of the load per unit length of the electric wire and the maximum tension of the electric wire and the design sag representing the specified range at a specified outside air temperature, at least one of the span lengths of the multiple transmission towers, the vertical height of each support point of the multiple transmission towers and the difference in elevation between each support point of the multiple transmission towers, and the actual length of the electric wire.

2. The learning unit a reward unit that determines a reward for a result of selecting the amount of input and output; a function update unit that updates a function representing a value selected from the amount of input and output based on the reward; The wire slack control device described in claim 1, characterized in that the learning model that estimates the new input and output amounts is constructed so that the reward calculated by the reward unit is maximized by repeatedly updating the function in the function update unit.

3. the sag-related data includes at least one of a combination of the load and the maximum tension and the design sag at a predetermined outside air temperature, the span length, at least one of the vertical height and the elevation difference, and the actual length; 3. The wire sag control device according to claim 2, wherein the reward is set to increase as the sag corresponding to the selected input / output amount approaches the design sag.

4. an estimation unit that estimates the new inflow / outflow amounts corresponding to the new slack-related data based on the learning model constructed by the learning unit; an execution unit that executes driving of the holding means based on the new inflow / outflow amount estimated by the estimation unit, The execution unit: a signal generating unit that generates a drive signal that reflects the new inflow / outflow amount estimated by the estimating unit; a drive unit that drives the holding means; 4. The wire slack control device according to claim 1, further comprising a transmitter that transmits the drive signal to the driver.

5. A wire sag control method for controlling an amount of inflow and outflow of a wire held by a holding means so that the sag of the wire supported by a plurality of transmission towers is within a specified range, an acquisition step of acquiring slack-related data relating to the slack and characteristic data including the amount of input and output corresponding to the slack-related data; a learning step of constructing a learning model for estimating new inflow / outflow amounts corresponding to new slack-related data from the feature data; A method for controlling sag of an electric wire, characterized in that the sag-related data includes at least one of a combination of the load per unit length of the electric wire and the maximum tension of the electric wire and a design sag representing the specified range at a predetermined outside air temperature, at least one of the span lengths of the plurality of transmission towers, the vertical height of each support point of the plurality of transmission towers and the difference in elevation between each support point of the plurality of transmission towers, and the actual length of the electric wire.

Images