Rubble mound construction method

Abstract

To calculate an appropriate height for dropping a weight during leveling construction of a rubble mound.SOLUTION: Measured value acquisition means 111 acquires a measured value of a height of a top surface 30 of a rubble mound 3 before tamping, and a measured value of the displacement amount of the height of the rubble mound 3 after tamping. Teacher data generating means 112 generates first teacher data 121 by specifying or calculating an explanatory variable and a target variable from the conditions of the executed tamping and the acquired measurement values. Learning model construction means 113 uses the generated first teacher data 121 to calculate a group of parameters for deriving the objective variable from the explanatory variables, and constructs a first learning model 122 using these. Fall height estimating means 114 estimates, based on the constructed first learning model 122, the height at which the weight 212 should be lifted and dropped in the next instructed tamping. Display control means 115 instructs a display unit 15 to display an estimated value of a fall height.SELECTED DRAWING: Figure 11

Description

[Technical Field]

[0001] The present invention relates to a method for constructing a rubble mound. [Background technology]

[0002] In order to install revetment caissons (concrete blocks) horizontally, the rubble mound that serves as the base must be leveled approximately evenly to the specified height (design height) before installation. Leveling methods include roller compaction, weight-bearing, hydraulic hammer, and vibro-hammer, but currently the most commonly used compaction methods are weight-bearing, hydraulic hammer, and vibro-hammer.

[0003] Compaction by weight mass is a method in which a weight mass is dropped freely from a predetermined height and the resulting impact energy is used to compact the rubble in a rubble mound.

[0004] Patent Document 1 describes that in order to bring the top surface of the rubble mound closer to the design height, the height position of a weight after it has fallen onto the rubble mound is measured to obtain the height of the top surface of the rubble mound. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 10-152839 Summary of the Invention [Problem to be solved by the invention]

[0006] For example, to increase the displacement (subsidence) of the top surface of a rubble mound, the weight can be dropped from a higher position. However, dropping the weight from a higher position than necessary will result in the top surface being displaced below the design height. Once the top surface is displaced below the design height, it is difficult to restore it to its original height, resulting in significant rework. Furthermore, dropping the weight from a relatively low position will not displace the rubble mound, and adjusting the top surface to the design height will require many drops, reducing construction efficiency. Efficient construction requires as few drops as possible to compact the rubble mound firmly and achieve a smooth surface at the design height. However, even if the current height of the rubble mound's top surface can be measured and the difference from the design height can be determined, determining the appropriate height for dropping the weight to achieve the required displacement to bring the current height closer to the design height must rely on the experience of the worker. Furthermore, when a certain section on the top surface of a rubble mound is struck, the amount of height displacement caused by the strike may differ depending on whether the surrounding sections have already been struck.

[0007] In view of the above background, the present invention provides a means for calculating an appropriate height for dropping a weight during rubble mound leveling work without relying on the empirical rules of the worker. [Means for solving the problem]

[0008] The construction method according to claim 1 of the present invention includes the steps of: measuring the height of the top surface of a rubble mound before striking with a weight for each of a plurality of sections obtained by dividing the top surface of the rubble mound; dropping the weight from a set height onto each of the plurality of sections to strike the top surface of the rubble mound a plurality of times and measuring the amount of displacement of the height of the rubble mound due to the striking; calculating the ratio of the amount of displacement to the square root of the product of the height from which the weight was dropped and the number of times it was dropped for each of the plurality of sections and the entire area of the top surface; This is a rubble mound construction method comprising the steps of: constructing a first learning model using training data in which the height, the number of times, a previous displacement amount indicating the displacement amount of the rubble mound due to the previous strike, a previous cumulative displacement amount indicating the cumulative displacement amount of the rubble mound due to strikes up to the previous time, and the ratio are used as explanatory variables and the displacement amount is used as the objective variable; using the first learning model to estimate the height at which the weight bob will be dropped onto the rubble mound; and dropping the weight bob from the height estimated in the estimating step to strike the top surface of the rubble mound with the weight bob.

[0009] A construction method according to claim 2 of the present invention is a construction method for a rubble mound according to claim 1, wherein in the step of calculating the ratio, the ratio of the entire area is any one of the arithmetic mean, mode, median, geometric mean or maximum value of all the ratios of the multiple plots.

[0010] A construction method according to claim 3 of the present invention is a construction method for a rubble mound according to claim 1 or 2, wherein the step of calculating the ratio is a step of calculating the ratio for each of the plurality of plots and the entire area, as well as for each group of two or more plots among the plurality of plots onto which the weight is moved and dropped, and the step of constructing is a step of constructing the first learning model using training data in which the height at which the weight is dropped, the number of times, the previous displacement amount, the accumulated displacement amount up to the previous time, and the ratio are used as explanatory variables for the plots, the groups, and the entire area, and the displacement amount is used as a target variable.

[0011] A construction method according to claim 4 of the present invention is a construction method for a rubble mound according to claim 3, wherein in the step of calculating the ratio, the ratio of the group is any one of the arithmetic mean, mode, median, geometric mean or maximum value of the ratios of all plots belonging to the group.

[0012] A construction method according to claim 5 of the present invention is a construction method for a rubble mound according to any one of claims 1 to 4, comprising the steps of: constructing a second learning model using training data in which the height at which the weight was dropped, the number of times, the previous displacement amount, the cumulative displacement amount up to the previous time, and the ratio for the multiple compartments and the entire area are used as explanatory variables, and the displacement amount of a compartment other than the compartment on which the weight was dropped is used as a target variable; calculating the displacement amount of another compartment caused by dropping the weight on a certain compartment among the multiple compartments using the second learning model; and notifying the calculated displacement amount of the other compartment.

[0013] A construction method according to claim 6 of the present invention is a construction method for a rubble mound according to any one of claims 1 to 5, comprising the steps of: using the first learning model to predict whether a state will occur in which the displacement amount caused by dropping the weight will be less than a threshold value in any of the plurality of sections; notifying that a state in which the displacement amount will be less than the threshold value has been predicted; and comparing the predicted state in which the displacement amount will be less than the threshold value and the top surface of the rubble mound with a predetermined height.

[0014] A construction method according to claim 7 of the present invention is a construction method for a rubble mound as described in claim 5, comprising the steps of: using the second learning model to predict whether a state will occur in any of the plurality of sections in which the amount of displacement of the other section will be less than a threshold value due to the dropping of the weight; notifying that a state in which the amount of displacement of the other section will be less than the threshold value has been predicted; and comparing the predicted state in which the amount of displacement of the other section will be less than the threshold value with a predetermined height.

[0015] A construction method according to claim 8 of the present invention is a construction method for a rubble mound described in any one of claims 1 to 7, in which the step of measuring the amount of displacement in height of the rubble mound due to the striking uses the depth measured by a measuring instrument at at least three points on the upper surface of the weight. [Effects of the Invention]

[0016] According to the present invention, it is possible to calculate the appropriate height from which to drop a weight when leveling a rubble mound. [Brief explanation of the drawings]

[0017] [Figure 1] A diagram showing the top surface 30 of the rubble mound 3 being leveled using construction method M. [Figure 2] A top view of the leveling machine 2. [Figure 3] FIG. 2 is a diagram showing an example of the configuration of a leveling machine 2. [Figure 4] 2 is a diagram showing an example of the configuration of a striking machine 21. FIG. [Figure 5] FIG. 2 is a diagram showing an example of leveling height sensors 23 arranged in the width direction. [Figure 6] FIG. 10 is a diagram showing an example of a leveling frame R. [Figure 7] FIG. 1 is a diagram showing an example of the configuration of a learning device 1. [Figure 8] FIG. 10 is a diagram showing an example of first teacher data 121. [Figure 9] FIG. 2 is a diagram showing an example of the configuration of a control device 4. [Figure 10] FIG. 2 is a diagram showing an example of the configuration of a management device 5. [Figure 11] FIG. 2 is a diagram showing an example of the functional configuration of the learning device 1. [Figure 12] FIG. 10 is a flow chart showing an example of the processing flow of a construction method M. [Figure 13] FIG. 10 is a diagram showing the state before the leveling machine 2 levels the top surface 30. [Figure 14] FIG. 2 is a diagram showing how the leveling machine 2 performs rake leveling. [Figure 15] FIG. 10 is a flowchart showing an example of a process flow for leveling a weight. [Figure 16] FIG. 10 is a diagram showing the state before the weight 212 strikes the top surface 30. [Figure 17] FIG. 10 is a diagram showing the state after the weight 212 strikes the top surface 30. [Figure 18] FIG. 10 is a diagram showing an example of the configuration of a learning device 1 according to a modified example. [Figure 19] FIG. 10 is a diagram showing an example of second teacher data 123 in a modified example. [Figure 20] FIG. 10 is a diagram showing an example of the functional configuration of a learning device 1 according to a modified example. [Figure 21] FIG. 10 is a diagram showing an example of the functional configuration of a learning device 1 according to a modified example. DETAILED DESCRIPTION OF THE INVENTION

[0018] <Embodiment> <Overall structure> Below, we will explain a construction method M for the top surface of a rubble mound according to one embodiment of the present invention. Figure 1 is a diagram showing how the top surface 30 of a rubble mound 3 is leveled using construction method M. In the following explanation, directions in the rubble mound 3 will be described using the X-axis, Y-axis, and Z-axis indicated by three arrows in the figure. Here, -Z is the direction of gravity, that is, the downward direction. Also, +X is, for example, the north direction, and +Y is, for example, the west direction.

[0019] Construction method M is carried out using a learning device 1, a control device 4, and a management device 5 mounted on a ship stationary on the sea surface W, and a leveling machine 2 placed on top of a rubble mound 3.

[0020] Rubble mound 3 is constructed by dumping rubble on seabed B. Rubble mound 3 constructed on seabed B will serve as the foundation for port structures such as caissons that will be constructed on top of it.

[0021] The learning device 1, the control device 4, and the management device 5 are connected to each other via a wired or wireless communication line so as to be able to communicate with each other.

[0022] The control device 4 is a device that controls the leveling machine 2, and is, for example, a programmable logic controller. The control device 4 may include, for example, an FPGA (Field Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit). In this embodiment, the control device 4 accepts operations from an operator.

[0023] The management device 5 is a device that acquires and stores data for managing the leveling machine 2 from the leveling machine 2 and the control device 4. The management device 5 is, for example, a computer.

[0024] The learning device 1 is a device that performs machine learning by acquiring at least a portion of the above-mentioned data from the management device 5. The learning device 1 is, for example, a computer.

[0025] The leveling machine 2 is lowered from the sea to the seabed by a crane ship or the like, and placed in a planned location on the top surface 30 of the rubble mound 3. Figure 2 is a diagram showing the leveling machine 2 as viewed from above. Figure 3 is a diagram showing an example of the configuration of the leveling machine 2.

[0026] <Configuration of the leveling machine> As shown in FIG. 3 , the leveling machine 2 includes a stand 20 , a striking machine 21 , a rake device 22 , a leveling height sensor 23 , a communication line 24 , a positioning device 25 , a control line 26 , and an orientation inclinometer 27 .

[0027] The platform 20 is a structure that supports each component of the leveling machine 2. The platform 20 has three or more legs that contact the top surface 30 and connecting parts that interconnect these legs. The leveling machine 2 shown in FIG. 3 has eight legs (not shown). Of these eight legs, four are legs that support the leveling machine 2, and the remaining four are legs for movement that contact the top surface 30 when the leveling machine 2 is moved. The legs and connecting parts are made of, for example, steel. The striking machine 21, rake device 22, and leveling height sensor 23 are supported on a frame that extends horizontally so that they can move along the X and Y axes.

[0028] Fig. 4 is a diagram showing an example of the configuration of the striking machine 21. The striking machine 21 shown in Fig. 4 has a hoisting wire 210, a weight height sensor 211, a weight 212, a weight fitting table 213, a table height sensor 214, a weight lifting winch 215, and a weight traveling carriage 216.

[0029] The weight-traveling traversing carriage 216 is a carriage that travels in a horizontal direction or the like along the frame of the platform 20. The weight-traveling traversing carriage 216 is equipped with a weight-lifting winch 215, and moves it to above the area where the weight 212 is expected to fall.

[0030] The weight lifting winch 215 is a winch mounted on the weight traveling traverse carriage 216 and winds up and unwinds the hoisting wire 210 .

[0031] The suspending wire 210 is a wire that suspends and holds (also referred to as "suspending") the weight fitting base 213.

[0032] The weight fitting base 213 is a base whose upper surface is suspended by a suspending wire 210. The weight fitting base 213 has a fitting member such as a latch that fits (also referred to as "fitting") the weight 212 on its lower surface, for example.

[0033] The weight 212 is a rectangular parallelepiped object made of metal such as steel. The upper surface of this weight 212 is detachable from the weight fitting base 213 by the fitting member described above. When the weight 212 is detached from the weight fitting base 213 and dropped onto the top surface 30 of the rubble mound 3, it strikes and compacts the top surface 30 of the rubble mound 3, making the height of the top surface 30 uniform.

[0034] The weight 212 is suspended together with the weight fitting base 213 by a hoisting wire 210 from a weight lifting winch 215, which travels on the frame via a weight traveling carriage 216 and is located at a predetermined position. The height at which the weight 212 falls is adjusted by the amount of the hoisting wire 210 wound by the weight lifting winch 215. When the weight lifting winch 215 adjusts the amount of the hoisting wire 210 wound, the weight fitting base 213 is suspended at a specified height. In order to maintain the length of the hoisting wire 210, the weight lifting winch 215 holds the wound hoisting wire 210 with a force that resists gravity acting on the weight 212 and the weight fitting base 213.

[0035] When the weight fitting base 213 disengages the fitting member at this adjusted height and releases the force holding the weight 212, the weight 212 detaches from the weight fitting base 213 and falls freely onto the top surface 30, which is the upper surface of the rubble mound 3.

[0036] That is, the weight lifting winch 215 winds up or unwinds the hoisting wire 210 by a predetermined length, thereby suspending the weight 212 at a specified height. Then, when the weight fitting base 213 subsequently releases the weight 212, the weight 212 falls toward the top surface 30 and strikes the position where it fell.

[0037] After striking, the weight lifting winch 215 pays out the hoisting wire 210 to lower the weight fitting base 213 to the height of the weight 212. The lowered weight fitting base 213 is again fitted with the weight 212 by the fitting member. Then, when the weight lifting winch 215 winds up the hoisting wire 210, the weight fitting base 213 and the weight 212 rise from the top surface 30 and are lifted up to a height according to the amount wound up. In other words, the weight 212 that struck the top surface 30 is again fitted with the weight fitting base 213 connected to the end of the hoisting wire 210 suspended from the weight lifting winch 215, and is thereby suspended.

[0038] The table height sensor 214 is a sensor that is installed on the underside of the weight-traveling traversing carriage 216 and measures the distance to the weight-fitting table 213. The table height sensor 214 is attached to the underside of the weight-traveling traversing carriage 216. The table height sensor 214 measures the distance from the underside of the weight-traveling traversing carriage 216 to the upper surface of the weight-fitting table 213, which corresponds to the amount of payout of the hoisting wire 210.

[0039] Therefore, the distance measurement result by the table height sensor 214 may be used to verify the amount of payout of the hoisting wire 210 measured by a rotary encoder or the like provided on the weight lifting winch 215. The amount of payout of the hoisting wire 210 or the distance measurement result by the table height sensor 214 is used to calculate the depth of the upper surface of the weight fitting table 213 based on the position of the lower surface of the weight traveling traversing carriage 216.

[0040] The weight height sensor 211 is a measuring instrument that measures the depth on the upper surface of the weight 212, and is an ultrasonic distance meter. The weight height sensor 211 is attached to the lower surface of the weight fitting base 213. The weight height sensor 211 has a wave transmitting unit that transmits ultrasonic waves toward the upper surface of the weight 212 located below. The weight height sensor 211 also has a wave receiving unit that receives ultrasonic waves reflected by the upper surface of the weight 212. The weight height sensor 211 measures the distance from the lower surface of the weight fitting base 213 to the upper surface of the weight 212 based on the time it takes for the ultrasonic waves transmitted from the wave transmitting unit to be reflected and return to the wave receiving unit.

[0041] For example, the depth of the bottom surface of the weight fitting base 213 can be found by offsetting the thickness of the weight fitting base 213 from the depth of the top surface of the weight fitting base 213 calculated from the payout amount of the suspending wire 210. The top surface of the weight 212 is located at a position further down from this depth by a distance measured by the weight height sensor 211. By offsetting the thickness of the weight 212 from the position of the top surface of this weight 212, the depth of the bottom surface of the weight 212, i.e., the top surface 30, can be measured.

[0042] The weight height sensors 211 are preferably attached at least three locations on the lower surface of the weight fitting base 213 at equal intervals in the horizontal direction, and it is more preferable that the intervals be as wide as possible, because this allows the upper surface of the weight 212 to be perceived as a plane and its inclination to be identified.

[0043] The leveler 2 may have a guide that contacts the side of the weight 212 as it falls to guide the falling position. The guide is, for example, a rail-shaped steel material or the like that is set up perpendicular to the top surface 30.

[0044] The rake device 22 shown in Figure 3 is a member that can move in the Y-axis and Z-axis directions, and its tip comes into contact with the top surface 30 and is moved in the Y-axis direction to scrape off and level rubble and the like on the top surface of the top surface 30. It is preferable that the rake device 22 be moved in a direction away from the center line C shown in Figure 2 to level it.

[0045] For example, as shown in Figure 2, when the leveling frame R is on the +Y side of the center line C, the rake device 22 scrapes off the riprap from the upper surface of the top surface 30 while moving in the +Y direction. As a result, the scraped off riprap is dropped onto the slope on the +Y side of the top surface 30. In this way, the rake device 22 scrapes off the riprap in a direction away from the center line C and drops it down the slope, so the scraped off riprap does not gather in the center of the top surface 30.

[0046] The rake device 22 is equipped with leveling height sensors 23 on the rear side in the direction of travel of the rake device 22. These leveling height sensors 23 measure the shape of the top surface 30 of the rubble mound 3 by moving the rake device 22 back and forth. In this embodiment, the leveling height sensors 23 move in conjunction with the movement of the rake device 22.

[0047] Measurements using the leveling height sensor 23 are carried out three times: before measurement, during measurement, and after measurement. The before measurement measures the shape of the top surface 30 of the rubble mound 3 into which rubble has been dumped before it is leveled by the rake device 22. The intermediate measurement is performed by measuring the shape of the mound after it has been leveled by the rake device 22. The post-measurement is performed by measuring the shape of the top surface 30 after it has been leveled by the rake device 22 and then compacted by the weight 212. The leveling height sensor 23 is, for example, an ultrasonic sensor.

[0048] A plurality of leveling height sensors 23 may be arranged in the width direction of the rake device 22. Fig. 5 is a diagram showing an example of leveling height sensors 23 arranged in the width direction. Fig. 5 shows a schematic view of the rake device 22 and the leveling height sensor 23 viewed from above. On the -Y side of the rake device 22 shown in Fig. 5, five leveling height sensors 23-1, 23-2, 23-3, 23-4, and 23-5 (hereinafter, when there is no need to distinguish between them, they will simply be referred to as "leveling height sensors 23") are arranged from the +X side to the -X side.

[0049] The leveling height sensors 23-1, 23-2, 23-3, 23-4, and 23-5 measure the height of the top surface 30 of the portion passed by the rake device 22, as well as unevenness in the width direction. In this explanation, five leveling height sensors are installed, but the number does not have to be five, and six or more sensors are also acceptable as long as they can measure the entire lateral width range of the leveler.

[0050] The positioning device 25 measures the position of the leveling machine 2. The positioning device 25 has, for example, a GNSS (Global Navigation Satellite System) receiver. Furthermore, the azimuth inclinometer 27 has a gyro sensor.

[0051] The GNSS receiver of the positioning device 25 receives radio waves from multiple satellites to measure the three-dimensional position of the leveling machine 2 on Earth. However, since radio waves from satellites may not reach underwater, the positioning device 25 is installed above the sea surface W. This construction method M may also involve installing a GNSS receiver and transceiver on a ship stopped above the sea surface W, and installing a transponder at a known position on the mounting base 20.

[0052] The GNSS receiver measures the coordinates of a positioning point P shown in FIG. 2, and the position of the leveling machine 2 is identified based on this positioning point P.

[0053] The gyro sensor of the azimuth inclinometer 27 measures the attitude of the leveling machine 2, that is, the tilt of the leveling machine 2 with respect to the X-axis direction, Y-axis direction, and Z-axis direction. This allows the global coordinates indicating the directions of north, south, east, and west on the Earth to correspond to the local coordinates indicating the relative positional relationship between the components of the leveling machine 2.

[0054] Then, by determining the installation position of the leveling machine 2 and associating the global coordinates with the local coordinates, the range of the top surface 30 that the weight 212 can drop and strike is determined as the leveling frame R shown in FIG.

[0055] The communication line 24 connects the positioning device 25 arranged above the sea level W with each component arranged below the sea level W so that communication is possible.

[0056] The control line 26 is a communication wire, and is a composite cable that transmits and receives control instructions and measurement data for the leveling machine 2, as well as transmits power to the leveling machine 2, and connects the leveling machine 2 to the control device 4 and management device 5 so that they can communicate with each other. The control line 26 may be wireless for transmitting and receiving control instructions and measurement data. When communicating wirelessly, radio waves may not reach underwater, so the leveling machine 2 only needs to be equipped with a communication device that communicates wirelessly with the control device 4 and management device 5, installed above the sea surface W. The control line 26 allows the leveling machine 2 to communicate with the control device 4 and management device 5, and place it under the control of the control device 4.

[0057] The control line 26 transmits control signals and the like from the control device 4 to a drive device (not shown) of the leveling machine 2. As a result, the striking machine 21, rake device 22, weight-traveling traverse carriage 216, and weight-fitting table 213, which are driven by this drive device, are controlled by the control device 4, and the leveling height sensor 23, table height sensor 214, and weight height sensor 211 are controlled by the management device 5. The various sensors may be controlled by the control device 4 rather than the management device 5 depending on the communication format used.

[0058] Fig. 6 is a diagram showing an example of the leveling frame R. The leveling frame R shown in Fig. 6 is configured by 8 sections R1 to R8 arranged in two rows and four columns. The surface (striking surface) that is the top surface 30 and that is struck in one go by the bottom surface of the weight 212 is, for example, a square measuring 2 meters on each side. In this case, because these striking surfaces are arranged in two rows and four columns, the set area for the leveling frame R is a rectangle measuring 4 meters in the X-axis direction and 8 meters in the Y-axis direction.

[0059] Using the control device 4, the operator first divides the entire area of the top surface 30 of the rubble mound 3 into multiple leveling frames R in order to level the entire area. Then, the operator assigns orders to the divided multiple leveling frames R and specifies them in order to the control device 4. The control device 4 then instructs the leveling machine 2 to move to a position for leveling the specified leveling frame R. Furthermore, the control device 4 instructs the leveling machine 2 to measure the height of the top surface 30 and to continue rake leveling until the measured height meets the set height conditions suitable for striking by the striking machine 21.

[0060] When the rake device 22 moves in the leveling machine 2 that has moved to the above position and the top surface 30 is leveled, the leveling height sensor 23 that moves with the rake device 22 measures the height of the top surface 30 after leveling and sends it to the control device 4 via the management device 5. When the measured height of the top surface 30 meets the condition of the set height for starting striking, the operator assigns orders to the multiple sections (R1 to R8) that the specified leveling frame R has been further divided into, and specifies these sections to the control device 4 in order.

[0061] The operator then instructs the control device 4 to measure the height of the top surface 30 of the rubble mound 3 before striking a certain section specified by the control device 4, to drop a weight 212 onto the section to strike the top surface 30, and to measure the amount of displacement in height of the section after striking it.

[0062] Before the first learning model 122 is constructed and the drop height cannot be estimated, the operator of the control device 4 instructs the leveling machine 2 via the operation unit to strike at a preset drop height. Also, after the first learning model 122 is constructed and the drop height is estimated, the operator of the control device 4 confirms or refers to this drop height displayed on the display unit 15 and inputs the drop height into the operation unit, thereby instructing the leveling machine 2 to strike at the input drop height.

[0063] The leveling machine 2, which has received this instruction via the control line 26, first operates the weight lifting winch 215 to wind up the hoisting wire 210 and hoist the weight fitting base 213 and weight 212 to the instructed drop height. Next, the leveling machine 2 releases the fitting member of the weight fitting base 213 to drop the weight 212. Thereafter, the leveling machine 2 measures the amount of displacement in height of the riprap mound 3 caused by the striking using the weight height sensor 211.

[0064] The management device 5 acquires the measured displacement amount via the control line 26. The management device 5 acquires and stores the measured displacement amount and each condition at the time of striking. When performing machine learning, the learning device 1 acquires data such as the above-mentioned displacement amount from the control device 4.

[0065] <Configuration of learning device> Fig. 7 is a diagram showing an example of the configuration of learning device 1. Learning device 1 has processor 11, memory 12, and interface 13. Learning device 1 may also have operation unit 14 and display unit 15 as shown in Fig. 7. These components are connected to each other so that they can communicate with each other, for example, by a bus.

[0066] Processor 11 reads and executes a computer program (hereinafter simply referred to as a program) stored in memory 12 to control each part of learning device 1. Processor 11 is, for example, a CPU (Central Processing Unit).

[0067] The memory 12 is a storage means for storing an operating system, various programs, data, etc., which are loaded into the processor 11. The memory 12 includes a RAM (Random Access Memory) and a ROM (Read Only Memory). The memory 12 may also include a solid state drive, a hard disk drive, etc.

[0068] The memory 12 also stores first teacher data 121 and a first learning model 122. Fig. 8 is a diagram showing an example of the first teacher data 121. The first teacher data 121 stores various measurement values measured each time the weight 212 strikes the top surface 30, and various conditions, in association with each other. The first teacher data 121 shown in Fig. 8 assigns a data ID, which is identification information for identifying data, to each strike, and stores, as explanatory variables, the number of falls, the fall height, a previous section constant which is the ratio of the amount of displacement to the square root of the product of the height at which the weight was dropped and the number of times it was dropped, a previous frame constant, a previous overall constant, a previous displacement amount which indicates the amount of displacement in the height of the rubble mound due to the previous strike, and a previous cumulative displacement amount which indicates the cumulative amount of displacement in the height of the rubble mound due to strikes up to the previous time.

[0069] Here, the number of drops is the number of times that the weight 212 has been dropped onto a section of the top surface 30 that is the target of the strike, including the current strike. The drop height is the height from which the weight was dropped in the current strike.

[0070] The previous section constant, previous frame constant, and previous overall constant are all the striking constant α. The striking constant α will now be explained. It is known that weight compaction has the relationship shown in the following equation (1).

[0071]

number

[0072] Therefore, according to this formula (1), the weight penetration amount is proportional to the weight mass, impact speed, and square root of the number of drops. Also, according to this formula (1), the weight penetration amount is inversely proportional to the weight base area.

[0073] Incidentally, the impact velocity v0 of the weight 212 when it falls the distance indicated by H and impacts the top surface 30 is expressed by the following equation (2) using the gravitational acceleration g.

[0074]

number

[0075]

number

[0076] The difference between the weight penetration amount in the Nth strike and the weight penetration amount in the previous strike (that is, the (N-1)th strike) is expressed by the following formula (4).

[0077]

number

[0078] The striking constant α is updated each time the weight is struck, and is calculated based on measurement values that differ for each section. The striking constant α calculated in association with a leveling frame made up of multiple sections is calculated as a representative value of the striking constant α calculated for each section. Furthermore, the striking constant α calculated in association with the entire area of the top surface 30 of the rubble mound 3 is calculated as a representative value of the striking constant α calculated for each of all sections included in this entire area.

[0079] That is, the previous section constant is the strike constant α calculated when the previous strike to a certain section ended.

[0080] The previous frame constant is a representative value of the strike constant α calculated for each of all the sections that make up the equalizing frame R when the previous strike was completed. This representative value is, for example, the arithmetic mean of the strike constant α calculated for each of the eight sections R1 to R8 that make up the equalizing frame R. The equalizing frame R is a group consisting of two or more sections. In other words, this previous frame constant is an example of the arithmetic mean of the ratios of all the sections that belong to the group.

[0081] The previous overall constant is a representative value of the striking constant α calculated for each of all the sections that make up the entire area of the top surface 30 at the time the previous striking was completed. This representative value is, for example, the arithmetic mean of the striking constant α calculated for each of all the sections that make up the entire area of the top surface 30. In this case, the previous overall constant is an example of the arithmetic mean of all the ratios of the multiple sections obtained by dividing the top surface of the rubble mound. Note that this representative value is not limited to the arithmetic mean of the striking constant α for each section, but may also be any of various statistical quantities such as the mode, median, geometric mean, or maximum value.

[0082] The first teacher data 121 stores, as a response variable, the displacement of the top surface 30 due to the strike, i.e., the displacement due to the current strike, in association with a data ID indicating the strike. The first teacher data 121 also stores a previous displacement, which is the previous displacement identified based on the displacement identified for each strike. The first teacher data 121 also stores a previous cumulative displacement, which is the cumulative amount of displacement up to the previous time.

[0083] The first learning model 122 shown in Fig. 7 is generated by machine learning based on the first teacher data 121. This first learning model includes a group of parameters used in a formula, matrix, etc. for predicting a dependent variable from explanatory variables stored in the first teacher data 121. For example, linear regression, decision tree, xgboost, etc. can be applied to the machine learning for generating the first learning model.

[0084] The interface 13 is a communication circuit that connects the learning device 1 to the management device 5 and the control device 4 so that they can communicate with each other via a wired or wireless communication line.

[0085] The operation unit 14 is equipped with operation buttons, a keyboard, a touch panel, a mouse, and other operators for issuing various instructions, and receives operations and sends signals according to the operation content to the processor 11. These operations include, for example, pressing keys on the keyboard or gestures on the touch panel.

[0086] Display unit 15 has a display screen such as a liquid crystal display, and displays images under the control of processor 11. A transparent touch panel of operation unit 14 may be placed on top of the display screen. Note that learning device 1 does not necessarily have operation unit 14 and display unit 15. Learning device 1 may be operated from an external device via interface 13, or may present information to an external device.

[0087] <Control device configuration> Fig. 9 is a diagram showing an example of the configuration of the control device 4. The control device 4 shown in Fig. 9 includes a processor 41, a memory 42, an interface 43, an operation unit 44, and a display unit 45. These components are connected to each other so that they can communicate with each other, for example, by a bus.

[0088] The processor 41 reads and executes a program stored in the memory 42 to control each part of the control device 4. The processor 41 is, for example, a CPU.

[0089] The interface 43 is a communication circuit that connects the control device 4 to other devices via wire or wirelessly so that they can communicate with each other.

[0090] The operation unit 44 is equipped with operation buttons, a lever box, a touch panel, and other operators for issuing various instructions, and receives operations and sends signals to the processor 41 according to the operation content.

[0091] The display unit 45 has a display screen such as a liquid crystal display, and displays images under the control of the processor 41. A transparent touch panel of the operation unit 44 may be placed on top of the display screen. Note that the control device 4 does not necessarily have the operation unit 44 and the display unit 45. The control device 4 may be operated by an external device via the interface 43, or may present information to an external device.

[0092] The memory 42 is a storage means for storing an operating system, various programs, data, etc., which are loaded into the processor 41. The memory 42 includes RAM and ROM. The memory 42 may also include a solid state drive, a hard disk drive, etc.

[0093] <Configuration of management device> Fig. 10 is a diagram showing an example of the configuration of the management device 5. The management device 5 shown in Fig. 10 includes a processor 51, a memory 52, an interface 53, an operation unit 54, and a display unit 55. These components are connected to each other via, for example, a bus so that they can communicate with each other.

[0094] The processor 51 reads and executes programs stored in the memory 52 to control each part of the management device 5. The processor 51 is, for example, a CPU.

[0095] The interface 53 is a communication circuit that connects the management device 5 to other devices via wire or wirelessly so that they can communicate with each other.

[0096] The operation unit 54 is provided with operation buttons, a keyboard, a touch panel, a mouse, and other operators for issuing various instructions, and receives operations and sends signals according to the operation content to the processor 51.

[0097] The display unit 55 has a display screen such as a liquid crystal display, and displays images under the control of the processor 51. A transparent touch panel of the operation unit 54 may be placed on top of the display screen. Note that the management device 5 does not necessarily have the operation unit 54 and the display unit 55. The management device 5 may be operated from an external device via the interface 53, or may present information to an external device.

[0098] The memory 52 is a storage means for storing an operating system, various programs, data, etc., which are loaded into the processor 51. The memory 52 includes RAM and ROM. The memory 52 may also include a solid state drive, a hard disk drive, etc.

[0099] <Functional configuration of the learning device> 11 is a diagram showing an example of the functional configuration of the learning device 1. When the processor 11 executes a program, the learning device 1 functions as a measurement value acquisition means 111, a teacher data generation means 112, a learning model construction means 113, a drop height estimation means 114, and a display control means 115.

[0100] The measurement value acquisition means 111 acquires the measurement value of the height of the top surface 30 of the rubble mound 3 before striking and the measurement value of the displacement of the height of the rubble mound 3 after striking from the management device 5 via the interface 13.

[0101] The teacher data generating means 112 generates first teacher data 121 by identifying or calculating explanatory variables and objective variables from the conditions of the executed strike and the acquired measurement values.

[0102] The learning model construction means 113 uses the generated first teacher data 121 to calculate a group of parameters for deriving a target variable from the explanatory variables, and constructs a first learning model 122 using these.

[0103] Based on the constructed first learning model 122, the drop height estimation means 114 estimates the height (drop height) from which the weight 212 should be lifted and dropped during the next instructed strike. First, the drop height estimation means 114 specifies a drop height, for example, 1 m, at which the weight 212 will not sink below the designed leveling height even if it falls and compacts the top surface 30, and estimates the displacement of a certain section when the weight 212 is dropped at that drop height. Then, the drop height estimation means 114 varies the drop height so that the estimated displacement approaches a predetermined desired displacement. In this way, the drop height estimation means 114 estimates the "drop height," which is the appropriate height from which the weight 212 should be dropped. The estimated drop height is transmitted to the display control means 115. The display control means 115 instructs the display unit 15 to display the estimated drop height.

[0104] <Construction method processing> FIG. 12 is a flow diagram showing an example of the processing flow of construction method M. The operator instructs the leveling machine 2 via the operation unit of the control device 4 to adjust the work start position (step S001). This determines the position of the leveling machine 2 and determines the leveling frame R. Next, the operator instructs the leveling machine 2 via the management device 5 to perform preliminary measurement with the leveling height sensor 23 and to perform rake leveling until the set height is reached via the operation unit of the control device 4 (step S002). This causes the leveling frame R to be roughly leveled.

[0105] The leveling machine 2 measures (intermediately measures) the height of the top surface 30 after rake leveling using the leveling height sensor 23 in accordance with instructions from the management device 5. Based on the measurement results, the management device 5 determines whether the top surface 30 meets the condition for the set height at the start of striking (step S003). If the management device 5 determines that the top surface 30 does not meet the condition for the set height at the start of striking (step S003; NO), the control device 4 instructs the leveling machine 2 to perform rake leveling again.

[0106] Figure 13 is a diagram showing the state before the leveling machine 2 levels the top surface 30. Under the control of the control device 4 and the management device 5, the leveling machine 2 scans the leveling height sensor 23 together with the rake device 22 to measure (pre-measure) the height of the top surface 30. At this time, the rake device 22 lifts the tip of the rake and moves it away from the top surface 30 to measure. The control device 4 pre-determines the upper limit of the height of the top surface 30 and the range of the difference in height between its highest and lowest points as conditions for the set height at which to start striking. The height measured by the leveling height sensor 23 is sent from the leveling machine 2 to the management device 5, which determines whether or not the set height conditions are met.

[0107] Figure 14 is a diagram showing how the leveling machine 2 performs rake leveling. Under the control of the control device 4, the leveling machine 2 drives the rake device 22 in the direction of the arrow shown in Figure 14 to perform "rake leveling," scraping off riprap and other debris from the top surface 30 and leveling it. The scraped riprap and other debris from the top surface 30 is moved to the outside of the top surface 30. After the rake device 22 has leveled the water and the turbidity has settled, the leveling height sensor 23 measures the height of the top surface 30 as an intermediate measurement, and transmits this to the control device 4 via the management device 5.

[0108] On the other hand, when it is determined that the top surface 30 meets the condition of the set striking start height (step S003; YES), the control device 4 instructs the leveling machine 2 to execute the weight leveling process (step S100).

[0109] When the weight leveling process is completed, the control device 4 determines whether leveling of the entire area of the top surface 30 has been completed (step S004). If it is determined that rake leveling and weight leveling of the entire area of the top surface 30 have not been completed (step S004; NO), the operator returns the process to step S001. That is, the operator specifies a leveling frame R on the top surface 30 that has not been leveled, and instructs the leveling machine 2 to perform a series of processes of position adjustment → rake leveling → weight leveling on the new leveling frame R. On the other hand, if it is determined that leveling of the entire area of the top surface 30 has been completed, the operator ends the process.

[0110] FIG. 15 is a flow diagram showing an example of the flow of the weight-dropping process (step S100). The control device 4 further divides the specified leveling frame R to determine sections R1 to R8. The operator then assigns an order to these sections R1 to R8 and determines which section will start weight-drop compaction in accordance with that order. The operator instructs the leveling machine 2 to measure (intermediately measure) the height of the determined section before striking it (step S101). The result of this measurement is transmitted from the leveling machine 2 to the control device 4 via the management device 5, and the data is accumulated in the management device 5. The learning device 1 acquires the recorded data from the management device 5, generates first training data, and constructs a first learning model.

[0111] This step S101 is performed sequentially for each of the plurality of sections obtained by dividing the top surface 30 of the rubble mound 3, for each leveling frame R. Therefore, this step S101 is an example of a step for measuring the height of the top surface of the rubble mound before striking with the weight, for each of the plurality of sections obtained by dividing the top surface of the rubble mound.

[0112] The learning device 1 determines whether or not a first learning model exists (has already been constructed) in the memory 12 (step S102). Note that, for a new construction project, re-learning may be performed using a learning model that has already been generated and used in another construction project. If it is determined that the first learning model does not exist in the memory 12 (step S102; NO), the operator specifies a predetermined drop height for the leveling machine 2 to the control device 4 (step S103).

[0113] On the other hand, if it is determined that the first learning model is present in the memory 12 (step S102; YES), the learning device 1 uses this first learning model to estimate the drop height of the weight 212, and displays the estimated drop height on the display unit 15 for confirmation by the operator (step S104). This step S104 is an example of a step in which the first learning model is used to estimate the height at which the weight will be dropped onto the rubble mound. After confirming the drop height display, the operator adjusts the drop height of the control device 4 via the operation unit if the drop height indication value from the first learning model needs to be adjusted, or if no adjustment is necessary, instructs the leveling machine 2 to strike from the estimated drop height.

[0114] The striking machine 21 of the leveling machine 2, under the control of the control device 4, hoists the weight 212 to the instructed drop height, and then releases the weight 212 to allow it to fall freely, striking the above-mentioned section (step S105). This step S105 is an example of a step in which the weight is dropped from the height estimated in the estimating step, and strikes the top surface of the rubble mound with the weight.

[0115] 16 is a diagram showing the state before the weight 212 strikes the top surface 30. The weight-traveling carriage 216 moves to above the specified section of the leveling frame R. When the movement is complete, the weight-lifting winch 215 pays out the hoisting wire 210 and lowers the weight-fitting table 213 to a height corresponding to the specified drop height.

[0116] Figure 17 is a diagram showing the state after the weight 212 has struck the top surface 30. The weight fitting base 213 has been lowered to a predetermined position, and the fitting member is removed to release the weight 212. This causes the weight 212 to fall freely and strike the top surface 30 of the rubble mound 3. As a result, the weight 212 penetrates the top surface 30, and its height is displaced.

[0117] This step S105 is performed for each of the multiple sections obtained by dividing the top surface 30 of the rubble mound 3 until the height falls within a predetermined range from the design compaction height. Therefore, this step S105 is an example of a step of dropping a weight from a set height a predetermined number of times onto each of the multiple sections obtained by dividing the top surface of the rubble mound.

[0118] When a strike is made and the weight 212 penetrates the top surface 30, the leveling machine 2, under the control of the management device 5, measures the amount of displacement in the height of the riprap mound 3 due to the strike (step S106). At this time, the leveling machine 2 measures the amount of displacement using the weight height sensor 211 attached to the weight 212 that has penetrated the top surface 30. This measurement is performed every time a strike is made. Therefore, step S106 is an example of a step for measuring the amount of displacement in the height of the riprap mound due to the strike.

[0119] In addition, when the leveling machine 2 measures the amount of height displacement using a weight height sensor 211, which is a transponder attached to at least three points on the top surface of the weight 212, this step S106 is an example of a step of measuring the amount of height displacement of the rubble mound due to striking, using the depth of at least three points on the top surface of the weight measured by a measuring instrument.

[0120] Next, the learning device 1 calculates the square root of the product of the height from which the weight 212 was dropped (drop height) and the number of times it was dropped (number of drops), and calculates the ratio of the displacement amount to the square root (step S107). This step S107 is performed for each of the multiple leveling frames R that make up the top surface 30, the multiple sections that make up these leveling frames R, and the entire area of the top surface 30. In other words, this step S107 is an example of a step that calculates the ratio of the displacement amount to the square root of the product of the height from which the weight 212 was dropped and the number of times it was dropped, for each of the multiple sections obtained by dividing the top surface of the rubble mound, and for the entire area of the top surface of the rubble mound.

[0121] Furthermore, the leveling frame R is a group consisting of two or more sections into which the weight 212 is moved and dropped. In other words, step S107 is an example of a step of calculating the ratio for each of the multiple sections on the top surface, the entire area of the top surface, and for each group consisting of two or more sections into which the weight is moved and dropped among the multiple sections.

[0122] Once the ratio is calculated, the learning device 1 uses this ratio along with the conditions of the strike to generate first teacher data 121. Furthermore, if first teacher data 121 has already been generated, the learning device 1 updates this first teacher data 121 using the new calculated ratio (step S108). As a result, the first teacher data 121 shown in FIG. 8 is generated in the memory 12 of the learning device 1. In other words, step S108 is an example of a step for generating teacher data in which the explanatory variables are the height at which the weight was dropped, the number of times, the previous displacement indicating the displacement in the height of the rubble mound due to the previous strike, the cumulative displacement in the height of the rubble mound due to previous strikes up to the previous time, and the ratio of the displacement to the square root of the product of the drop height and the number of drops in multiple sections of the top surface and the entire area of the top surface, and the objective variable is the displacement due to the current strike.

[0123] Furthermore, the leveling frame R is a group consisting of two or more sections into which the weight 212 is moved and dropped. In other words, step S108 is an example of a step for generating training data in which the explanatory variables are the height at which the weight has been dropped, the number of times, the previous displacement, the cumulative displacement up to the previous time, and the ratio of the displacement to the square root of the product of the drop height and the number of drops for the multiple sections on the top surface, the group of two or more sections among those sections into which the weight is moved and dropped, and the entire area of the top surface, and the displacement due to the current strike is the objective variable.

[0124] After generating or updating the first teacher data 121, the learning device 1 constructs or updates a first learning model using the first teacher data 121 (step S109). Step S109 is an example of a step of constructing a first learning model using the generated or updated teacher data.

[0125] The control device 4 determines whether the weight leveling process has been completed for all sections of the top surface 30 (step S110). If it determines that the weight leveling process has not been completed for all sections (step S110; NO), the control device 4 assigns an unprocessed section among the multiple sections of the leveling frame R to the leveling machine 2 in the above-mentioned order, and returns the process to step S101. On the other hand, if it determines that the weight leveling process has been completed for all sections of the leveling frame R (step S110; YES), the control device 4 ends the process.

[0126] When step S110 becomes YES (when the control device 4 determines that weight compaction of all sections within the leveling frame R has been completed), the operator instructs the leveling machine 2 to scan the leveling height sensor 23 and measure the height of the top surface 30 as the aforementioned post-measurement. The result of this measurement is transmitted from the leveling machine 2 to the control device 4 via the management device 5, and the data is stored in the management device 5.

[0127] After confirming in a post-measurement that the entire area within the leveling frame R has been compacted to the design height, the operator moves the leveling machine 2 to the next leveling frame R. As a result, the process of construction method M returns to the caller of step S100 and proceeds to step S004 (see FIG. 12).

[0128] As described above, according to the construction method M, first training data is generated in which the explanatory variables are the height at which the weight has been dropped, the number of times, the previous displacement of the height of the riprap mound due to the previous strike, the cumulative displacement of the height of the riprap mound due to the previous strikes up to the previous time, and the ratio of the displacement to the square root of the product of the drop height and the number of drops, for multiple sections of the top surface and the entire area of the top surface, and the displacement is the objective variable. And according to this construction method M, the height at which the weight 212 will be dropped is estimated by the first learning model constructed using the first training data, so it is possible to set the drop height of the weight 212 during weight compaction without relying on the empirical rules of workers, etc.

[0129] <Modification> The above is a description of the embodiment, but the contents of this embodiment can be modified as follows: In addition, the following modifications may be combined.

[0130] <1> In the above-described embodiment, the first training data 121 was training data in which the explanatory variables were the height at which the weight was dropped, the number of times, the previous displacement of the height of the rubble mound due to the previous strike, the cumulative displacement of the height of the rubble mound due to the previous strikes up to the previous time, and the ratio of the displacement to the square root of the product of the drop height and the number of drops in multiple sections of the top surface and the entire area of the top surface, and the displacement amount was the objective variable, but training data in which other indicators are used as objective variables may also be generated. For example, the objective variable of the training data may be the displacement of "another section" different from the section that was struck.

[0131] Fig. 18 is a diagram showing an example of the configuration of a learning device 1 in a modified example. The learning device 1 shown in Fig. 18 stores second teacher data 123 and a second learning model 124 in memory 12 instead of or in addition to first teacher data 121 and first learning model 122.

[0132] 19 is a diagram showing an example of second teacher data 123 in a modified example. Like the first teacher data 121, the second teacher data 123 assigns a data ID, which is identification information that identifies the data, to each strike, and stores the number of drops, drop height, previous section constant, previous frame constant, previous overall constant, previous displacement amount, and cumulative displacement amount up to the previous time as explanatory variables.

[0133] On the other hand, the second teacher data 123 differs from the first teacher data 121 in that the objective variable is the area surrounding a certain area where a strike has been made, and stores the displacement amount (peripheral displacement amount) of any area different from the certain area (also called a peripheral area) within the smoothing frame R. This second teacher data 123 is used to construct a second learning model 124 for estimating the peripheral displacement amount.

[0134] Here, the amount of peripheral displacement is not limited to the amount of subsidence, because when a certain section is struck and the section sinks, rubble and other debris may be pushed out into the surrounding sections, causing them to rise.

[0135] 20 is a diagram showing an example of the functional configuration of a learning device 1 in a modified example. Teacher data generation means 112 shown in Fig. 20 generates second teacher data 123 in addition to or instead of the first teacher data 121 described above.

[0136] In addition, the learning model construction means 113 shown in Figure 20 constructs a second learning model 124 using second teacher data 123 in addition to or instead of constructing a first learning model 122 using the above-mentioned first teacher data 121.

[0137] Since the second teacher data 123 uses the peripheral displacement amount as the dependent variable, the second learning model 124 includes a group of parameters for deriving the peripheral displacement amount, which is the dependent variable, from the above-mentioned explanatory variables.

[0138] In other words, in this modified example, construction method M includes a step of generating second teaching data 123 in which the explanatory variables are the section onto which the weight was dropped, height, number of times, previous displacement amount, cumulative displacement amount up to the previous time, and the ratio of displacement amount to the square root of the product of the drop height and the number of drops in multiple sections of the top surface and the entire area of the top surface, and the objective variable is the displacement amount of other sections different from the section onto which the weight was dropped.

[0139] In addition, in this modified example, construction method M includes a step of constructing second learning model 124 using generated second teacher data 123.

[0140] The processor 11 shown in FIG. 20 also functions as a peripheral displacement amount calculation means 116 and a notification means 117.

[0141] The peripheral displacement amount calculation means 116 calculates the amount of displacement caused in the peripheral section by the current strike using the constructed second learning model 124. In other words, the peripheral displacement amount calculation means 116 is an example of a means for executing the step of calculating the amount of displacement of another section caused by dropping a weight on a certain section using the second learning model.

[0142] The notification means 117 notifies the estimated displacement of the surrounding section calculated by the surrounding displacement amount calculation means 116 by displaying it on the display unit 15, or by displaying it on the display unit 55 of the management device 5 via the interface 13. In other words, the notification means 117 is an example of a means for executing a step of notifying the calculated estimated displacement of another section.

[0143] According to the construction method M of this modification, it is possible to know the estimated amount of displacement that will occur in other sections due to the current strike.

[0144] <2> In the above-described embodiment, the learning device 1 used the first learning model 122 to estimate the "drop height," which is the appropriate height from which the weight 212 should be dropped, but it may also predict a state in which the displacement amount will be less than the threshold value regardless of the drop height.

[0145] Fig. 21 is a diagram showing an example of the functional configuration of a learning device 1 in a new modified example. By executing a program, the processor 11 shown in Fig. 21 functions as a measurement value acquisition means 111, a teacher data generation means 112, and a learning model construction means 113, as well as a displacement amount prediction means 118, a comparison means 119, and a notification means 117.

[0146] The displacement amount prediction means 118 uses the first learning model 122 to assume multiple drop heights for each of the multiple sections, and determines whether the section is in a state where any strike will cause the displacement amount to be less than the threshold, regardless of the drop heights. When determining that this state is occurring, the displacement amount prediction means 118 notifies the comparison means 119 and the notification means 117 that it has predicted this state. The displacement amount prediction means 118 is an example of a means for executing a step of using the first learning model to predict whether a state will occur in which the displacement amount caused by dropping a weight will be less than the threshold in any of the multiple sections.

[0147] When the displacement amount prediction means 118 predicts the above-mentioned state, the comparison means 119 compares the current height of the top surface 30 of the rubble mound 3 with a predetermined height, which is the design height. In other words, the comparison means 119 is an example of a means for executing a step of comparing the top surface of the rubble mound predicted to be in a state in which the displacement amount will be less than the threshold value with the design compaction height.

[0148] When the displacement amount prediction means 118 predicts the above-mentioned state, the notification means 117 notifies the user that this state has been predicted by display unit 15 or by displaying it on display unit 55 of management device 5 via interface 13. Furthermore, the notification means 117 notifies the user of the comparison result by the comparison means 119 by display unit 15 or by displaying it on display unit 55 of management device 5. In other words, the notification means 117 is an example of a means for executing a step of notifying that a state in which the displacement amount will be less than the threshold has been predicted.

[0149] According to this modified example, it is possible to determine whether the height of the top surface 30 of the rubble mound 3 will no longer be displaced by more than a threshold value due to the impact of the free fall of the weight 212. Furthermore, according to this modified example, when the height of the top surface 30 of the rubble mound 3 has stopped being displaced by more than a threshold value, it is possible to determine how close that height is to a predetermined height, which is the design height.

[0150] In this modified example, the displacement amount prediction means 118 predicts whether or not the above-mentioned state will occur in any of the multiple sections using the first learning model 122, but it may also predict this using the second learning model 124. In this case, the displacement amount prediction means 118 is an example of a means for executing a step of predicting, using the second learning model, whether or not a state will occur in any of the multiple sections in which the displacement amount of another section due to the dropping of a weight will be less than a threshold value.

[0151] <3> In the above-described embodiment, the construction method M is performed by the learning device 1, the leveling machine 2, the control device 4, and the management device 5, but it may also be performed by other configurations. For example, the construction method M may be performed using a crane mounted on a crane ship stopped above the sea surface W, a weight suspended from the crane, and an information processing device installed in the crane ship.

[0152] In addition, construction method M may simultaneously use first teacher data 121 and second teacher data 123 generated by teacher data generation means 112 to construct a first learning model 122 and a second learning model 124, and use both learning models in combination. [Explanation of symbols]

[0153] 1...Learning device, 11...Processor, 111...Measurement value acquisition means, 112...Teacher data generation means, 113...Learning model construction means, 114...Fall height estimation means, 115...Display control means, 116...Peripheral displacement calculation means, 117...Notification means, 118...Displacement prediction means, 119...Comparison means, 12...Memory, 121...First teacher data, 122...First learning model, 123...Second teacher data, 124...Second learning model, 13...Interface, 14...Operation unit, 15 ...Display unit, 2...Leveling machine, 20...Base, 21...Thrusting machine, 210...Lifting wire, 211...Weight weight height sensor, 212...Weight weight, 213...Weight weight fitting base, 214...Base height sensor, 215...Weight weight lifting winch, 216...Weight weight traveling carriage, 22...Rake device, 23 (23-1, 23-2, 23-3, 23-4, 23-5)...Leveling height sensor, 24...Communication line, 25...Positioning device, 26...Control line, 27...Azimuth inclinometer, 3...Riprap mound, 30...Top surface, 4...Control device, 41...processor, 42...memory, 43...interface, 44...operation unit, 45...display unit, 5...management device, 51...processor, 52...memory, 53...interface, 54...operation unit, 55...display unit, C...center line, D1 to D5...arrows (movement path of leveling machine sensor), P...positioning point, R...leveling frame, R1 to R8...sections

Claims

1. a step of measuring the height of the top surface of the rubble mound before striking with a weight for each of a plurality of sections obtained by dividing the top surface of the rubble mound; a step of dropping the weight from a set height onto each of the plurality of sections a plurality of times to strike the section, and measuring the amount of displacement in height of the rubble mound due to the strikes; Calculating a ratio of the displacement amount to the square root of the product of the height at which the weight is dropped and the number of times the weight is dropped for each of the plurality of sections and the entire area of ​​the top surface; the height at which the weight was dropped, the number of times, a previous displacement amount indicating the displacement amount of the height of the rubble mound due to the previous strike, and a previous cumulative displacement amount indicating the cumulative amount of displacement of the height of the rubble mound due to the previous strikes, in the plurality of sections and the entire area; constructing a first learning model using training data in which the ratio is an explanatory variable and the displacement amount is a response variable; using the first learning model to estimate a height at which the weight will be dropped onto the rubble mound; a step of dropping the weight from the height estimated in the estimating step and striking the top surface of the rubble mound with the weight; A construction method for a rubble mound equipped with:

2. In the step of calculating the ratio, the ratio of the entire region is any one of an arithmetic mean value, a mode value, a median value, a geometric mean value, and a maximum value of all the ratios of the plurality of sections. A method for constructing a rubble mound according to claim 1.

3. the step of calculating the ratio includes calculating the ratio for each of the plurality of sections, the entire area, and for each group of two or more sections among the plurality of sections into which a weight is moved and dropped, The constructing step is a step of constructing the first learning model using training data in which the height at which the weight is dropped, the number of times, the previous displacement amount, the cumulative displacement amount up to the previous time, and the ratio are used as explanatory variables for the section, the group, and the entire region, and the displacement amount is used as a response variable. A construction method for a rubble mound according to claim 1 or 2.

4. In the step of calculating the ratio, the ratio of the group is any one of the arithmetic mean, mode, median, geometric mean, or maximum value of the ratios of all the sections belonging to the group. A method for constructing a rubble mound according to claim 3.

5. constructing a second learning model using training data in which the height at which the weight was dropped, the number of times, the previous displacement amount, the cumulative displacement amount up to the previous time, and the ratio in the plurality of sections and the entire region are used as explanatory variables, and the displacement amount of a section other than the section on which the weight was dropped is used as a response variable; calculating, using the second learning model, displacements of other compartments caused by dropping the weight onto one of the plurality of compartments; notifying the calculated displacement amount of the other section; 5. A rubble mound construction method according to claim 1, further comprising:

6. Using the first learning model, predicting whether a state will occur in which the displacement amount caused by dropping the weight will be less than a threshold value in any of the plurality of sections; notifying that a state in which the displacement amount will be less than the threshold value is predicted; a step of comparing a state in which the displacement amount is less than the threshold value with a predetermined height of the top surface of the rubble mound predicted; 6. A method for constructing a rubble mound according to any one of claims 1 to 5, comprising:

7. Using the second learning model, a step of predicting whether a state will occur in any of the plurality of compartments in which the displacement amount of the other compartment due to the dropping of the weight will be less than a threshold value; notifying that a state in which the displacement amount of the other section is predicted to be less than the threshold value; A step of comparing a state in which the displacement amount of the other section is less than the threshold value with a predetermined height and the top surface of the rubble mound predicted; 6. The method for constructing a rubble mound according to claim 5, comprising:

8. The step of measuring the displacement of the height of the rubble mound due to the striking uses the depth of at least three points on the top surface of the weight measured by a measuring instrument. A construction method for a rubble mound according to any one of claims 1 to 7.

Images