Blasting information processing device, blasting information processing method, and trained model

Abstract

To provide a blasting information processing device capable of easily collecting teacher data for forming a learned model used for setting a blasting pattern.SOLUTION: A blasting information processing device 5 comprises: an input data acquiring part 502 for acquiring the position of a hole H formed in a prescribed cross section perpendicular to a cutting face R out of multiple holes formed in the cutting face R, hole-drilling energy when drilling the hole H and an amount of explosive charged in the hole H, as input data; an output data acquiring part 503 for acquiring the shape of a prescribed cross section C of a region A formed by explosion of the explosive charged in the multiple holes H, as output data; and a model forming part which executes machine learning on the basis of the input data and the output data and creates a learned model showing relative relations of the position of the hole H, hole-drilling energy and the explosive amount with the shape of the prescribed cross section C.SELECTED DRAWING: Figure 5

Description

[Technical Field]

[0001] The present invention relates to a blasting information processing device, a blasting information processing method executed in the blasting information processing device, and a trained model generated by the blasting information processing device. [Background technology]

[0002] Conventionally, when constructing mountain tunnels through medium-hard rock, the excavation is generally carried out by blasting. The positions of the drill holes into which explosives are to be loaded and the amount of explosives to be loaded into the drill holes are determined based on the experience of skilled engineers.

[0003] However, even for experienced engineers, it is difficult to achieve the designed shape of the ground after blasting. Attempts have been made to machine-learn the judgment of the quality of blasting by experienced blasters, using blasting conditions based on the scattering and accumulation of debris after blasting and the excavation condition of the face as indicator values, without relying on the skills, experience, and judgment of experienced blasters. By using this trained model, it may be possible to set the optimal blasting pattern (Patent Document 1). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2019-190189 Summary of the Invention [Problem to be solved by the invention]

[0005] However, conventional technology is complicated by the need to measure and input the state of debris scattering and accumulation. Furthermore, the learning process requires the judgment of skilled workers. Furthermore, because indicators of the blasting status and data indicating the quality of the blasting from a single blasting are used as training data for machine learning, only one set of training data can be obtained from a single blasting, and collecting training data takes a lot of time and money.

[0006] The present invention aims to provide a blasting information processing device that can easily collect training data for generating a learning model used to set blasting patterns and perform machine learning using the collected training data, a blasting information processing method executed in the blasting information processing device, and a trained model generated by the blasting information processing device. [Means for solving the problem]

[0007] The invention of claim 1 is a blasting information processing device used in a blasting method in which explosives loaded in each of multiple drill holes arranged at a working face are detonated to perform a single blast, and the blasting information processing device comprises: an input data acquisition unit that acquires, as input data for each of the specified cross sections, the positions of the drill holes formed within a specified cross section perpendicular to the working face, the drilling energy when the drill holes are drilled, and the amount of explosives loaded in the drill holes; an output data acquisition unit that acquires, as output data for each of the specified cross sections, the shape of the specified cross section of the area formed by the explosion of the explosives loaded in the multiple drill holes; and a model generation unit that performs machine learning based on the input data and the output data to generate a trained model showing the correlation between the positions of the drill holes, the drilling energy, and the amount of explosives and the shape of the specified cross section.

[0008] The invention of claim 2 is a blasting information processing device as described in claim 1, further comprising a calculation unit that calculates the projection position of another perforation formed at a position within a predetermined distance from the specified cross section, projected onto the specified cross section, and the input data acquisition unit further acquires the projection position calculated by the calculation unit, the drilling energy when the other perforation hole is drilled, and the amount of explosives loaded in the other perforation hole as other input data, and the model generation unit performs machine learning based on the input data, the other input data, and the output data, and generates a trained model that shows the correlation between the position of the perforation hole, the projection position, the drilling energy, and the amount of explosives and the shape of the specified cross section.

[0009] The invention of claim 3 is a blasting information processing device as described in claim 1 or claim 2, wherein the specified cross section includes at least one of a vertical cross section in which the face R is divided by drawing multiple vertical cutting lines in a left-right direction relative to the face, a horizontal cross section in which the face R is divided by drawing multiple horizontal cutting lines in a top-bottom direction relative to the face, and a cross section inclined from either of these.

[0010] The invention of claim 4 is a blasting information processing method used in a blasting method in which explosives loaded in each of a plurality of boreholes arranged at a working face are detonated to perform a single blast, and includes the steps of: acquiring, as input data for each of the predetermined cross sections, the positions of the boreholes formed within a predetermined cross section perpendicular to the working face, the drilling energy when the boreholes are drilled, and the amount of explosives loaded in the boreholes; acquiring, as output data for each of the predetermined cross sections, the shape of the predetermined cross section of the area formed by the explosion of the explosives loaded in the plurality of boreholes; and performing machine learning based on the input data and the output data to generate a trained model showing the correlation between the positions of the boreholes, the drilling energy, and the amount of explosives, and the shape of the predetermined cross section.

[0011] The invention of claim 5 is a blasting information processing method as described in claim 4, further comprising calculating a projection position of another perforation formed at a position within a predetermined distance from the specified cross section, projecting the projection position onto the specified cross section, further acquiring the projection position, the drilling energy when the other perforation hole is drilled, and the amount of explosives loaded in the other perforation hole as other input data, and performing machine learning based on the input data, the other input data, and the output data to generate a trained model showing a correlation between the position of the perforation hole, the projection position, the drilling energy, and the amount of explosives and the shape of the specified cross section.

[0012] The invention of claim 6 is a blasting information processing method as described in claim 4 or claim 5, wherein the specified cross section includes at least one of a vertical cross section in which the face R is divided by drawing multiple vertical cutting lines in a left-right direction relative to the face, a horizontal cross section in which the face R is divided by drawing multiple horizontal cutting lines in a top-bottom direction relative to the face, and a cross section inclined from either of these.

[0013] The invention of claim 7 is a trained model generated based on input data and output data in a blasting information processing device used in a blasting method in which explosives loaded in each of multiple drill holes arranged at a working face are detonated to perform a single blast, wherein the input data is obtained for each of the predetermined cross sections from the positions of the drill holes formed within a predetermined cross section perpendicular to the working face, the drilling energy when the drill holes are drilled, and the amount of explosives loaded in the drill holes, and the output data is obtained for each of the predetermined cross sections from the shapes of the predetermined cross sections of the areas formed by the explosion of the explosives loaded in the drill holes, and the trained model is generated by performing machine learning based on the input data and the output data and shows the correlation between the positions of the drill holes, the drilling energy, and the amount of explosives and the shape of the predetermined cross section.

[0014] The invention of claim 8 is a trained model generated based on the input data and output data in addition to other input data, wherein the other input data is obtained for each of the predetermined cross sections by projecting onto the predetermined cross section the projection position of another perforation formed at a position within a predetermined distance from the predetermined cross section, the drilling energy when the other perforation is drilled, and the amount of explosives loaded in the other perforation, and is a trained model as described in claim 7 that indicates a correlation between the position of the perforation, the projection position, the drilling energy, and the amount of explosives and the shape of the predetermined cross section, generated by performing machine learning based on the input data, the other input data, and the output data.

[0015] The invention of claim 9 is a trained model as described in claim 7 or claim 8, wherein the predetermined cross section includes at least one of a vertical cross section obtained by dividing the face R by drawing multiple vertical cutting lines aligned in the left-right direction relative to the face, a horizontal cross section obtained by dividing the face R by drawing multiple horizontal cutting lines aligned in the up-down direction relative to the face, and a cross section obtained by inclining any of these. [Effects of the Invention]

[0016] The present disclosure makes it possible to easily collect training data for generating learning models that are used to set blasting patterns. [Brief explanation of the drawings]

[0017] [Figure 1] FIG. 1 is a diagram illustrating an example of work performed using a blasting method. [Figure 2] FIG. 1 shows an example of a device used to excavate a tunnel. [Figure 3] FIG. 10 is a diagram illustrating an example of drilling positions indicated by a blasting pattern. [Figure 4] FIG. 2 is a diagram illustrating an example of the hardware configuration of a blasting information processing device. [Figure 5] FIG. 2 is a diagram illustrating an example of the function of a blasting information processing device. [Figure 6] FIG. 2 is a diagram showing an example of a predetermined cross section perpendicular to the working face. [Figure 7] FIG. 10 is a diagram showing an example of a predetermined cross-sectional shape. [Figure 8] FIG. 1 is a diagram illustrating an example of machine learning using a neural network. [Figure 9] FIG. 10 is a diagram illustrating an example of the flow of processing executed by a blasting information processing device. [Figure 10] FIG. 2 is a block diagram illustrating an example of the function of a blasting information processing device. [Figure 11] FIG. 10 is a diagram illustrating a process executed by a calculation unit. DETAILED DESCRIPTION OF THE INVENTION

[0018] An embodiment of the present invention will be described below with reference to the drawings. Note that not all combinations of features described in the following embodiment are necessarily required to solve the problem. In addition, more detailed description than necessary may be omitted. Furthermore, the following description of the embodiment and the drawings are provided to enable those skilled in the art to fully understand the present invention, and are not intended to limit the scope of the claims.

[0019] First Embodiment When constructing mountain tunnels and excavating rock masses such as medium-hard rock, the blasting method is generally used. This method involves drilling holes in the rock mass and detonating explosives inside to break up the rock.

[0020] Figure 1 is a diagram illustrating an example of work using the blasting method. A drilling machine 1 drills multiple holes (not shown) in the tunnel face R, which are required to excavate one excavation length S. The drilling machine 1 drills holes at positions determined based on a predetermined blasting pattern.

[0021] The boreholes are loaded with explosives, the amount of explosives to be loaded into each borehole also being determined based on a predetermined blasting pattern.

[0022] The blasting takes place when explosives are loaded into the holes drilled at the positions specified by the blasting pattern. After the blasting, the debris is removed using a wheel loader (not shown). If necessary, checks and marks are taken before or after the debris is removed.

[0023] After the debris has been removed, concrete is sprayed onto the wall surface using a sprayer (not shown) to reinforce the natural ground. Once the concrete has hardened, rock bolts are driven into the wall.

[0024] After the rock bolts have been driven, drill holes are drilled at the face R to excavate another excavation length S, and explosives are loaded into each hole and detonated. These operations are repeated to excavate the tunnel.

[0025] Next, the equipment used to excavate the tunnel will be described.

[0026] 2 is a diagram showing an example of equipment used to excavate a tunnel. The equipment used to excavate a tunnel includes, for example, the above-mentioned drilling machine 1, as well as a blasting pattern management device 2, a communication line 3, and a shape measuring device 4.

[0027] The blasting pattern management device 2 is a device that manages the blasting patterns for one excavation length S. The blasting pattern includes information on the combination of the drilling position of the working face R and the amount of explosives to be loaded into the drilled hole. Blasting patterns include, for example, those that are predetermined based on the experience of skilled workers, known ones, and also those that are arranged according to the characteristics of the site.

[0028] Figure 3 is a diagram illustrating an example of drilling positions indicated by a blasting pattern. The blasting pattern includes information indicating the positions of multiple drill holes H to be drilled in the tunnel face R. The information indicating the positions of the multiple drill holes H is expressed, for example, by coordinate values with a predetermined position as the origin. If the cross-sectional shape of the tunnel is a horseshoe shape, part of which coincides with the shape of part of an ellipse, the origin is, for example, the intersection of the major and minor axes of the ellipse.

[0029] The blasting pattern may include data such as the angle of each of the drill holes H relative to the face R, the diameter of the drill holes H, and the depth of the drill holes H.

[0030] The blasting pattern management device 2 may manage a plurality of blasting patterns. That is, the blasting pattern management device 2 may manage a plurality of blasting patterns according to the size of the tunnel to be excavated, the hardness of the bedrock, etc.

[0031] The blasting pattern management device 2 is configured by, for example, a PC (Personal Computer). The blasting pattern management device 2 transmits data indicating a blasting pattern to the drilling machine 1 via the communication line 3.

[0032] The communication line 3 connects the blasting pattern management device 2, the drilling machine 1, and the shape measuring device 4 to each other. The communication line 3 may be a wired line or a wireless line. The communication line 3 is, for example, an internet line.

[0033] The drilling machine 1 is a device that drills holes in the working face R. The drilling machine 1 drills holes using, for example, a drill. The drilling machine 1 receives a blasting pattern from a blasting pattern management device 2. The drilling machine 1 drills holes based on the position of the drilling hole H indicated by the blasting pattern. The drilling machine 1 is, for example, a fully automatic computer jumbo or a semi-automatic computer jumbo.

[0034] The drilling machine 1 is equipped with, for example, a projector, which displays an image on the working face R showing the drilling positions, the amount of explosives loaded at each drilling position, etc. based on the blasting pattern. The drilling machine 1 uses a drill to drill holes at the drilling positions displayed on the working face R. Furthermore, a worker loads explosives into a predetermined drilling hole H using the amount of explosives displayed on the working face R as a guide. The amount of explosives loaded into the drilling hole H is actually measured, and the measured amount of explosives is associated with the position of the drilling hole H and input to, for example, the blasting pattern management device 2.

[0035] The drilling machine 1 is equipped with multiple guide shells, each of which has a hydraulic drifter that slides in the drilling direction. Each hydraulic drifter slides in the drilling direction to drill a hole in the face R. The drilling machine 1 is equipped with multiple sensors that detect predetermined physical quantities when drilling.

[0036] Each sensor detects, for example, the hydraulic pressure applied to the hydraulic drifter, the sliding speed of the hydraulic drifter, and the sliding distance of the hydraulic drifter. The drilling machine 1 obtains the drilling energy consumed when drilling each drilling hole H using the physical quantities detected by each sensor.

[0037] The shape measuring instrument 4 measures the shape of a portion corresponding to one excavation length S after the debris accumulated by blasting has been removed. The shape measuring instrument 4 is, for example, a three-dimensional shape measuring instrument. The three-dimensional shape measuring instrument is, for example, a three-dimensional laser scanner. The shape measuring instrument 4 identifies the area crushed by blasting based on the shape of the face R before and after blasting. For example, the shape measuring instrument 4 sets a predetermined coordinate system for the face R and identifies the shape of the natural ground before and after blasting using coordinate values in the predetermined coordinate system. Note that the shape measuring instrument 4 may measure the shape before the debris is removed, or may perform measurements excluding work such as digging and marking. In other words, the shape excavated by blasting is measured.

[0038] Next, the blasting information processing device 5 will be described. The blasting information processing device 5 learns the correlation between the position of drill hole H to be drilled in the working face R, the drilling energy, the amount of explosives loaded in drill hole H, and the shape of a predetermined cross section. The blasting information processing device 5 is connected to the communication line 3 shown in Figure 2, and acquires various data from the blasting pattern management device 2, the drilling machine 1, and the shape measuring device 4. The blasting information processing device 5 is composed of, for example, a PC. The blasting information processing device 5 is implemented with the blasting pattern management device 2, but may be provided separately.

[0039] 4 is a diagram showing an example of the hardware configuration of the blasting information processing device 5. The blasting information processing device 5 includes a CPU (Central Processing Unit) 51, a bus 52, a ROM (Read Only Memory) 53, a RAM (Random Access Memory) 54, a non-volatile memory 55, and an input / output interface 56.

[0040] The CPU 51 is a processor that, in accordance with a system program, controls the entire blasting information processing device 5. The CPU 51 reads the system program stored in the ROM 53 via the bus 52, and performs various processes based on the system program.

[0041] The bus 52 is a communication path that connects each piece of hardware within the blasting information processing device 5. Each piece of hardware within the blasting information processing device 5 exchanges data via the bus 52.

[0042] The ROM 53 is a storage device that stores a system program for controlling the entire blasting information processing device 5. The ROM 53 is a computer-readable storage medium.

[0043] The RAM 54 is a storage device that temporarily stores various data and functions as a work area for the CPU 51 to process various data.

[0044] The nonvolatile memory 55 is a storage device that retains data even when the power to the blasting information processing device 5 is turned off. The nonvolatile memory 55 stores various information such as machine learning programs and trained models. The nonvolatile memory 55 is a computer-readable storage medium. The nonvolatile memory 55 is configured, for example, by an SSD (Solid State Drive).

[0045] The input / output interface 56 is a communication path that connects an external device (not shown) to the bus 52. The blasting information processing device 5 is connected to the communication line 3 via the input / output interface 56.

[0046] Next, the function of the blasting information processing device 5 will be described.

[0047] 5 is a diagram illustrating an example of the function of the blasting information processing device 5. As described above, the blasting information processing device 5 learns the correlation between the position of the drill hole H drilled in the working face R, the drilling energy, the amount of explosives loaded in the drill hole H, and the shape of a predetermined cross section.

[0048] The blasting information processing device 5 includes a reception unit 501 , an input data acquisition unit 502 , an output data acquisition unit 503 , a model generation unit 504 , a model storage unit 505 , and an output unit 506 .

[0049] The reception unit 501 and output unit 506 are realized by the input / output interface 56 exchanging various types of data via the communication line 3. The input data acquisition unit 502, output data acquisition unit 503, and model generation unit 504 are realized by the CPU 51 executing arithmetic processing using the system program stored in the ROM 53, the machine learning program stored in the nonvolatile memory 55, and various types of data. The model storage unit 505 is realized by storing the results of arithmetic processing by the CPU 51 in the RAM 54 or the nonvolatile memory 55.

[0050] The reception unit 501 receives input of various data. For example, the reception unit 501 receives data indicating the blasting pattern and information indicating the actual measured value of the amount of explosives loaded in each bore hole H from the blasting pattern management device 2 via the communication line 3. The reception unit 501 also receives data indicating the drilling energy consumed when drilling each bore hole H from the drilling machine 1. The reception unit 501 also receives data indicating the shape after blasting from the shape measuring instrument 4.

[0051] The input data acquisition unit 502 acquires, from the data received by the reception unit 501, the positions of the drill holes H formed in a predetermined cross section perpendicular to the face R among the multiple drill holes H formed in the face R, the drilling energy when drilling the drill holes H, and the amount of explosives loaded in the drill holes H as input data. The amount of explosives loaded in the drill holes H is the amount of explosives actually loaded in the drill holes H. In other words, the amount of explosives loaded in the drill holes H is an actually measured value. However, the amount of explosives loaded in the drill holes H may not be an actually measured value, but may be an amount specified by the blasting pattern.

[0052] Fig. 6 is a diagram showing an example of a predetermined cross section perpendicular to the working face R. The cross section C perpendicular to the working face R is a cross section determined, for example, based on a blasting pattern. The position at which the cross section C is set may be determined, for example, based on the experience of a skilled engineer, or may be allocated at a predetermined interval.

[0053] Cross section C perpendicular to the face R can be, for example, multiple vertical cross sections obtained by dividing the face R by drawing multiple vertical cutting lines aligned horizontally relative to the face R, as shown in Figure 6. Cross section C perpendicular to the face R can also be multiple horizontal cross sections obtained by dividing the face R by drawing multiple horizontal cutting lines aligned vertically relative to the face R. Cross section C perpendicular to the face R can also be a so-called mesh that includes both of these. While meshing increases the complexity of the learning model, it also increases accuracy and can be applied to optimal blasting design. Cross section C perpendicular to the face R can also be inclined upward. Of course, inclined cross sections can also be combined to form a mesh. Cross sections C perpendicular to the face R do not have to be equally spaced. Cross section C perpendicular to the face R can also be a curved or bent surface, but note that excessive bending reduces the accuracy of learning. Figure 6 shows an example of multiple vertical cross sections dividing the face R into multiple sections horizontally.

[0054] The position of the cross section C perpendicular to the face R may be determined, for example, so that one cross section C includes many drill holes H. The cross section C perpendicular to the face R does not necessarily have to pass through the drill hole H, but may be determined so as to pass through the drill hole H approximately.

[0055] The input data is a dataset of data indicating the positions of perforations H to be formed within a predetermined cross section C perpendicular to the face R, the drilling energy when each perforation H is drilled, and the amount of explosives loaded in each perforation H. The input data includes a plurality of datasets corresponding to each of the plurality of cross sections C.

[0056] The output data acquisition unit 503 acquires the shape of a predetermined cross section C of the area formed by the explosion of the explosives loaded in the borehole H as output data.

[0057] 7 is a diagram showing an example of the shape of a certain predetermined cross section C indicated by the output data acquired by the output data acquisition unit 503. The dashed line indicates the designed excavation line L. The dashed-dotted line indicates the position of the drill hole H loaded with explosives. The output data acquisition unit 503 acquires the shape of the predetermined cross section C of the area A formed by the explosion of the explosives loaded in the drill hole H, based on the shape of the natural ground near the drilling face before and after blasting, which is identified by the shape measuring instrument 4.

[0058] The model generation unit 504 performs machine learning based on the input data and output data to generate a trained model that indicates correlations between at least the positions of the boreholes H, the drilling energy, and the amount of explosives, and the shape of a predetermined cross-section C. In other words, the model generation unit 504 learns the effects of the arrangement of the boreholes H within the cross-section C specified by the positions of the boreholes H, the drilling energy used to drill each borehole H, and the amount of explosives loaded in each borehole H on the shape of the cross-section C after blasting. The model generation unit 504 performs machine learning using a set of training data, including input data and output data obtained for one cross-section C. The model generation unit 504 performs supervised learning using, for example, a neural network.

[0059] FIG. 8 is a diagram illustrating an example of machine learning using a neural network. The neural network NN includes an input layer, a hidden layer, and an output layer. Each of the input layer, hidden layer, and output layer includes a plurality of neurons N. In the example shown in FIG. 8, the hidden layer has a single-layer structure, but the hidden layer may also have a multi-layer structure.

[0060] Input data is input to neurons N in the input layer. The output value output from neurons N in the input layer is multiplied by a weight and input to neurons N in the hidden layer. A transfer function is set in neurons N in the hidden layer, and output values are output from the hidden layer after passing through the transfer function. The output value output from the hidden layer is further multiplied by a weight and input to neurons N in the output layer. An output result is output from each neuron N in the output layer.

[0061] The weights in the intermediate layer are adjusted, for example, by backpropagation so that the output results from each neuron N in the output layer approach the output data. In other words, machine learning is performed so that each weight multiplied by the data input to each neuron N approaches the optimal solution.

[0062] The model storage unit 505 stores the trained model generated by the model generation unit 504.

[0063] The output unit 506 outputs the trained model stored in the model storage unit 505. The output unit 506 transmits the trained model to the blasting pattern management device 2 via the communication line 3, for example.

[0064] Next, the flow of processing executed by the blasting information processing device 5 will be described.

[0065] 9 is a diagram illustrating an example of the flow of processing executed by the blasting information processing device 5. First, the receiving unit 501 receives various data (step S1). As described above, the various data includes data indicating the blasting pattern, data indicating the drilling energy, and data indicating the shape of the natural ground after blasting.

[0066] Next, the input data acquisition unit 502 acquires input data (step S2). The input data is data indicating at least the positions of the drill holes H formed in a predetermined cross section C perpendicular to the face R among the multiple drill holes H formed in the face R, the drilling energy when the drill holes H are drilled, and the amount of explosives loaded in the drill holes H.

[0067] Next, the output data acquisition unit 503 acquires output data (step S3). The output data is data indicating the shape of a predetermined cross section C of the region A formed by the explosion of the explosive loaded in the borehole H.

[0068] Next, the model generation unit 504 performs machine learning to generate a trained model (step S4). The trained model indicates correlations between at least the position, drilling energy, and amount of explosive of the drilling hole H and the shape of the predetermined cross section C.

[0069] Next, the model storage unit 505 stores the trained model (step S5).

[0070] Finally, the output unit 506 outputs the trained model stored in the model storage unit 505 (step S6), and the process ends.

[0071] As described above, the blasting information processing device 5 includes an input data acquisition unit 502 that acquires as input data the positions of the drill holes H formed within a predetermined cross section C perpendicular to the face R among the multiple drill holes H formed at least in the face R, the drilling energy when the drill holes H are drilled, and the amount of explosives loaded in the drill holes H; an output data acquisition unit 503 that acquires as output data the shape of the predetermined cross section C of the area A formed by the explosion of the explosives loaded in the drill holes H; and a model generation unit 504 that performs machine learning based on the input data and output data to generate a trained model that shows the correlation between the positions of the drill holes H, the drilling energy, and the amount of explosives and the shape of the predetermined cross section C.

[0072] Therefore, the blasting information processing device 5 can easily collect a large amount of training data for generating a learning model used to set a blasting pattern, and can perform machine learning using the collected training data. That is, the blasting information processing device 5 can obtain training data for the number of cross sections C set at the working face R with a single blasting. As a result, the blasting information processing device 5 can easily generate a trained model that shows a correlation between at least the position of the drill hole H to be drilled at the working face R, the drilling energy consumed when drilling the drill hole H at the working face R, and the amount of explosives loaded in the drill hole H, and the shape of a predetermined cross section C in the area A formed by the explosion of the explosives loaded in the drill hole H.

[0073] The trained model output by the output unit 506 is used, for example, in the blasting pattern management device 2 to optimally design blasting patterns. For example, the drilling position and the amount of explosives loaded can be optimized based on the correlation between the position of drill hole H to be drilled in the face R, the drilling energy consumed when drilling drill hole H in the face R, the amount of explosives loaded in drill hole H, and the shape of a predetermined cross section C of area A formed by the explosion of the explosives loaded in drill hole H. As a result, excavation costs can be reduced. Furthermore, the face shape after blasting can be made closer to the design shape.

[0074] In the above-described embodiment, the input data acquisition unit 502 acquires, as input data, the drilling energy consumed when drilling the borehole H and the amount of explosives loaded in the borehole H, but this is not limited to this. The input data acquisition unit 502 may further acquire, as input data, information indicating the depth of the borehole H. In this case, it is possible to deal with cases where one excavation length is changed, and the model generation unit 504 generates a trained model indicating a correlation between the position of the borehole H formed within the specified cross section C, the depth of the borehole H, the drilling energy consumed when drilling the borehole H, and the amount of explosives loaded in the borehole H, and the shape of the specified cross section C.

[0075] <Second embodiment> The second embodiment of the present invention will be described below with reference to Figures 10 and 11. Note that a description of the same parts as in the first embodiment will be omitted, and the differences will be mainly described.

[0076] In the second embodiment, an example is described in which the blasting information processing device 5 further acquires the projection positions of other perforations that are not on the specified cross-section C but are formed at a position within a specified distance from the specified cross-section C, projected onto the specified cross-section C, and performs machine learning.

[0077] Figure 10 is a block diagram illustrating an example of the functions of the blasting information processing device 5. The blasting information processing device 5 shown in Figure 10 further includes a calculation unit 511 in addition to the functions of the blasting information processing device 5 shown in Figure 5. The calculation unit 511 is realized by the CPU 51 performing calculation processing using the system program stored in the ROM 53, the machine learning program stored in the non-volatile memory 55, and various data.

[0078] The receiving unit 501 receives, for example, data indicating the blasting pattern and information indicating the actual measured value of the amount of explosives loaded in each bore hole H from the blasting pattern management device 2 via the communication line 3. The receiving unit 501 also receives data indicating the drilling energy consumed when drilling each bore hole H from the drilling machine 1. The receiving unit 501 also receives data indicating the shape after blasting from the shape measuring instrument 4.

[0079] The calculation unit 511 calculates the projection positions of other drill holes formed at positions within a predetermined distance from the predetermined cross section C perpendicular to the working face R, projected onto the predetermined cross section C. The predetermined distance is determined taking into consideration whether the explosive force generated by the explosion of the explosives loaded in the other drill holes will have a significant effect on the shape of the drill holes at the predetermined cross section C after blasting. The predetermined distance may be determined based on experiments carried out in advance. Alternatively, the predetermined distance may be determined based on the experience of a skilled engineer.

[0080] Fig. 11 is a diagram illustrating the processing executed by the calculation unit 511. In Fig. 11, a position at a predetermined distance d from the predetermined cross section C is indicated by a dashed line. The calculation unit 511 projects another perforation Ho located at a position equal to or less than the predetermined distance d from the predetermined cross section C onto the predetermined cross section C, and calculates the projection position at which the other perforation Ho is projected. Note that the other perforation Ho may correspond to a different predetermined cross section C from the one to be projected, for example, an adjacent other predetermined cross section C, or may correspond to a further distant other predetermined cross section C.

[0081] When the predetermined cross section C is a vertical cross section, the calculation unit 511 calculates the projection position of the other perforation Ho projected in the horizontal direction toward the predetermined cross section C. When the predetermined cross section C is a horizontal cross section, the calculation unit 511 calculates the projection position of the other perforation Ho projected in the vertical direction toward the predetermined cross section C.

[0082] The input data acquisition unit 502 acquires, from the data accepted by the acceptance unit 501, the positions of drilling holes formed in a predetermined cross section C perpendicular to the face R among the multiple drilling holes formed in the face R, the drilling energy when the drilling holes are drilled, and the amount of explosives loaded in the drilling holes as input data. The input data acquisition unit 502 further acquires, as other input data, the projection positions calculated by the calculation unit 511, the drilling energy when other drilling holes Ho are drilled, and the amount of explosives loaded in the other drilling holes Ho.

[0083] The model generation unit 504 performs machine learning based on the input data, other input data, and output data to generate a trained model that indicates the correlation between the position, drilling energy, and amount of explosives of a drilling hole and the shape of a predetermined cross section C. Here, the input data is data indicating the position of a drilling hole formed in a predetermined cross section C perpendicular to the face R among multiple drilling holes formed in the face R, the drilling energy when the drilling hole is drilled, and the amount of explosives loaded in the drilling hole. In addition, the other input data is data indicating the projection position calculated by the calculation unit 511, the drilling energy when another drilling hole Ho is drilled, and the amount of explosives loaded in the other drilling hole Ho. In addition, the output data is data indicating the shape of the predetermined cross section C of the area A formed by the explosion of the explosives loaded in the multiple drilling holes.

[0084] The model generation unit 504 performs machine learning based on the input data, other input data, and output data, and generates a trained model that indicates correlations between at least the drilling position, projection position, drilling energy, and amount of explosives and the shape of a predetermined cross section C. The model generation unit 504 performs supervised learning using, for example, a neural network NN.

[0085] The model storage unit 505 stores the trained model generated by the model generation unit 504.

[0086] The output unit 506 outputs the trained model stored in the model storage unit 505. The output unit 506 transmits the trained model to the blasting pattern management device 2 via the communication line 3, for example.

[0087] As described above, the blasting information processing device 5 further includes a calculation unit 511 that calculates the projection position of another drill hole Ho formed at a position within a predetermined distance d from the predetermined cross section C, projected onto the predetermined cross section C, and the input data acquisition unit 502 further acquires the projection position calculated by the calculation unit 511, the drilling energy when the other drill hole Ho is drilled, and the amount of explosives loaded in the other drill hole Ho as other input data, and the model generation unit 504 performs machine learning based on the input data, the other input data, and the output data, and generates a trained model that shows the correlation between the hole position, projection position, drilling energy, and amount of explosives and the shape of the predetermined cross section C.

[0088] Therefore, the blasting information processing device 5 can easily collect training data for generating a learning model used to set blasting patterns, and perform machine learning using the collected training data.

[0089] [Other Modifications] The present invention is not limited to the above-described embodiment, and may also include the following, for example.

[0090] In this embodiment, the input data acquisition unit acquires, as input data, the positions (arrangement) of the drilling holes, the drilling energy consumed when the drilling holes are drilled, and the amount of explosives loaded in the drilling holes. However, this is not limited thereto, and the input data acquisition unit may also acquire, as input data, information indicating the depth of the drilling holes (drilling length). Furthermore, other data such as the type of explosives and the diameter of the drilling holes may also be acquired as input data. In this case, the model generation unit generates a trained model that shows a correlation between information including this and the shape of the predetermined cross-section C. Furthermore, information such as the rock type, fracture density, and spring water pressure at the drilling positions may also be acquired as other input data, and a trained model that shows a correlation between information including this and the shape of the predetermined cross-section C may be generated.

[0091] In this embodiment, a full-section blasting method is used, but the present invention is not limited to this and may also be applied to a full-section method with an auxiliary bench or an upper semi-circular method.

[0092] The technical aspects of any of the embodiments may be combined to form an example. [Explanation of symbols]

[0093] 1 drilling machine 2 Blasting pattern control device 3. Communication lines 4 Shape measuring device 5. Blasting information processing device 51 CPU 52 Bus 53 ROM 54 RAM 55 Non-volatile memory 56 Input / Output Interface 501 Reception 502 Input data acquisition unit 503 Output data acquisition unit 504 Model Generation Unit 505 Model Memory Unit 506 Output section 511 Calculation Unit

Claims

1. A blasting information processing device used in a blasting method in which explosives loaded in each of a plurality of drill holes arranged in a face are detonated to perform a single blast, an input data acquisition unit that acquires, as input data for each predetermined cross section, the positions of the drilling holes formed within a predetermined cross section perpendicular to the face, the drilling energy when the drilling holes are drilled, and the amount of explosives loaded in the drilling holes; an output data acquisition unit that acquires, as output data for each predetermined cross section, a shape of the predetermined cross section of an area formed by the explosion of the explosives loaded in the plurality of perforations; a model generation unit that performs machine learning based on the input data and the output data to generate a trained model that indicates a correlation between the position of the drilling, the drilling energy, and the amount of explosives and the shape of the predetermined cross section; A blasting information processing device comprising:

2. a calculation unit that calculates a projection position of another perforation formed at a position within a predetermined distance from the predetermined cross section, projected onto the predetermined cross section; The input data acquisition unit further acquires the projection position calculated by the calculation unit, the drilling energy when the other drilling hole is drilled, and the amount of explosives loaded in the other drilling hole as other input data, The blasting information processing device described in claim 1, wherein the model generation unit performs machine learning based on the input data, the other input data, and the output data, and generates a trained model showing the correlation between the position of the drilling hole, the projection position, the drilling energy, and the amount of explosives and the shape of the specified cross section.

3. The blasting information processing device according to claim 1 or claim 2, wherein the predetermined cross section includes at least one of a vertical cross section in which the face R is divided by drawing a plurality of vertical cutting lines in a left-right direction relative to the face, a horizontal cross section in which the face R is divided by drawing a plurality of horizontal cutting lines in a top-bottom direction relative to the face, and a cross section inclined from either of these.

4. A blasting information processing method used in a blasting method in which explosives loaded in each of a plurality of drill holes arranged in a face are detonated to perform a single blast, Acquiring, as input data for each predetermined cross section, the positions of the drilling holes formed within a predetermined cross section perpendicular to the face, the drilling energy when the drilling holes are drilled, and the amount of explosives loaded in the drilling holes; acquiring, as output data for each of the predetermined cross sections, a shape of the predetermined cross section of an area formed by the explosion of the explosives loaded in the plurality of boreholes; Performing machine learning based on the input data and the output data to generate a trained model that indicates a correlation between the position of the drilling hole, the drilling energy, and the amount of explosives and the shape of the predetermined cross-section; A blasting information processing method including:

5. further comprising calculating a projection position of another perforation formed at a position within a predetermined distance from the predetermined cross section, projected onto the predetermined cross section; Further obtaining the projection position, the drilling energy when the other drilling holes are drilled, and the amount of explosives loaded in the other drilling holes as other input data; Performing machine learning based on the input data, the other input data, and the output data to generate a trained model that indicates a correlation between the position of the drilling hole, the projection position, the drilling energy, and the amount of explosives and the shape of the predetermined cross section; 5. The blasting information processing method according to claim 4, further comprising:

6. The blasting information processing method described in claim 4 or claim 5, wherein the specified cross section includes at least one of a vertical cross section in which the face R is divided by drawing a plurality of vertical cutting lines in a left-right direction relative to the face, a horizontal cross section in which the face R is divided by drawing a plurality of horizontal cutting lines in a top-bottom direction relative to the face, and a cross section inclined from any of these.

7. In a blasting information processing device used in a blasting method in which explosives loaded in each of a plurality of drill holes arranged at a working face are detonated in a single blast, a trained model generated based on input data and output data, The input data includes the positions of the drilling holes formed within a predetermined cross section perpendicular to the face, the drilling energy when the drilling holes are drilled, and the amount of explosives loaded in the drilling holes, which are acquired for each of the predetermined cross sections, The output data is obtained by acquiring the shape of the predetermined cross section of the region formed by the explosion of the explosive loaded in the borehole for each predetermined cross section, A trained model that indicates the correlation between the position of the drilling hole, the drilling energy, and the amount of explosives and the shape of the specified cross-section, generated by performing machine learning based on the input data and the output data.

8. A trained model generated based on the input data and output data as well as other input data, The other input data is obtained for each of the predetermined cross sections, and includes a projection position calculated by projecting another perforation formed at a position within a predetermined distance from the predetermined cross section onto the predetermined cross section, a drilling energy when the other perforation is drilled, and an amount of explosives loaded in the other perforation, The trained model of claim 7, which indicates the correlation between the position of the drilling hole, the projected position, the drilling energy, and the amount of explosives and the shape of the specified cross section, is generated by performing machine learning based on the input data, the other input data, and the output data.

9. The predetermined cross section includes at least one of a vertical cross section obtained by dividing the face R by drawing a plurality of vertical cutting lines aligned in the left-right direction relative to the face, a horizontal cross section obtained by dividing the face R by drawing a plurality of horizontal cutting lines aligned in the up-down direction relative to the face, and a cross section obtained by inclining these. The trained model described in claim 7 or 8.

Images