Crusher control system, crusher operation control device and method

Abstract

To stabilize the quality of a crushed material.SOLUTION: A crusher control system (SYS1) includes: a particle size distribution estimation device (4) which estimates the particle size distribution of crushed materials by a crusher (12); and an operation control device (5) which controls the operation of the crusher on the basis of the comparison result between the particle size distribution estimated by the particle size distribution estimation device and the target particle size distribution set in advance. The particle size distribution estimation device includes: detection means that calculates a contour feature amount from a photographed image of the crushed material that freely falls; determination means that determines a particle size category being a first candidate on the basis of the proximity of the distribution of a contour feature amount for leaning for each particle size category obtained in advance on the basis of an image for learning and the contour feature amount identified by the detection means for each crushed material; and area integration means that calculates the total area for each particle size category by classifying all or part of the crushed material area calculated for each crushed material into the particle size category of the first candidate.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a crusher control system, and an apparatus and method for controlling the operation of a crusher. [Background technology]

[0002] The conventional method of controlling crushers involves manually measuring the particle size distribution at regular intervals and controlling the gap of the crushing section and the load factor of the motor. However, manual particle size adjustment requires constant monitoring, which places a heavy burden on the operator.

[0003] In the "Manufacturing method and plant for crushed sand" disclosed in JP-A-8-173835 (Patent Document 1), grain size information is detected from the discharged crushed sand and compared with previously collected reference grain size information. Based on the comparison result, grain size change parameters (feed amount, raw material grain size, etc.) related to grain size adjustment of the crushed sand are automatically controlled so that crushed sand with the desired coarse grain ratio can be produced.

[0004] In the "Manufacturing method and manufacturing device for crushed sand" disclosed in JP 2003-10726 A (Patent Document 2), the whole or part of the crushed sand is photographed to produce image information, and the particle size and particle size distribution or weight distribution are calculated using an image analysis means. The material supply means or crushing means is controlled so that the calculated particle size distribution or weight distribution falls within a predetermined allowable range.

[0005] In the "Automatic Control Method of Coal Particle Size" disclosed in JP 2004-16983 A (Patent Document 3), an image of coal on a belt conveyor that has been leveled by a pelletizing plate is captured by a CCD camera, and the image is processed to measure the average value of the coal particle size or a particle size distribution within a specified range. Based on the measured value, the rotation speed of the crusher is changed to automatically control the coal particle size. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Japanese Patent Application Publication No. 8-173835 [Patent Document 2] JP 2003-10726 A [Patent Document 3] JP 2004-16983 A Summary of the Invention [Problem to be solved by the invention]

[0007] In Patent Document 1, equipment (sieves, etc.) is required for particle size measurement, which increases the installation space and equipment costs.

[0008] In Patent Documents 2 and 3, the particle size distribution is measured by analyzing the photographed image of the material, but since the posture of the material on the conveyor that is photographed is not constant, the size of the material cannot be accurately grasped, and the particle size distribution may not be accurately measured. Therefore, even if the crusher is automatically controlled based on the measured particle size distribution, the target particle size distribution may not be obtained. In other words, the quality of the material crushed by the crusher (i.e., the crushed product) may not be stable.

[0009] The present invention has been made to solve the above-mentioned problems, and its object is to provide a crusher control system, a crusher operation control device, and a method that can stabilize the quality of crushed materials (aggregates, coal, iron ore, etc.). [Means for solving the problem]

[0010] A crusher control system according to an aspect of the present invention includes a particle size distribution estimation device that estimates the particle size distribution of crushed materials by the crusher, and an operation control device that controls the operation of the crusher based on a comparison result between the particle size distribution estimated by the particle size distribution estimation device and a target particle size distribution set in advance. The particle size distribution estimation device includes a detection means that detects a minor axis equivalent part and a major axis equivalent part of each crushed material from a photographed image of the crushed material falling freely and calculates a contour feature amount including at least the length of the minor axis equivalent part, a determination means that determines a first candidate particle size class for each crushed material based on the closeness of the distance between the distribution of learning contour feature amounts of each particle size class determined in advance based on a learning image and the contour feature amount specified by the detection means, a calculation means that calculates a crushed material area as an index value representing the amount of each crushed material from the photographed image, and an area accumulation means that calculates a total area for each particle size class by classifying all or a part of the crushed material area calculated by the calculation means into the first candidate particle size class for each crushed material.

[0011] Preferably, the operation control device includes an adjustment processing means for adjusting control parameters including at least one of the gap value of the crushing space in the crusher, the rotation speed of the rotating shaft of the crusher, and the amount of crushed material supplied to the crusher when the particle size distribution estimated by the particle size distribution estimation device is outside a target range.

[0012] Preferably, the crusher control system further includes a storage means for storing a learning model generated by inputting the current control parameters and the current particle size distribution estimated at that time and specifying a target particle size distribution so as to output an adjustment amount from the current control parameters, and the adjustment processing means calculates the adjustment amount of the control parameters by using the learning model.

[0013] It is desirable that the crusher control system further includes a learning device that generates the learning model.

[0014] A crusher operation control device according to another aspect of the present invention is an operation control device that controls the operation of a crusher, and includes an acquisition means for acquiring a particle size distribution of crushed material by the crusher estimated by a particle size distribution estimation device, and an adjustment processing means for adjusting control parameters including at least one of a clearance value of a crushing space in the crusher, a rotation speed of a rotating shaft of the crusher, and a crushed material supply amount to the crusher based on a comparison result between the particle size distribution acquired by the acquisition means and a target particle size distribution set in advance. The particle size distribution acquired by the acquisition means is a value obtained by calculating a total area for each particle size division by classifying all or a part of the crushed material area calculated from the photographed image into a first candidate particle size division determined based on the closeness of the distance between the distribution of learning contour feature values ​​of each particle size division previously obtained based on the learning image and the contour feature values ​​of each crushed material calculated from the photographed image of the crushed material.

[0015] A method for controlling the operation of a crusher according to another aspect of the present invention is a method for controlling the operation of a crusher, and includes the steps of acquiring a particle size distribution of the material crushed by the crusher, estimated by a particle size distribution estimation device, and adjusting control parameters including at least one of the gap value of the crushing space in the crusher, the rotation speed of the rotating shaft of the crusher, and the amount of crushed material supplied to the crusher, based on a comparison result between the acquired particle size distribution and a target particle size distribution set in advance. Effect of the Invention

[0016] According to the present invention, the operation of the crusher is controlled by comparing the accurately estimated particle size distribution of the crushed material with the target particle size distribution, so that the particle size distribution of the crushed material can be appropriately brought close to the target particle size distribution, and as a result, the quality of the crushed material can be stabilized. [Brief description of the drawings]

[0017] [Figure 1] FIG. 1 is a schematic diagram showing an overview of a crusher control system according to first and second embodiments of the present invention. [Diagram 2]FIG. 1A is a functional block diagram showing a functional configuration of an operation control device in a first embodiment of the present invention, and FIG. 1B is a diagram showing control items of a crusher. [Diagram 3] 4 is a flowchart showing the operation of the operation control device according to the first embodiment of the present invention. [Figure 4] FIG. 10(A) is a functional block diagram showing the functional configuration of an operation control device and a learning device in a second embodiment of the present invention, and FIG. 10(B) is a diagram showing variables learned by the learning device. [Diagram 5] 6 is a flowchart showing the operation of the operation control device according to the second embodiment of the present invention. [Figure 6] 1 is a schematic diagram showing an overview of a particle size distribution estimation system mounted on a crusher control system according to each embodiment of the present invention. FIG. [Figure 7] FIG. 1A is a schematic diagram for explaining a comparative example of a particle size distribution estimation system according to an embodiment of the present invention, and FIG. 1B and FIG. 1C are schematic diagrams showing the posture of aggregates freely falling from a conveying section. [Figure 8] FIG. 13 is a diagram showing a method for measuring the size of aggregates in an experiment based on a comparative example of a particle-size distribution estimation system according to an embodiment of the present invention. [Figure 9] FIG. 1A is a functional block diagram showing the functional configuration of a learning device in a particle size distribution estimation system according to an embodiment of the present invention, and FIG. 1B is a functional block diagram showing the functional configuration of a particle size distribution estimation device according to an embodiment of the present invention. [Figure 10] 1A and 1B are diagrams for explaining a method for estimating a gradation range in an embodiment of the present invention, in which (A) shows an approximation ellipse of an aggregate, and (B) shows a Gaussian distribution for each gradation range. [Figure 11] FIG. 4 is a diagram for explaining a method for estimating a grain size division in an embodiment of the present invention. [Figure 12] 1(A) and (B) are diagrams showing schematic diagrams illustrating the estimated particle size distribution. [Figure 13] 11 is a flowchart showing a distribution information generating process according to an embodiment of the present invention. [Figure 14]4 is a flowchart showing a particle size distribution estimation process according to an embodiment of the present invention. [Figure 15] FIG. 1A is a graph showing an example of an estimated particle size distribution in comparison with an allowable range of the particle size distribution, and FIG. 1B is a diagram showing an example of a determination result for each particle size category. [Figure 16] FIG. 4 is a functional block diagram showing a functional configuration of an operation control device in a modified example of the first embodiment of the present invention. [Figure 17] FIG. 13A is a diagram showing a specific example of a judgment table, and FIG. 13B is a diagram showing a specific example of parameter adjustment data. [Figure 18] FIG. 4 is a diagram showing a schematic diagram of a test result for an operation control device according to a modified example of the first embodiment of the present invention. [Figure 19] FIG. 4 is a diagram showing a schematic diagram of a test result for an operation control device according to a modified example of the first embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail with reference to the drawings. In the drawings, the same or corresponding parts are designated by the same reference characters and their description will not be repeated.

[0019] [Embodiment 1] <Summary> With reference to FIG. 1, an overview of a crusher control system SYS1 according to this embodiment will be described.

[0020] The crusher control system SYS1 mainly controls the crusher 12. The crusher control system SYS1 includes a particle size distribution estimation device (hereinafter abbreviated as "estimation device") 4 that estimates the particle size distribution of the crushed material by the crusher 12, and an operation control device 5 that controls the operation of the crusher 12. The "particle size distribution" is expressed as a cumulative passing ratio, with the mesh size of the sieve (each opening side of a flat-woven sieve) used in the sieve classification test being used as a particle size division. In this embodiment, an example will be described in which the crushed material by the crusher 12 is aggregate. Note that the crushed material by the crusher 12 is not limited to aggregate, and may be other types of processed materials, such as coal and iron ore.

[0021] First, a brief description will be given of the crusher 12. The crusher 12 crushes aggregates (raw materials) supplied from the material supply device 11, and discharges the crushed aggregates (crushed material) onto a conveyor 21. The material supply device 11 supplies raw materials in a hopper onto a belt using a material supply feeder such as a chute, and discharges the raw materials by rotating the belt. Note that the material supply device 11 is not limited to a belt type, and other types may be adopted.

[0022] The crusher 12 is typically a rotary crusher, and includes a rotating shaft that is rotated around a vertical axis inside a vertically placed cylindrical casing, a mantle that rotates integrally with the rotating shaft, and a bowl liner (cone cave) that is fixed to the inner wall of the casing so as to face the outer circumferential surface of the mantle. A hopper is provided at the upper end of the casing, and the raw material supplied from the hopper is crushed by the mantle and bowl liner. The space in which the raw material is crushed, such as the space (gap) between the mantle and the bowl liner, is called the "crushing space."

[0023] The estimation device 4 continuously captures images of the aggregates freely falling from the conveyor 21 with the camera 24 and estimates the particle size distribution of the aggregates by analyzing (image processing) the captured images. The estimation device 4 is typically realized by a particle size distribution estimation device described in the specification of Japanese Patent Application No. 2021-164452 filed by the present applicant. Such an estimation device 4 includes a detection means for detecting the part corresponding to the short diameter and the part corresponding to the long diameter of each aggregate from a photographed image of the freely falling aggregate and calculating a contour feature including at least the length of the part corresponding to the short diameter, a determination means for determining a first candidate gradation class for each aggregate based on the closeness of the distance between the distribution of learning contour feature values ​​for each gradation class determined in advance based on the learning image and the contour feature value specified by the detection means, a calculation means for calculating an aggregate area (crushed area) as an index value representing the amount of each aggregate from the photographed image, and an area accumulating means for classifying all or a part of the aggregate area calculated by the calculation means into the first candidate gradation class for each aggregate, thereby calculating a total area for each gradation class. Details of this estimation device 4 will be described later.

[0024] The operation control device 5 compares the particle size distribution estimated by the estimation device 4 with a preset target particle size distribution, and controls the operation of the crusher 12 according to the comparison result. In other words, the operation control device 5 performs feedback control of the crusher 12 so that aggregate having a target coarseness can be discharged from the crusher 12. The operation control device 5 is realized by an arithmetic processing device such as a PLC (Programmable Logic Controller).

[0025] <Estimation device> The estimation device 4 described in Japanese Patent Application No. 2021-164452 will be described in detail with reference to Fig. 6 to Fig. 14. As shown in Fig. 6, the estimation device 4 of the present embodiment is included in a particle size distribution estimation system 1.

[0026] Referring to Figure 6, the particle-size distribution estimation system 1 includes a photographing mechanism 2 for photographing a material 9 consisting of a large number of aggregates as it flows down the stream, and an estimation device 4 for estimating the particle-size distribution of the material 9 by image processing the image data photographed by the photographing mechanism 2.

[0027] The photographing mechanism 2 includes a conveyor (conveyor) 21 that conveys the material 9 forward and allows the material 9 to fall freely from the front end, a screen 22 provided on the rear side of the flow path of the material 9, an illumination unit 23 that illuminates the screen 22 from the front, and a camera (photographing means) 24 that is arranged facing the screen 22 and photographs the material 9 flowing down the flow path. The camera 24 continuously photographs the material 9 falling freely from the conveyor 21, i.e., the group of aggregates falling freely. The camera 24 may directly photograph the aggregates falling freely from the crusher 12.

[0028] The estimation device 4 estimates the grading classification of each aggregate by image processing of the image captured by the camera 24. This makes it possible to estimate the grading distribution of the material 9 without performing a manual sieve test.

[0029] Here, ideally, as shown in Fig. 7(A), the minimum circumscribing square R1 of the aggregate region 91 is regarded as a sieve mesh, and it is determined to which gradation class the side length L10 of the minimum circumscribing square R1 belongs, so that the gradation class of the aggregate (original aggregate detected as the aggregate region 91) can be appropriately estimated. However, it is considered that a flat aggregate 92 as shown by hatching in Figs. 7(B) and (C) takes a stable posture on the conveyor 21 with the long side x middle side facing upward. Fig. 7(B) is a top view of the aggregate 92 on the conveyor 21, and Fig. 7(C) is a side view of the aggregate 92 on the conveyor 21. Note that in these figures, the flat shape of the aggregate 92 is exaggerated.

[0030] Therefore, flat aggregate 92 tends to fall with its long side and middle side facing forward (toward camera 24) as shown in Fig. 7(C). Some aggregates fall while rotating, but if the falling aggregate is imaged from the front, the outline shape of the long side and middle side is easy to recognize.

[0031] On the other hand, in manual sieve testing, the size of the sieve mesh that will pass or not depends on the median side x short side. This is because even if the mesh does not pass when the long side x mid side is used, it may pass when the mid side x short side is used. Therefore, when estimating the aggregate's grain size classification using the minimum circumscribing square R1, the estimated grain size classification tends to be larger than the actual grain size obtained by the sieve testing, and this tendency is thought to be more pronounced for aggregates with a significant difference between the long side and short side (flat aggregates).

[0032] As an example, we measured the three sides (short side, middle side, long side) of aggregate in the 30-20mm gradation range with a vernier caliper, calculated the side length of a square circumscribing an ellipse formed by the middle side x short side or the long side x middle side (see Figure 8), and conducted an experiment to examine the effect of differences in the photographing direction on the estimation accuracy of the gradation range.The results of this experiment showed that when the photograph was taken with the middle side x short side facing forward, the side length of the square was concentrated on the actual sieve mesh (30-20mm), whereas when the photograph was taken with the long side x middle side facing forward, the side length of the square was greater than the actual sieve mesh (30-20mm).

[0033] Therefore, when estimating the gradation classification of the aggregate by extracting the minimum circumscribing square R1 of the aggregate area 91 as shown in Fig. 7(A), the aggregate may be classified into a gradation classification larger than the actual classification. Since the difference between the long side and the short side tends to increase as the size (grain size) of the aggregate increases, the error may increase as the gradation classification increases. If the middle side x short side of the aggregate could be photographed by the camera 24, such errors could be reduced, but this is difficult to achieve in practice.

[0034] Therefore, the estimation device 4 in this embodiment determines at least one gradation class as a candidate by a statistical method based on the "contour feature value" representing the contour shape of the aggregate, and estimates the gradation distribution of the material 9 by classifying the "aggregate area" into the gradation class determined as the candidate. This makes it possible to correct the above-mentioned errors.

[0035] The "contour feature amount" in this embodiment includes the short axis length and long axis length of an approximation ellipse of the aggregate region 91 extracted from the captured image (see FIG. 10(A) described later). The "approximation ellipse" is a typical example of an abstracted figure that abstracts the contour shape of the aggregate region 91. The "abstracted figure" is a figure in which the part corresponding to the short axis and the part corresponding to the long axis of the aggregate (aggregate region 91) can be expressed by two straight lines (preferably perpendicular to each other). Note that the part corresponding to the short axis and the part corresponding to the long axis do not depend on the actual short side, middle side and long side of the aggregate, but represent the short side part and the long side part that can be specified from the aggregate region 91.

[0036] "Aggregate area" is an index value representing the amount of each aggregate, and in this embodiment, the projected area of ​​the aggregate, i.e., the area of ​​aggregate region 91, is calculated as the aggregate area. Note that the contour feature amount and aggregate area can be calculated by converting the number of pixels (pixel) of the captured image into an actual value (mm).

[0037] 6, the estimation system 1 according to the present embodiment further includes a learning device 3 so as to enable determination of candidates for grain size classification by a statistical method. The learning device 3 and the estimation device 4 are configured by an information processing device (such as a general-purpose computer) including a processor such as a CPU (Central Processing Unit) and a non-volatile memory.

[0038] The learning device 3 calculates in advance the distribution of the contour feature values ​​of the aggregates (hereinafter referred to as "learning contour feature values") for each gradation class based on learning images of the aggregates for each gradation class. As a result, distribution information indicating the distribution of the learning contour feature values ​​for each gradation class is generated, and the generated distribution information is stored in the storage device of the estimation device 4.

[0039] In this embodiment, the particle size classification includes the smallest classification of less than 5 mm, and six classifications of (1) 5-13 mm, (2) 13-20 mm, (3) 20-25 mm, (4) 25-30 mm, (5) 30-40 mm, and (6) 40-50 mm. The method of setting the particle size classification is not particularly limited.

[0040] (About the learning device) FIG. 9(A) is a functional block diagram showing the functional configuration of the learning device 3.

[0041] The learning device 3 is equipped with an input unit 31 that inputs (acquires) multiple learning image data obtained by continuously photographing an aggregate group (sample) 9A (Figure 6) divided into grain size classifications, a learning image memory unit 32 that stores the multiple acquired learning image data by grain size classification, a contour detection unit 33 that performs image processing on the learning image data stored in the learning image memory unit 32 to detect an approximate ellipse of each aggregate and calculates learning contour features based on the detected approximate ellipses, a calculation unit 34 that calculates the distribution of the learning contour features for each grain size classification, and a distribution information memory unit 35 that stores the calculation results by the calculation unit 34 as distribution information D1.

[0042] The learning image storage unit 32 includes, for example, learning files equal in number to the number of granularity divisions, and each learning file stores learning image data of the corresponding granularity division.

[0043] The contour detection unit 33 performs image processing using, for example, deep learning to extract (detect) an aggregate region from the learning image, and calculates an approximation ellipse R2 of the aggregate region 91 as an "abstracted figure" that abstracts the contour shape of the aggregate region 91, as shown in Fig. 10(A). Also, it calculates the major axis length L1 and the minor axis length L2 of the approximation ellipse R2.

[0044] The calculation unit 34 calculates a two-dimensional Gaussian distribution for each grain size class using the major axis length L1 and the minor axis length L2 of the approximation ellipse R2 as variables. The major axis length L1 and the minor axis length L2 of the approximation ellipse R2 used in calculating the Gaussian distribution are "learning contour feature values."

[0045] 10B is an example of a graph showing two-dimensional Gaussian distributions of learning contour features for each granularity class using a probability ellipse. Here, Gaussian distributions G1 to G6 corresponding to the above-mentioned granularity classes (1) to (6) respectively are shown. Information on such Gaussian distributions G1 to G6 is stored as distribution information D1 in the distribution information storage unit 35. The information on the two-dimensional Gaussian distribution stored as distribution information D1 may be information that directly represents the probability distribution, or may be information on a normal distribution obtained by normalizing the probability distribution.

[0046] The functions of the contour detection unit 33 and the calculation unit 34 are realized by a processor executing software. The learning image storage unit 32 and the distribution information storage unit 35 are typically configured with a non-volatile memory. The input unit 31 is, for example, an interface that inputs image data from the camera 24 via a wired or wireless connection.

[0047] (Functional configuration of the estimation device) FIG. 9(B) is a functional block diagram showing the functional configuration of the estimation device 4.

[0048] The estimation device 4 includes a distribution information storage unit 46 for storing the distribution information D2 (D1) generated by the learning device 3, an input unit 41 for inputting (acquiring) a plurality of image data from the camera 24, i.e., a plurality of image data obtained by continuously photographing the material 9 freely falling from the conveyor 21, an image storage unit 42 for storing the acquired plurality of image data, a contour detection unit 43 for processing the image data stored in the image storage unit 42 to detect an approximate ellipse of each aggregate and calculate a contour feature value based on the detected approximate ellipse, an estimation processing unit 44 for determining at least a first candidate grain size category for each aggregate based on the contour feature value and classifying the aggregate area based on the determination result to estimate the grain size distribution of the material 9, and an output unit 45 for outputting the estimation result by the estimation processing unit 44. The distribution information D2 can be received from the learning device 3, for example, via a communication interface (not shown).

[0049] The contour detection unit 43, like the contour detection unit 33 of the learning device 3, extracts (detects) aggregate regions from the captured image by performing image processing using, for example, deep learning, and calculates (detects) an approximation ellipse R2 for each aggregate region 91, as shown in Fig. 10(A). In addition, the contour detection unit 43 calculates the major axis length L1 and the minor axis length L2 of the approximation ellipse R2 as the contour feature amount of the aggregate.

[0050] The estimation processing unit 44 determines candidates for the granularity classification by evaluating the closeness of the positional relationship between the calculated contour feature value and the Gaussian distributions G1 to G6 of the learning contour feature value of each granularity classification indicated by the distribution information D2 for each aggregate. Specifically, the estimation processing unit 44 determines a first candidate granularity classification based on the closeness of the distance between the calculated contour feature value and the Gaussian distributions G1 to G6. Then, the estimation processing unit 44 classifies all or part of the aggregate area into the first candidate granularity classification for each aggregate to calculate the total area for each granularity classification.

[0051] The closeness of the positional relationship with the Gaussian distributions G1 to G6 can typically be evaluated by the "Mahalanobis distance" which quantifies the degree of distance from a data group. Specifically, as shown in FIG. 11, the Mahalanobis distance between point P, whose X- and Y-coordinates are the long axis length L1 and short axis length L2 of a certain aggregate A, and each of the Gaussian distributions G1 to G6 (populations) is calculated. For example, if point P is closest to Gaussian distribution G3, the above granulometry classification (3) can be determined as the first candidate. In addition, the second candidate, third candidate, ... can be determined in order of distance.

[0052] When the estimation processing unit 44 determines candidates for the gradation class of the aggregate A as described above, it classifies the aggregate area (area of ​​the aggregate region 91) by weighting it according to the Mahalanobis distance. As a method for classifying the aggregate area of ​​the aggregate A after the candidate determination, the following method (classification method 1) can be adopted.

[0053] Classification method 1: The weighting coefficients of the granularity classes of multiple candidates including the first candidate are calculated and normalized to a total of 1 according to the distance, and the area of ​​the aggregate region 91 of aggregate A is weighted for each of the granularity classes of the multiple candidates and classified.

[0054] When the above-mentioned processing is completed for all the captured images, the estimation processing unit 44 calculates the total area (mm 2 ) to calculate the area frequency (%) for each particle size division, thereby calculating (estimating) the particle size distribution of the material 9. In this way, the estimation processing unit 44 functions as a determination means for determining candidates for particle size division, and as an area accumulation means for calculating the total area for each particle size division.

[0055] The output unit 45 outputs the particle size distribution of the material 9 estimated by the estimation processing unit 44. The output unit 45 is configured, for example, by a display unit including a display, and displays the area and area frequency for each particle size division in a table format as shown in FIG. 12(A). Alternatively, a particle size distribution curve as shown in FIG. 12(B) may be generated and displayed. The particle size distribution curve is a curve obtained by plotting the cumulative values ​​of the area frequency in ascending order of particle size division on a graph with particle size (mm) on the horizontal axis and cumulative passing ratio (%) on the vertical axis.

[0056] The functions of the contour detection unit 43 and the estimation processing unit 44 shown in Fig. 9(B) are realized by a processor executing software. The distribution information storage unit 46 is typically composed of a non-volatile memory, and the image storage unit 42 is composed of a volatile or non-volatile memory. The input unit 41 is, for example, an interface that inputs image data from the camera 24 via a wired or wireless connection.

[0057] In addition, the distribution information D2 (D1) generated by the learning device 3 may be stored on a cloud server (not shown), and in such a case, the estimation device 4 may download the distribution information D2 from the cloud server and store it in the distribution information memory unit 46 at the timing of estimating the particle size distribution of the material 9.

[0058] Furthermore, the output unit 45 may be configured with a communication interface for transmitting and receiving data to and from other computers, or may be configured with a writing device for writing data to a removable recording medium.

[0059] (How distribution information is generated) A distribution information generating method executed by learning device 3 will be described with main reference to Fig. 9(A) and Fig. 13. Fig. 13 is a flowchart showing the distribution information generating process.

[0060] 13, first, aggregate (sample 9A) is fed onto conveyor 21 for each gradation division, and the aggregate (hereinafter also referred to as "learning aggregate") freely flowing down from the end of conveyor 21 is continuously photographed by camera 24 (step S1). As a result, a plurality of image data (learning image data) for each gradation division are input to input unit 31 in chronological order.

[0061] The conveying speed of the sample 9A by the conveyor 21 may be the same as that of a general material conveyor, for example, about 1 m / s. The falling speed of the sample 9A when it falls freely from the front end of the conveyor 21 increases at an accelerating rate. If the installation height of the conveyor 21 based on the installation height of the camera 24 is about 1 m, the falling speed of the sample 9A flowing down the space in front of the camera 24 (flow path) is about 4.4 m / s. When the falling speed increases to 4.4 m / s, the dense aggregate group that had been flowing at 1 m / s up until that point spreads, and the overlapping of the aggregates is eliminated or reduced. Therefore, according to this embodiment, the camera 24 can capture the aggregate group in a spread state. The camera 24 continuously captures the falling sample 9A so that each aggregate is captured at least once.

[0062] The multiple (large number) image data input to the input unit 31 are recorded in chronological order as learning image data in a file of the corresponding granularity classification in the learning image storage unit 32.

[0063] When image data is acquired for each granularity classification, the contour detection unit 33 executes image processing of each image data, and detects (extracts) the area of ​​each aggregate, i.e., aggregate area 91, for each image data (step S3). In this embodiment, the shape of each aggregate is determined by image processing using instance segmentation, and the area of ​​each aggregate is extracted according to the determined shape. Instance segmentation is a type of image processing technology using deep learning, and is a technology that detects the position of each object and classifies the pixels that make up the object.

[0064] By using instance segmentation in image processing, shape information can be output while distinguishing each aggregate. In addition, since objects belonging to the same class can be classified as separate objects, the boundaries of areas where aggregates overlap can be obtained. Note that, although instance segmentation is used in this embodiment, other types of image processing techniques may be used as long as they are capable of detecting each individual aggregate.

[0065] Next, the contour detection unit 33 detects the approximate ellipse R2 of each aggregate region 91 detected in step S3 for each granularity classification as the "abstract figure" of each training aggregate (step S5). In addition, the major axis length L1 and minor axis length L2 of the approximate ellipse R2 are calculated as the contour feature of each training aggregate, i.e., the "training contour feature" (step S7). Depending on the shape of the aggregate, the approximate ellipse R2 may be a perfect circle, in which case the major axis length L1 and minor axis length L2 may be the same length.

[0066] The calculation unit 34 calculates a Gaussian distribution of the training contour feature values ​​obtained for each aggregate in step S7 for each gradation class (step S9). That is, a two-dimensional Gaussian distribution is calculated with the major axis length L1 and minor axis length L2 of the approximation ellipse R2 as variables. This generates distribution information showing the Gaussian distribution for each gradation class. In order to reduce the variation in accuracy of the Gaussian distribution, it is desirable to calculate the Gaussian distribution with the same number of training aggregates for each gradation class.

[0067] In the present embodiment, an example has been described in which learning device 3 directly inputs image data from camera 24 of imaging mechanism 2, but the invention is not limited to such an example. In other words, input unit 31 in learning device 3 may be configured with a communication interface that receives image data captured in advance from another computer, or a reading device that reads image data from a removable recording medium.

[0068] (Method of estimating particle size distribution) The particle size distribution estimation method executed by the estimation device 4 will be described with reference to the flowchart of Fig. 14. Fig. 14 is a flowchart showing the particle size distribution estimation process.

[0069] 14, first, material 9 to be measured, the particle size distribution of which is unknown, is conveyed onto conveyor 21, and material 9 flowing freely down from the end of conveyor 21 is continuously photographed by camera 24 (step S11). A plurality of photographed image data are temporarily stored in image storage unit 42 in chronological order. The installation height and conveying speed of conveyor 21 during photographing are typically the same as those during learning, and camera 24 photographs still images of the aggregate group in a dispersed state. Camera 24 also photographs the falling material 9 continuously so that each aggregate is photographed at least once.

[0070] First, similar to step S3 in FIG. 13, the contour detection unit 43 determines the shape of each aggregate for each image data by image processing using instance segmentation, and detects (extracts) the area of ​​each aggregate, i.e., aggregate area 91, according to the determined shape (step S13).

[0071] In this embodiment, for example, at this timing, the side length L10 of the minimum circumscribing square R1 of each aggregate region 91 as shown in FIG. 7(A) is calculated (step S15).

[0072] Thereafter, the contour detection unit 43 detects an approximation ellipse R2 of each aggregate region 91 as shown in Fig. 10(A) as an "abstract figure" of each aggregate (step S17), and calculates the major axis length L1 and the minor axis length L2 of the approximation ellipse R2 as the "contour feature amount" of each aggregate (step S19). In this way, the contour detection unit 43 specifies the contour feature amount of each aggregate by a calculation process using an image processing technique.

[0073] If the side length L10 of the minimum circumscribing square R1 calculated in step S15 is less than 5 mm (NO in step S21), the estimation processing unit 44 determines the grain size classification from the side length L10 of the minimum circumscribing square R1 (step S23), since it is assumed that the lengths of the long and short sides of the aggregate region 91 do not change significantly (the difference is small).

[0074] On the other hand, if the side length L10 of the minimum circumscribing square R1 of the aggregate region 91 is 5 mm or more (YES in step S21), the Mahalanobis distance from the diameter lengths L1, L2 (contour feature amount) of the approximation ellipse R2 to the Gaussian distributions G1 to G6 generated by the above-mentioned method is calculated as the similarity with each grain size division (step S25). The estimation processing unit 44 determines the first candidate, the second candidate, ... in order of the shortest distance.

[0075] Next, the estimation processing unit 44 sets a weighting factor according to the calculated Mahalanobis distance (step S27). In this embodiment, the weighting factors of the granularity classifications of multiple candidates including the first candidate are calculated by normalizing them so that the total is 1 according to the distance according to the above-mentioned classification method 1. As a result, for example, the weighting factors of the six granularity classifications (1) to (6) are calculated as [0.0], [0.1], [0.1], [0.3], [0.5], and [0.0]. According to this example, the granularity classification (5) is determined as the first candidate, the granularity classification (4) as the second candidate, and the granularity classification (2) as the third candidate, and the granularity classifications (1) and (6) are considered to be non-candidates.

[0076] The estimation processing unit 44 calculates the area of ​​the aggregate region 91 as the "aggregate area" at this timing (step S29), for example, and outputs it to the estimation processing unit 44. Note that the method of calculating the aggregate area is not limited to this example, and for example, the area of ​​an approximation ellipse R2 of the aggregate region 91 may be regarded as the aggregate area. In this case, the approximation ellipse R2 may be specified directly (without detecting the aggregate region 91) by image processing the captured image. Also, the calculation of the aggregate area may be simultaneous with the calculation of the weighting coefficient by the estimation processing unit 44, or may be earlier than that.

[0077] Based on the calculation results of the weighting coefficients in step S27 and the calculation results of the aggregate area in step S29, the estimation processing unit 44 weights the aggregate area for each aggregate region 91 by the weighting coefficient and classifies the aggregate area into multiple grain size divisions (step S30).

[0078] Specifically, for example, if the weighting coefficients of the six gradation classifications (1) to (6) calculated for a certain aggregate region 91 are [0.0], [0.1], [0.1], [0.3], [0.5], and [0.0], the area of ​​the aggregate area x 0.1 is classified (accumulated) into gradation classifications (2) and (3), the area of ​​the aggregate area x 0.3 is classified into gradation classification (4), and the area of ​​the aggregate area x 0.5 is classified into gradation classification (5).

[0079] When classification of the aggregate area for all aggregates detected as aggregate regions 91 is completed in this manner, the total area (accumulated area value) of each gradation class is obtained, and the area frequency for each gradation class can be calculated from the total area of ​​aggregates contained in material 9 (step S31). As described above, when extracting the region of each aggregate using instance segmentation, small aggregates less than 5 mm may not be detected, so it is desirable to prepare an image (background image) of the state before material 9 is allowed to flow down, extract the difference from the background image for each captured image, and calculate the sum of these differences as the total area.

[0080] The output unit 45 outputs the calculation result in step S31, that is, the particle size distribution information of the material 9 (step S33). With this, the particle size distribution estimation process is completed.

[0081] As described above, in this embodiment, the grain size distribution of each aggregate is estimated by processing the captured image of the freely falling material 9, so that the grain size distribution of the material 9 can be easily measured without performing a sieve test. Therefore, the estimation device 4 according to this embodiment makes it possible to easily perform quality control of the material 9.

[0082] Furthermore, for aggregates in which the side length L10 of the minimum circumscribing square R1 is equal to or greater than a predetermined value (e.g., 5 mm), a correction process for the grading classification is performed using a statistical method, so that the grading distribution of the material 9 can be estimated with high accuracy. For aggregates in which the side length L10 of the minimum circumscribing square R1 is less than a predetermined value (e.g., 5 mm), the statistical method is not used, and the grading classification is determined based on the side length L10 of the minimum circumscribing square R1, so that the processing load on the estimation device 4 can be reduced.

[0083] It is preferable that extraction of aggregate region 91 and calculation of contour feature amount are performed only once for one aggregate. Therefore, the transport speed of conveyor 21 and the shooting speed of camera 24 may be adjusted so that one aggregate does not appear in multiple captured images, or only aggregates appearing in a partial area of ​​each captured image may be subject to image processing. In addition, when the resolution of the captured image is low, preprocessing such as trimming or dividing the captured image may be performed before image processing.

[0084] (Other classification methods for aggregate area) In this embodiment, the aggregate area is classified into a plurality of candidates according to the above-mentioned classification method 1, but other classification methods may be adopted.

[0085] Classification method 2: The weighting coefficient for the first candidate gradation class is set to 1 (100%), and the entire aggregate area is classified into the first candidate gradation class.

[0086] According to classification method 2, it is possible to suppress an increase in error caused by classifying the aggregate area into a gradation class that is far away (for example, the fifth candidate).

[0087] It is also possible to use classification method 1 and classification method 2 in combination. For example, classification method 1 may be used as a base, and classification method 2 may be adopted only when the weighting coefficient for the first candidate calculated by classification method 1 is equal to or greater than a predetermined value (e.g., 0.6).

[0088] Alternatively, a method may be adopted in which the aggregate area is classified into only two gradation classes, the first candidate and the second candidate. In this case, the second candidate may be determined in association with the gradation class of the first candidate, as in classification method 3 shown below.

[0089] Classification method 3: The class one level above or below the first candidate is designated as the second candidate, and weighting factors are set that are normalized according to the distance from the first candidate and the second candidate, and the aggregate area is weighted according to the grain size class of the first candidate and the second candidate, respectively, for classification.

[0090] Considering the results of an experiment on the effect of differences in the direction in which aggregate is photographed on the accuracy of estimating the gradation class, one possible example is to set the class one level lower than the first candidate as the second candidate (for example, when the first candidate is equal to or higher than the specified gradation class). As another example, the class one level higher than the first candidate may be set as the second candidate (for example, when the first candidate is below the specified gradation class). Note that each of the above "specified gradation classes" can be determined by experiment.

[0091] (Modification) In this embodiment, the major axis length L1 and the minor axis length L2 of the approximate ellipse R2 of each aggregate are set as the "contour feature" and the "learning contour feature", and a two-dimensional Gaussian distribution is used, but this is not limited to the example. For example, the minor axis length L2 of the approximate ellipse R2 of each aggregate may be set as the "contour feature" and the "learning contour feature", and a one-dimensional Gaussian distribution may be used. Note that the variable of the one-dimensional Gaussian distribution may be either the major axis length L1 or the minor axis length L2.

[0092] In the present embodiment, the area (aggregate area 91), the approximate ellipse, and the contour feature amount of each aggregate are detected (calculated) by image processing technology, but the present invention is not limited to this example, and the detection (calculation) may be performed manually, for example. That is, the contour feature amount of each aggregate may be input via an operation means (not shown) or the like. Therefore, the detection means for detecting the approximate ellipse and calculating the contour feature amount may be realized manually, and the learning device 3 and the estimation device 4 may be provided with a means for specifying the contour feature amount of the approximate ellipse by calculation or input value.

[0093] In the present embodiment, the approximate ellipse of the aggregate region 91 is detected as the "abstract figure" of the aggregate. However, a figure other than the approximate ellipse may be used as long as the part corresponding to the short diameter and the part corresponding to the long diameter of the aggregate region 91 can be expressed by two straight lines (which are perpendicular to each other). For example, the minimum circumscribing rectangle surrounding the aggregate region 91 may be detected as the "abstract figure" of the aggregate. In this case, the short side length and the long side length of the minimum circumscribing rectangle may be used as the contour feature. In this way, the contour feature includes the short diameter or short side (part corresponding to the short diameter) and the long diameter or long side (part corresponding to the long diameter) of a predetermined abstract figure. Alternatively, the part corresponding to the short diameter and the part corresponding to the long diameter of the aggregate region 91 may be detected based on a predetermined rule without detecting an abstract figure from the aggregate region 91.

[0094] In addition, in this embodiment, the contour feature includes two variables, the minor axis equivalent and the major axis equivalent, but other variables may be included. In other words, the closeness of the positional relationship between the contour feature including the detected three variables and the three-dimensional Gaussian distribution may be evaluated.

[0095] In this embodiment, for aggregates with small particle size, the side length L10 of the minimum circumscribing square R1 of the aggregate region 91 is used as is to determine the particle size classification, but this is not limiting. Therefore, when the estimation device 4 estimates the particle size distribution, the process of detecting the minimum circumscribing square R1 may be omitted.

[0096] Furthermore, in the present embodiment, the learning device 3 and the estimation device 4 have been described as being separate devices, but they may be realized by a common device.

[0097] <Operation control device> Next, the operation control device 5 will be described in detail with reference to FIG. 2 and FIG.

[0098] (Functional configuration) 2(A) is a functional block diagram showing the functional configuration of the operation control device 5. The operation control device 5 includes an acquisition unit 51, an adjustment processing unit 52, a crushing control unit 53, and a control data storage unit .

[0099] The control data storage unit 54 stores a plurality of control parameters related to the crusher 12. The crushing control unit 53 controls the crusher 12 based on the control parameters (control values ​​or levels) of each item stored in the control data storage unit 54. The crushing control unit 53 also controls the material supply device 11.

[0100] 2(B) shows a specific example of the control parameters of the crusher 12. Items that can be controlled by the operation control device 5 (referred to as "target items") include control parameters such as the gap value in the crushing space of the crusher 12, the rotation speed of the mantle (rotating shaft), and the amount of aggregate supplied to the crusher 12 (amount of crushed material supplied) adjusted by the feed speed of the material supply feeder. These control parameters are stored in the control data storage unit 54. The control data storage unit 54 may also store other measured values ​​such as the current value of the motor generated in the crusher 12 and the crushing pressure of the crusher 12 as needed. It is to be noted that the control parameters of the target items desirably include at least the gap value.

[0101] The acquisition unit 51 acquires an output value (particle size distribution) from the above-mentioned estimation device 4 and outputs it to the adjustment processing unit 52. When the particle size distribution acquired by the acquisition unit 51 is outside the target range (outside the allowable range of the target particle size distribution), the adjustment processing unit 52 adjusts at least one of the control parameters stored in the control data storage unit 54. That is, the adjustment processing unit 52 adjusts at least one of the gap value, the rotation speed, and the aggregate supply amount. The target particle size distribution is set, for example, by a user (operator). The set value of the target particle size distribution may also be stored in the control data storage unit 54.

[0102] In the following description, for ease of understanding, an example will be described in which the adjustment processing unit 52 adjusts the gap value. That is, when the particle size distribution is outside the target range (outside the allowable range of the target particle size distribution), the adjustment processing unit 52 calculates an adjustment amount for the current gap value according to a predetermined rule. The gap value (control parameter) in the control data storage unit 54 is updated according to the calculation result of the adjustment amount.

[0103] The acquisition unit 51 is realized by, for example, an input / output interface or a communication interface. The adjustment processing unit 52 and the crushing control unit 53 are realized by a processor executing software. The control data storage unit 54 is typically realized by a non-volatile storage device.

[0104] (About operation) 3 is a flowchart showing the operation of the operation control device 5. The operation control device 5 sets a target particle size distribution (target value of particle size distribution) based on an instruction from a user (step S102). The target particle size distribution is set, for example, as "the cumulative passing rate of 20 mm or less is 60%" (see FIG. 12(B) described later). Alternatively, it is set as "the passing rate of 13 mm or more and 20 mm or less is 60%". The set target particle size distribution is stored in the control data storage unit 54.

[0105] Also, an initial value of the control parameter of the target item is set (step S104). The initial value of the control parameter of the target item may be a predetermined default value or a value input by the user.

[0106] The crushing control unit 53 controls the material supplying device 11 and the crusher 12 based on the control parameters of each item stored in the control data storage unit 54, and starts the operation of the material supplying device 11 and the crusher 12 (step S106). At this time, the operation control device 5 outputs an operation command to the estimation device 4. As a result, the estimation device 4 calculates (estimates) the particle size distribution of the aggregate based on the captured image of the aggregate dropping from the conveyor 21 downstream of the crusher 12 while the crusher 12 is operating.

[0107] When the acquisition unit 51 acquires the estimation result by the estimation device 4, that is, the estimated value of the particle size distribution (step S108), the adjustment processing unit 52 compares the acquired estimated value with the target value set in step S102. Specifically, it is determined whether the particle size distribution estimated by the estimation device 4 is within a target range based on the set target particle size distribution. If the acquired estimated value is within the target range (YES in step S110), the process returns to step S108. That is, the control of the crusher 12 continues with the current parameters.

[0108] On the other hand, if the acquired estimated value is outside the target range (NO in step S110), the control parameter of the target item is adjusted, for example, by a predetermined amount (level), and the control parameter stored in the control data storage unit 54 is updated (step S112). Specifically, when the estimated particle size distribution is finer than the target particle size distribution (for example, the estimated particle size distribution is "the cumulative passing rate of 20 mm or less is 80%"), the adjustment amount may be set to, for example, "+1 level" so that the gap value becomes larger. Conversely, when the estimated particle size distribution is coarser than the target particle size distribution (for example, the estimated particle size distribution is "the cumulative passing rate of 20 mm or less is 40%), the adjustment amount may be set to, for example, "-1 level" so that the gap value becomes smaller. When this process is completed, the process returns to step S108. As a result, the crushing control unit 53 controls the crusher 12 with the adjusted parameters.

[0109] In practice, not only the gap value but also the rotation speed and / or the aggregate supply amount may be adjusted at the same time. Alternatively, the gap value may be maintained while both or either of the rotation speed and / or the aggregate supply amount is adjusted (each by a predetermined amount).

[0110] In addition, for stable operation of the crusher 12, it is desirable to adjust the control parameters such as the gap value while monitoring the measured values ​​of other control items (current value, crushing pressure). Therefore, the adjustment process of the control parameters may be performed manually as appropriate. That is, in this embodiment, the user may input the adjustment amount.

[0111] As described above, the crusher control system SYS1 according to this embodiment includes the estimation device 4 described with reference to Figs. 6 to 14, and the estimation device 4 accurately estimates the particle size distribution of the crushed material (material 9). Therefore, the operation control device 5 controls the operation of the crusher 12 by comparing the accurately estimated particle size distribution of the aggregate with the target particle size distribution, so that the particle size distribution of the crushed material can be appropriately brought close to the target particle size distribution. As a result, the operation efficiency of the crusher 12 is improved, and the quality of the crushed material is stabilized.

[0112] Furthermore, by providing the present system SYS1 with the estimation device 4, it is possible to improve the production speed and reduce the facility space and facility costs.

[0113] (Specific processing of the adjustment processing unit) In steps S110 and S112 in FIG. 3, the adjustment processing unit 52 determines whether the estimation result (particle size distribution of the material obtained by image processing) obtained from the estimation device 4 is within the target range (within the allowable range of the target particle size distribution) or not, and adjusts the control parameters (for example, the gap value) if it is outside the target range. Here, when the target particle size distribution is specified by the target frequency (or the target cumulative passing rate) in a plurality of particle size divisions, a case in which a particle size division lower (coarse) than the target frequency and a particle size division higher (fine) than the target frequency are mixed can be assumed. For example, as an extreme example, an example of the estimation result at each point (boundary particle size of each division) of 13 mm, 20 mm, 25 mm, and 30 mm shown in FIG. 12(B) is shown in FIG. 15(A).

[0114] In Fig. 15(A), the estimated cumulative passing ratios of each point (boundary particle size of each division) based on the estimation results by the estimation device 4 are plotted on a graph, and a graph (shown by a solid line) in which the plotted points are connected by straight lines is shown, together with a reference line (shown by a dashed line) passing through the target cumulative passing ratios of each point (boundary particle size of each division) and a boundary line (shown by a broken line) indicating the allowable range (upper and lower limits) of the same. An example of the judgment results for each point based on the estimation results in Fig. 15(A) is shown in Fig. 15(B).

[0115] In this example, only the estimated cumulative passing ratio at the fourth point (x4=30 mm) is within the target range, while the estimated cumulative passing ratios at the first point (x1=13 mm), second point (x2=20 mm), and third point (x3=25 mm) are outside the target range. The estimated cumulative passing ratios at the first and third points are smaller (coarse) than the target range, and the estimated cumulative passing ratio at the second point is larger (fine) than the target range.

[0116] The processing method (method of determining the estimation result and method of adjusting the control parameters) of the adjustment processing unit 52, which takes into account the case where there is a mixture of categories in which the estimated frequency or estimated cumulative passing rate is below the target range and categories in which it exceeds the target range, is described below as a modified example of embodiment 1.

[0117] (Method of determining the estimated results) Fig. 16 is a functional block diagram showing the functional configuration of the operation control device 5 in a modified example of this embodiment. As shown in Fig. 16, the adjustment processing unit 52 in this modified example includes, as its functional configuration, an individual judgment unit 521 and a comparison unit 522. The individual judgment unit 521 individually judges whether or not the estimated cumulative passing ratio for each of a plurality of (for example, four) particle size classes is within a target range of a target value according to the target particle size distribution. The estimated cumulative passing ratios for the plurality of particle size classes are obtained from the acquisition unit 51. When the estimated cumulative passing ratio is outside the target range, the individual judgment unit 521 also judges whether or not it is greater than or smaller than the target range.

[0118] The comparison unit 522 converts the judgment result by the individual judgment unit 521 into a score with a positive or negative numerical value (integer), and compares the total score with a preset upper threshold and a lower threshold. As a simple example of converting the judgment result by the individual judgment unit 521 into a score, it is possible to set the score to "+1" when the estimated cumulative passing ratio is below the target range (rough), the score to "-1" when the estimated cumulative passing ratio exceeds the target range (fine), and the score to "0" when the estimated cumulative passing ratio is within the target range (good). A specific example of the judgment table in this case is shown in FIG. 17(A). As shown in FIG. 16, the judgment table is stored in a judgment table storage unit 58 provided in the operation control device 5. The judgment table storage unit 58 is typically realized by a non-volatile storage means.

[0119] 17(A), the judgment table is a correspondence table that associates the number of points for each granularity class, which indicates the judgment result by the individual judgment unit 521, with the total value of the number of points. The judgment table makes it possible to easily specify the total points corresponding to the combination of judgment results for each granularity class. In this embodiment, the total points can take one of nine values: +4, +3, +2, +1, 0, -1, -2, -3, and -4.

[0120] 3, the comparison unit 522 compares the total score of the four points with the upper and lower thresholds set in advance. If the total score of the four points exceeds the upper or lower threshold, the adjustment processing unit 52 adjusts the control parameters (step S112). Here, it is assumed that the upper threshold is set to "+1" and the lower threshold is set to "-1".

[0121] With reference to FIG. 17(A), for example, a pattern in which points 1 to 4 are "0, +1, +1, 0" respectively has a total point of "+2" and therefore exceeds the upper threshold. In this case, the adjustment processing unit 52 determines that the overall particle size distribution of the target material is "coarser" than the target particle size distribution. A pattern in which points 1 to 4 are "0, +1, -1, -1" respectively has a total point of "-1" and therefore does not exceed the upper and lower thresholds. In this case, the adjustment processing unit 52 determines that the overall particle size distribution of the target material is within the target range (i.e., within the allowable range of the target particle size distribution).

[0122] In FIG. 17(A), for ease of understanding, the cells where the total points exceed the upper threshold (corresponding to a “coarse” pattern outside the target range) are shown in dark gray, and the cells where the total points exceed the lower threshold (corresponding to a “fine” pattern outside the target range) are shown in light gray.

[0123] (How to adjust control parameters) As shown in Fig. 17(B), when the total points exceed the upper threshold (in the above example, "+2 or more"), the adjustment processing unit 52 issues a command to, for example, set the gap value setting level of the crusher 12 to "-1" and decrease (narrow) the gap value. When the total points are less than the lower threshold (in the above example, "-2 or less"), the adjustment processing unit 52 issues a command to, for example, set the gap value setting level of the crusher 12 to "+1" and increase (widen) the gap value. In other cases (in the above example, "-1 to +1"), the current setting level is maintained.

[0124] It was confirmed by tests that the particle size distribution of the target material can be brought within the allowable range of the target particle size distribution by executing the above-mentioned determination method and adjustment method in the adjustment processing unit 52. Figures 18 and 19 show schematic results of Test 1 and Test 2, respectively.

[0125] In the test 1, as shown in Fig. 18(A), the estimation results at all points were below the target range (coarse), and as shown in Fig. 18(B), the judgment result (total points) of the crushed object to be tested was "+4" (above the upper threshold). Therefore, the adjustment processing unit 52 controlled the setting level of the gap value of the crusher 12 to "-1".

[0126] Test 2 was conducted immediately after Test 1 using the same material as in Test 1 as a sample. In Test 2, as shown in FIG. 19(A), the estimated results of the three points other than the first point were within the target range, and as shown in FIG. 19(B), the judgment result (total points) of the crushed material to be tested was "-1" (less than the upper threshold and more than the lower threshold). In this way, it was confirmed that the particle size distribution of the crushed material was within the allowable range of the target particle size distribution by the adjustment processing unit 52 reducing the gap value of the crusher 12 in response to the results of Test 1.

[0127] The allowable range (from the reference value) for each grain size division can be set arbitrarily. As shown in FIG. 15(B) and the like, for example, if the allowable ranges for each of the four points are "±α1", "±α2", "±α3", and "±α4", the values ​​of α1 to α4 may be constant (the same value) or may be different. In the latter case, at least one of α1 to α4 may be "0" (there may be no range). Furthermore, the user may be able to set these values ​​arbitrarily.

[0128] In the present embodiment, the individual judgment unit 521 sets the judgment value to "+1" when the estimated value (estimated frequency or estimated cumulative passing rate) is smaller (coarse) than the target range in any granularity division, and sets the judgment value to "-1" when the estimated value is larger (fine) than the target range, but the judgment value may be different for each granularity division. For example, the following settings may be used. · First point (grain size 13 mm) · "Coarse: +1, Fine: -2" · Second point (grain size 20 mm) · "Coarse: +1, fine: -1" · Third point (grain size 25 mm) · "Coarse: +1, fine: -1" · Fourth point (grain size 30 mm) · "Coarse: +2, fine: -1" As shown in Fig. 16, it is preferable that the operation control device 5 in the modified example of this embodiment further includes a display unit 56 that displays the judgment table stored in a judgment table storage unit 58, and an operation unit 57 that allows the user to set and change the number of points for each granularity classification shown in the judgment table. In addition, the user may be allowed to arbitrarily set an upper limit threshold and a lower limit threshold for the total points via the operation unit 57. From the viewpoint of operability, it is preferable that the display unit 56 is realized by a touch panel that also serves as the operation unit 57.

[0129] The determination method by individual determination unit 521 can also be adopted in the following second embodiment.

[0130] [Embodiment 2] It is also possible to operate the crusher 12 independently by accumulating data such as particle size distribution data acquired in time series by the operation control device 5 of the above-mentioned embodiment 1 and adjustment amounts calculated, and by machine learning. In this embodiment, a crusher control system SYS2 in this case will be described.

[0131] With reference to FIG. 1, a crusher control system SYS2 according to this embodiment includes an estimation device 4, a learning device 6, and an operation control device 5A.

[0132] <Learning device> The functional configuration of the learning device 6 will be described with reference to Fig. 4. The learning device 6 is configured by an information processing device (such as a general-purpose computer) equipped with a processor such as a CPU and a non-volatile memory. Fig. 4 is a functional block diagram showing the functional configuration of the learning device 6 and the operation control device 5A.

[0133] The learning device 6 includes a data acquisition unit (not shown) that directly or indirectly acquires multiple types of data from the operation control device 5 of the first embodiment, a data storage unit 61 that accumulates the multiple types of data acquired by the data acquisition unit, a learning unit 62 that performs machine learning on the multiple types of data accumulated in the data storage unit 61 to generate a learning model, and a model storage unit 63 that stores the learning model generated by the learning unit 62. The data acquisition unit is realized by, for example, an input / output interface or a communication interface.

[0134] The data accumulation unit 61 accumulates the control parameters stored and updated in the control data storage unit 54 of the operation control device 5 and the particle size distribution data sequentially acquired by the acquisition unit 51 (i.e., estimated by the estimation device 4). In addition, adjustment amount data calculated by the adjustment processing unit 52 is accumulated in association with the particle size distribution data before and after it.

[0135] The learning unit 62 performs machine learning on the data stored in the data storage unit 61, inputs the current control parameters and the current particle size distribution estimated at that time, and generates a learning model that outputs an adjustment amount from the current control parameters by specifying a target particle size distribution. As the machine learning algorithm used by the learning unit 62, for example, a stochastic gradient subduction method can be adopted.

[0136] 4(B), the learning unit 62 may learn by machine learning the correlation between (1) "the original control parameters of each target item" and (2) "the original particle size distribution estimated with the original control parameters" and the correlation between (3) "the item name of the control parameter adjusted by the adjustment processing unit 52 and the adjustment amount thereof" and (4) "the amount of change from the original particle size distribution due to the adjustment." In other words, the teacher data for machine learning may include the variable data of (1) to (4).

[0137] The variable data (1) to (3) are data that are sequentially acquired by performing the process of step S104 in Fig. 3 described above, and then repeatedly performing the process routine of steps S108 to S112. The variable data (4) are data that the learning unit 62 calculates from the "original particle size distribution" (2) (a value acquired in step S108 in a certain process routine) and a particle size distribution estimated thereafter (a value acquired in step S108 in a subsequent process routine). Note that the "particle size distribution estimated thereafter" may correspond to the variable data (2) (original particle size distribution) in other teacher data (i.e., teacher data that is later in time).

[0138] The function of the learning unit 62 is realized by a processor executing software. The data accumulation unit 61 and the model storage unit 63 are realized by a non-volatile memory (storage device). As a modified example (not shown), the operation control device 5 of the first embodiment may be equipped with the function of the learning device 6.

[0139] <Operation control device> The functional configuration of the operation control device 5A in this embodiment will be described with reference to FIG.

[0140] The operation control device 5A differs from the operation control device 5 of the first embodiment in that it includes a model storage unit 55 and an adjustment processing unit 52A instead of the above-mentioned adjustment processing unit 52. The operation control device 5A includes an acquisition unit 51, an adjustment processing unit 52A, a crushing control unit 53, a control data storage unit 54, and a model storage unit 55.

[0141] The model storage unit 55 stores the learning model generated by the learning device 6. The model storage unit 55 may be realized by a non-volatile storage device, or may be realized by a cloud server accessible from the operation control device 5A.

[0142] The adjustment processing unit 52A acquires the particle size distribution output from the acquisition unit 51, and also acquires the control parameters of each target item stored and updated in the control data storage unit 54. Measurement values ​​other than the target items may also be acquired. Then, at regular intervals (for example, one minute), a target particle size distribution set in advance is input to the learning model stored in the model storage unit 55, and (1) the current control parameters and (2) the current particle size distribution are input. The adjustment processing unit 52A calculates the output from the learning model as an adjustment amount from the current control parameters. Specifically, it is desirable to output the adjustment amount in association with item identification data that identifies the item to be adjusted.

[0143] The adjustment amount (and item identification data) calculated by the adjustment processing unit 52A is output to the crushing control unit 53 and the control data storage unit 54. This allows the crushing control unit 53 to control the material supply device 11 or the crusher 12 with the control parameters that take into account the adjustment amount calculated by the learning model. The control data storage unit 54 may store, for example, the initial value and adjustment amount of the control parameter for each target item.

[0144] The operation of the operation control device 5A in this embodiment will be described with reference to Fig. 5. In the flowchart of Fig. 5, the same processes as those shown in the flowchart of Fig. 3 are given the same step numbers, so detailed description of these processes will not be repeated.

[0145] In this embodiment, after starting the operation of the crusher 12, the operation control device 5A compares the target value set in step S102 with the acquired estimated value. If the acquired estimated value is within the target range (YES in step S110), the process returns to step S108. On the other hand, if the acquired estimated value is outside the target range (NO in step S110), the adjustment processing unit 52A inputs the estimated value of the particle size distribution and the current control parameters to the learning model (step S120). As a result, the learning model outputs an adjustment amount in association with the item identification data (step S122). That is, the adjustment amount of the control parameters to achieve the target particle size distribution is will be output.

[0146] After calculating the adjustment amount, the adjustment processing unit 52A updates the control parameters in the control data storage unit 54 (step S124). As a result, the crushing control unit 53 controls the crusher 12 with the adjusted control parameters.

[0147] According to this embodiment, the adjustment amount of the control parameter is calculated using a learning model previously learned in the learning device 6, so that the crusher 12 can be autonomously operated. Therefore, the control process of the crusher 12 can be omitted, and the effects of improving the production speed and reducing labor costs can be expected.

[0148] Each of the estimation method executed by the estimation device 4, the operation control method executed by the operation control devices 5 and 5A, and the learning method executed by the learning device 6 can be provided as a program. Such a program can be provided by being recorded on an optical medium such as a CD-ROM (Compact Disc-ROM) or a computer-readable non-transitory recording medium such as a memory card. Also, the program can be provided by downloading via a network.

[0149] The program according to the present invention may be one that executes processing by calling necessary modules in a predetermined sequence at a predetermined timing among program modules provided as part of a computer's operating system (OS). In this case, the program itself does not include the above modules, and executes processing in cooperation with the OS. Programs that do not include such modules may also be included in the program according to the present invention.

[0150] The program according to the present invention may be provided by being incorporated into a part of another program. In this case, the program itself does not include the modules included in the other program, and executes processing in cooperation with the other program. Such a program incorporated into another program may also be included in the program according to the present invention.

[0151] The embodiments disclosed herein should be considered to be illustrative and not restrictive in all respects. The scope of the present invention is defined by the claims, not the above description, and is intended to include all modifications within the meaning and scope of the claims. [Explanation of symbols]

[0152] SYS1, SYS2 crusher control system, 4 estimation device, 5, 5A operation control device, 6 learning device, 11 material supply device, 12 crusher, 51 acquisition unit, 52, 52A adjustment processing unit, 53 crushing control unit, 54 control data memory unit, 55, 63 model memory unit, 56 display unit, 57 operation unit, 58 judgment table memory unit, 61 data accumulation unit, 62 learning unit.

Claims

1. A particle size distribution estimation device for estimating the particle size distribution of crushed material by a crusher; and an operation control device that controls the operation of the crusher based on a comparison result between the particle size distribution estimated by the particle size distribution estimation device and a target particle size distribution set in advance, The particle size distribution estimation device includes: a detection means for detecting a part corresponding to a minor axis and a part corresponding to a major axis of each crushed object from a photographed image of the crushed object falling freely, and calculating a contour feature amount including a length of the part corresponding to the minor axis and a length of the part corresponding to the major axis; a determining means for determining a first candidate particle size class for each crushed object based on a distribution of learning contour feature values ​​for each particle size class determined in advance based on a learning image and a distance between the contour feature value identified by the detecting means; A calculation means for calculating an area of ​​each crushed object as an index value representing the amount of each crushed object from the captured image; A crusher control system comprising: an area accumulating means for calculating a total area for each crushed material by classifying all or a portion of the crushed material area calculated by the calculation means into the first candidate particle size category.

2. The operation control device includes an adjustment processing means for adjusting control parameters including at least one of a gap value of the crushing space in the crusher, a rotation speed of the rotating shaft of the crusher, and an amount of crushed material supplied to the crusher when the particle size distribution estimated by the particle size distribution estimation device is outside a target range. The crusher control system according to claim 1.

3. The present invention further includes a storage means for storing a learning model that is generated so as to output an adjustment amount from the current control parameters by inputting the current control parameters and the current particle size distribution estimated at that time and specifying a target particle size distribution, The crusher control system according to claim 2 , wherein the adjustment processing means calculates an adjustment amount of the control parameter by using the learning model.

4. The crusher control system according to claim 3 , further comprising a learning device that generates the learning model.

5. An operation control device for controlling the operation of a crusher, An acquisition means for acquiring the particle size distribution of the crushed material by the crusher, the particle size distribution being estimated by the particle size distribution estimation device; and an adjustment processing means for adjusting control parameters including at least one of a gap value of a crushing space in the crusher, a rotation speed of a rotating shaft of the crusher, and an amount of crushed material supplied to the crusher, based on a comparison result between the particle size distribution acquired by the acquisition means and a target particle size distribution set in advance; The particle size distribution acquired by the acquiring means is a value obtained by calculating a total area for each particle size division by classifying all or a part of the crushed object area calculated from the photographed image into a first candidate particle size division determined based on the closeness of the distance between the distribution of learning contour feature values ​​for each particle size division previously obtained based on the learning image and the contour feature values ​​of each crushed object calculated from the photographed image of the crushed object, The contour feature amount includes the length of a portion corresponding to a minor axis and the length of a portion corresponding to a major axis detected from a photographed image of the crushed object falling freely.

6. The acquiring means acquires an estimated frequency or an estimated cumulative passing ratio of a plurality of particle size classes from the particle size distribution estimated by the particle size distribution estimating device, The crusher operation control device according to claim 5, wherein the adjustment processing means includes individual judgment means for judging whether an estimated frequency or an estimated cumulative passing rate for each of the plurality of particle size classifications is greater than a target range corresponding to the target particle size distribution or is smaller than the target range.

7. The crusher operation control device according to claim 6, wherein the adjustment processing means includes a comparison means for converting the judgment results by the individual judgment means into points using positive and negative numerical values, and comparing the total value of the points with an upper threshold value and a lower threshold value.

8. 8. The crusher operation control device according to claim 7, further comprising a storage means for storing a judgment table in which the points for each particle size range representing the judgment result by the individual judgment means correspond to the total value of the points.

9. A display means for displaying the judgment table; The crusher operation control device according to claim 8 , further comprising an operation means for a user to set or change the number of points for each particle size category represented in the judgment table.

10. An operation control method for controlling an operation of a crusher, comprising: A step of acquiring a particle size distribution of crushed material by the crusher, the particle size distribution being estimated by a particle size distribution estimation device; and adjusting control parameters including at least one of a gap value of a crushing space in the crusher, a rotation speed of a rotating shaft of the crusher, and an amount of crushed material supplied to the crusher based on a comparison result between the acquired particle size distribution and a target particle size distribution set in advance; The particle size distribution obtained in the obtaining step is a value obtained by calculating a total area for each particle size division by classifying all or a part of the crushed object area calculated from the photographed image into a first candidate particle size division determined based on the closeness of the distance between the distribution of learning contour feature values ​​for each particle size division previously obtained based on the learning image and the contour feature values ​​of each crushed object calculated from the photographed image of the crushed object, The method for controlling operation of a crusher, wherein the contour feature amount includes the length of a portion corresponding to a minor axis and the length of a portion corresponding to a major axis detected from a photographed image of the crushed object falling freely.

Images