Method for predicting the quality of ready-mixed concrete, method for manufacturing ready-mixed concrete, and system for manufacturing ready-mixed concrete

Abstract

To improve prediction accuracy of quality of ready-mixed concrete.SOLUTION: A ready-mixed concrete quality prediction method according to one aspect of the present disclosure comprises: an acquisition step of acquiring input information including vibration information indicating a magnitude of vibration due to dropping or flowing-down of ready-mixed concrete manufactured by a mixer and image information of the ready-mixed concrete; a prediction step of predicting quality of the ready-mixed concrete on the basis of a prediction model that is previously constructed by machine learning so as to output quality information indicating the quality of the ready-mixed concrete according to the input of the input information and the input information acquired in the acquisition step; and a construction step of constructing the prediction model. The construction step comprises: acquiring input information for test including the vibration information for test and image information of ready-mixed concrete for test; and constructing the prediction model on the basis of the input information for test and a measurement value of the quality of the ready-mixed concrete for test.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to a quality prediction method for ready-mixed concrete, a method for producing ready-mixed concrete, and a production system for ready-mixed concrete. [Background Art]

[0002] Patent Document 1 discloses a quality prediction method that predicts the quality of ready-mixed concrete using a prediction model. [Prior Art Literature] [Patent Literature]

[0003] [Patent Document 1] Japanese Unexamined Patent Application Publication No. 2020-144099 [Summary of the Invention] [Problem to be Solved by the Invention]

[0004] The present disclosure provides a quality prediction method for ready-mixed concrete, a method for producing ready-mixed concrete, and a production system for ready-mixed concrete that can improve the accuracy of quality prediction. [Means for Solving the Problem]

[0005] [1] A method for predicting the quality of ready-mixed concrete relating to one aspect of the present disclosure includes: an acquisition step of acquiring input information including vibration information indicating the magnitude of vibrations caused by the dropping or flow of ready-mixed concrete produced by a mixer for mixing concrete materials, and image information of ready-mixed concrete; a prediction model pre-built by machine learning to output quality information indicating the quality of ready-mixed concrete in response to the input of input information; a prediction step of predicting the quality of ready-mixed concrete based on the input information acquired in the acquisition step; and a construction step of building a prediction model before executing the acquisition step and the prediction step. The construction step includes acquiring test input information including test vibration information indicating the magnitude of vibrations caused by the dropping or flow of test ready-mixed concrete produced by a mixer, and image information of test ready-mixed concrete; and building a prediction model based on the test input information and measured values ​​of the quality of test ready-mixed concrete associated with the test input information.

[0006] [2] In the quality prediction method described in [1] above, the acquisition step may include obtaining vibration information from the detection results of a sensor located below the mixer and attached to a hopper capable of containing ready-mixed concrete. The vibration information may also be information indicating the magnitude of vibration of the hopper when ready-mixed concrete is discharged to the hopper.

[0007] [3] In the quality prediction method described in [2] above, the acquisition step may include obtaining image information of the ready-mixed concrete by imaging the ready-mixed concrete in the hopper.

[0008] [4] In the quality prediction method described in [2] or [3] above, the acquisition step may include acquiring a reference time corresponding to the time when the ready-mixed concrete is discharged from the hopper, and acquiring time-series information as vibration information indicating the magnitude of vibration of the hopper during the period from a time before the reference time to a time after the reference time.

[0009] [5] In the quality prediction method described in any of [2] to [4] above, the vibration information may include information indicating the magnitude of hopper vibration in a direction intersecting the side wall of the hopper, and information indicating the magnitude of hopper vibration in a direction along the side wall of the hopper.

[0010] [6] In the quality prediction method described in any of [2] to [5] above, the acquisition step and the prediction step may be performed while batch processing, which includes mixing in a mixer, discharge from the mixer to a hopper, and loading from the hopper to a transport vehicle, is being repeated. In the acquisition step, input information may be acquired for each batch processing. In the prediction step, the quality of the ready-mixed concrete may be predicted for each batch processing.

[0011] [7] In the quality prediction method described in any of [1] to [6] above, both the slump and slump flow of the ready-mixed concrete may be predicted in the prediction step.

[0012] [8] In the quality prediction method described in any of [1] to [7] above, the prediction step may predict the amount of air in the fresh concrete.

[0013] [9] A method for producing ready-mixed concrete relating to one aspect of the present disclosure includes: a mixing step of mixing concrete materials in a mixer; an acquisition step of acquiring input information including vibration information indicating the magnitude of vibrations caused by the dropping or flow of ready-mixed concrete produced in the mixing step and image information of ready-mixed concrete; a prediction model pre-built by machine learning to output quality information indicating the quality of ready-mixed concrete in response to the input of input information; a prediction step of predicting the quality of ready-mixed concrete based on the input information acquired in the acquisition step; and a construction step of building a prediction model before executing the acquisition step and the prediction step. The construction step includes acquiring test input information including test vibration information indicating the magnitude of vibrations caused by the dropping or flow of test ready-mixed concrete produced by the mixer and image information of test ready-mixed concrete; and building a prediction model based on the test input information and measured values ​​of the quality of test ready-mixed concrete associated with the test input information.

[0014]

[10] A ready-mix concrete manufacturing system relating to one aspect of the present disclosure includes: a mixer for mixing concrete materials; a vibration sensor for detecting the magnitude of vibrations caused by the dropping or flow of ready-mix concrete produced in the mixer; an image sensor for imaging the ready-mix concrete; an information acquisition unit for acquiring input information including vibration information indicating the magnitude of vibrations caused by the dropping or flow of ready-mix concrete and image information of ready-mix concrete from the detection results by the vibration sensor and the imaging results by the image sensor; a prediction model pre-built by machine learning to output quality information indicating the quality of ready-mix concrete in response to the input of input information; a quality prediction unit for predicting the quality of ready-mix concrete based on the input information acquired by the information acquisition unit; and a model construction unit for constructing the prediction model. The model construction unit performs the following: acquire test input information including test vibration information indicating the magnitude of vibrations caused by the dropping or flow of test ready-mix concrete produced by the mixer and image information of test ready-mix concrete; and construct a prediction model based on the test input information and measured values ​​of the quality of test ready-mix concrete associated with the test input information. [Effects of the Invention]

[0015] This disclosure provides a method for predicting the quality of ready-mixed concrete, a method for manufacturing ready-mixed concrete, and a system for manufacturing ready-mixed concrete, all of which are capable of improving the accuracy of quality prediction. [Brief explanation of the drawing]

[0016] [Figure 1] Figure 1 is a schematic diagram showing an example of a ready-mix concrete manufacturing system. [Figure 2] Figure 2(a) is a schematic side view showing an example of a loading hopper. Figure 2(b) is a schematic bottom view showing an example of a loading hopper. [Figure 3] Figure 3 is a block diagram showing an example of the functional configuration of a control device. [Figure 4]Fig. 4(a) is a graph showing an example of detection values obtained by a vibration sensor. Fig. 4(b) is a graph showing an example of detection values when ready-mixed concrete is dropped. [Figure 5] Fig. 5 is a block diagram showing an example of a hardware configuration of a control device. [Figure 6] Fig. 6 is a flowchart showing an example of a series of processes executed in a learning phase. [Figure 7] Fig. 7 is a flowchart showing an example of a series of processes executed in an evaluation phase. DETAILED DESCRIPTION OF EMBODIMENTS

[0017] An embodiment will be described below with reference to the drawings. In the description, the same reference numerals are assigned to the same elements or elements having the same functions, and duplicate descriptions are omitted.

[0018] [Ready-mixed concrete manufacturing system] Fig. 1 schematically shows a ready-mixed concrete manufacturing system according to an embodiment. The manufacturing system 1 shown in Fig. 1 is a system for manufacturing ready-mixed concrete. The manufacturing system 1 manufactures ready-mixed concrete by kneading concrete materials. The concrete materials include cement, admixtures, coarse aggregate (for example, gravel), fine aggregate (for example, sand), water, admixtures, and the like.

[0019] The manufacturing system 1 loads the manufactured ready-mixed concrete onto a transport vehicle C. After the ready-mixed concrete is loaded, the transport vehicle C transports the ready-mixed concrete to a site where the ready-mixed concrete is used (for example, a construction site). Examples of the transport vehicle C include agitator trucks (mixer trucks) and dump trucks. The manufacturing system 1 may manufacture ready-mixed concrete from concrete materials so as to satisfy the target quality (required quality) set for each site. The target quality may include at least one target value of slump, slump flow, and air content. The manufacturing system 1 includes a manufacturing apparatus 10, a vibration sensor 40, an image sensor 50, and a control device 60.

[0020] (Apparatus for manufacturing ready-mixed concrete) The manufacturing apparatus 10 is an apparatus that manufactures ready-mixed concrete based on operation instructions from a control device 60. The manufacturing apparatus 10 may manufacture ready-mixed concrete having a slump of 21 cm or less, may manufacture ready-mixed concrete having a slump of 23 cm or less, and may manufacture ready-mixed concrete having a slump of 25 cm or less. When manufacturing ready-mixed concrete having a slump of 21 cm or less, 23 cm or less, or 25 cm or less, the target quality of the ready-mixed concrete may be set by both the target slump value and the target slump flow value. The manufacturing apparatus 10 may manufacture ready-mixed concrete having a slump greater than 21 cm. When manufacturing ready-mixed concrete having a slump greater than 21 cm, the target quality of the ready-mixed concrete may be set by a target slump flow value instead of slump.

[0021] The manufacturing apparatus 10 includes, for example, a storage bin 12, a weighing bin 14, a collecting hopper 16, a mixer 20, and a loading hopper 30 (hopper). In Fig. 1, the vertical direction is indicated by "direction D1" and "direction D2". Direction D1 is vertically upward, and direction D2 is vertically downward. In the manufacturing apparatus 10, the storage bin 12, the weighing bin 14, the collecting hopper 16, the mixer 20, and the loading hopper 30 are arranged in this order from top to bottom. Further, the manufacturing apparatus 10 may not include the collecting hopper 16, and in the manufacturing apparatus 10, the storage bin 12, the weighing bin 14, the mixer 20, and the loading hopper 30 may be arranged in this order from top to bottom.

[0022] The storage bin 12 temporarily stores various concrete materials. Various concrete materials are conveyed into the storage bin 12 from an aggregate yard, a cement silo, a water tank, or the like by a conveying device such as a belt conveyor. The storage bin 12 is configured to store various concrete materials individually. Hereinafter, "concrete material" may be simply referred to as "material". Various materials stored in the storage bin 12 are supplied to the weighing bin 14 as needed.

[0023] The measuring bottle 14 is located below the storage bottle 12. The measuring bottle 14 operates based on operation instructions from the control device 60 and weighs various materials individually. When the measuring bottle 14 detects the target amount of material instructed by the control device 60, it supplies that material to the collection hopper 16. When water is supplied to the measuring bottle 14, an admixture may be mixed into the water. The collection hopper 16 is located below the measuring bottle 14. The collection hopper 16 collects the various materials discharged from the measuring bottle 14 and supplies the collected materials to the mixer 20. If the collection hopper 16 is not provided, the various materials are supplied to the mixer 20 from the measuring bottle 14.

[0024] The mixer 20 is located below the collective hopper 16. The mixer 20 is a device for mixing concrete materials. The mixer 20 produces ready-mix concrete by mixing aggregate, cement, water, and admixtures. The mixer 20 may be a tilting mixer, a horizontal single-axis mixer, a horizontal double-axis mixer, or a pan-type mixer. The mixer 20 includes, for example, two stirring members 21 and a mixer drive unit 22.

[0025] The two agitators 21 are components that agitate the various materials supplied to the mixer 20. The two agitators 21 are arranged side by side inside the main body (container) of the mixer 20 and are rotatably mounted. Each of the two agitators 21 includes a rotation axis that extends in one horizontal direction. The mixer drive unit 22 rotates the rotation axes of each of the two agitators 21 based on operation instructions from the control device 60. The mixer drive unit 22 includes a drive source, such as a motor, that provides driving force to the agitators 21. An opening is provided at the bottom of the main body of the mixer 20 for discharging the manufactured ready-mixed concrete into the loading hopper 30.

[0026] The loading hopper 30 is located below the mixer 20 and is capable of accommodating ready-mixed concrete. The loading hopper 30 receives the ready-mixed concrete discharged from the mixer 20 (the opening of the mixer 20) and temporarily stores the concrete. When the ready-mixed concrete is discharged from the mixer 20 to the loading hopper 30, it falls into the loading hopper 30. For example, when discharged from the mixer 20, almost all of the ready-mixed concrete produced in the mixer 20 falls into the loading hopper 30. After temporarily storing the ready-mixed concrete produced in the mixer 20, the loading hopper 30 supplies the ready-mixed concrete to the transport vehicle C.

[0027] The loading hopper 30 may have any shape as long as it can accommodate ready-mixed concrete. The loading hopper 30 forms a space for accommodating ready-mixed concrete by, for example, a bottom and a side wall that extends upward from the outer edge of the bottom and has an open upper end. When the loading hopper 30 is viewed from vertically above, the bottom and the outer edge of the opening at the upper end of the side wall may have any shape. The loading hopper 30 may be truncated cone-shaped or truncated square pyramidal. Alternatively, the loading hopper 30 may have a shape in which the outer edge of the bottom is circular and the outer edge of the opening at the upper end of the side wall is other than circular (for example, square). In the following, the contents of this disclosure will be explained using the case in which the loading hopper 30 is truncated square pyramidal as an example, but the shape of the loading hopper 30 is not limited to this. The loading hopper 30 may have a bottom 32 and side walls 34a, 34b, 34c, and 34d, as shown in Figures 2(a) and 2(b). The bottom 32 is positioned horizontally, and in plan view (viewed from vertically above), the outer edge of the bottom 32 is rectangular. Each of the side walls 34a, 34b, 34c, and 34d is connected to one edge of the bottom 32 and is formed to extend in a vertically oblique direction. Each side wall is inclined vertically so as it moves upward, away from the bottom 32.

[0028] The bottom 32 and side walls 34a, 34b, 34c, and 34d form the storage space within the loading hopper 30. The storage space within the loading hopper 30 is open at the top, and the outer edge of the opening at the top is rectangular. In the storage space within the loading hopper 30, the area in the horizontal plane decreases from the top towards the bottom 32. The bottom 32 is provided with an opening (not shown) for discharging ready-mixed concrete to the transport vehicle C. Figure 2(b) illustrates the positional relationship between the loading hopper 30 and the opening 24 provided at the bottom of the main body of the mixer 20 when viewed from vertically below the loading hopper 30.

[0029] The opening 24 of the mixer 20 may be formed to extend along one horizontal direction. The opening 24 may be positioned between the pair of stirring members 21 described above and may be formed to extend along the axis of rotation of the stirring members 21. The loading hopper 30 may be positioned such that the pair of outer edges of the bottom 32 are aligned with the extending direction of the opening 24. The side walls 34a and 34b are positioned to sandwich the bottom 32 in the extending direction of the opening 24. The side walls 34c and 34d are positioned to sandwich the bottom 32 and the opening 24 in one horizontal direction perpendicular to the extending direction of the opening 24.

[0030] In the direction of extension of the opening 24, the length of the opening 24 may be longer than the bottom 32, and may be shorter than the length (shortest distance) between the upper end of the side wall 34a and the upper end of the side wall 34b. In this case, when fresh concrete is discharged from the opening 24, the fresh concrete may fall not only onto the upper surface of the bottom 32, but also onto the inner surfaces of the side walls 34a and 34b. The fresh concrete that falls onto the inner surface of the side wall 34a or the inner surface of the side wall 34b flows down along these inner surfaces towards the bottom 32.

[0031] In this disclosure, a unit of ready-mixed concrete produced in a single mixing cycle in the mixer 20 and loaded onto the transport vehicle C via the loading hopper 30 is defined as "1 batch." The process for producing 1 batch of ready-mixed concrete is defined as "batch processing." In one example, 1 to 3 batches of ready-mixed concrete are loaded onto one transport vehicle C. For example, if 2 batches of ready-mixed concrete are loaded onto one transport vehicle C, two batch processes, each following the same production conditions, are performed at different times (sequentially).

[0032] (Vibration sensor) The vibration sensor 40 is a sensor that detects vibrations (magnitude of vibrations) caused by the dropping or flow of ready-mixed concrete produced by the mixer 20. Vibrations caused by dropping or flowing include vibrations caused by both dropping and flowing of the ready-mixed concrete. The vibration sensor 40 may be capable of detecting the magnitude of vibrations in two or more directions caused by the dropping or flowing of the ready-mixed concrete. The vibration sensor 40 is attached to, for example, the loading hopper 30 to detect the magnitude of vibrations in the loading hopper 30. The vibration sensor 40 may be installed on either side wall of the loading hopper 30. The vibration sensor 40 may be installed on one of the side walls 34a and 34b from which ready-mixed concrete discharged from the mixer 20 can fall. In the example shown in Figures 2(a) and 2(b), the vibration sensor 40 is provided on the outer surface of side wall 34a.

[0033] The vibration sensor 40 may detect the magnitude of vibration of the loading hopper 30 in a direction intersecting the side wall 34a. In addition to vibration in a direction intersecting the side wall 34a, the vibration sensor 40 may also detect the magnitude of vibration of the loading hopper 30 in a direction along the side wall 34a. For example, the vibration sensor 40 detects the magnitude of vibration of the loading hopper 30 in a direction perpendicular to the side wall 34a, and the magnitude of vibration in two directions that are along the side wall 34a and perpendicular to each other. In this disclosure, the direction perpendicular to the side wall 34a is defined as the "Z-axis," and the horizontal and vertical directions when viewing the side wall 34a from the direction perpendicular to the side wall 34a are defined as the "X-axis" and "Y-axis," respectively.

[0034] The vibration sensor 40 may be of any type as long as it is capable of detecting the magnitude of vibration of the loading hopper 30. For example, the vibration sensor 40 is a sensor that detects the acceleration of the loading hopper 30. The vibration sensor 40 may detect the acceleration of the loading hopper 30 in the X-axis direction, the Y-axis direction, and the Z-axis direction. The vibration sensor 40 may detect the acceleration of each axis at a predetermined sampling period. The sampling period may be about 0.001 seconds to 1.0 second (for example, 0.5 seconds).

[0035] The vibration sensor 40 may, instead of detecting acceleration, detect the speed, angular velocity, position information, displacement, or both acceleration and angular velocity of the loading hopper 30. The sensor that detects position information may be a GPS sensor. The vibration sensor 40 may continuously detect physical quantities indicating the magnitude of vibration, such as acceleration, while the manufacturing system 1 (manufacturing equipment 10) is in operation. The vibration sensor 40 outputs the detected values, such as acceleration, to the control device 60. By the vibration sensor 40 directly detecting physical quantities indicating the vibration of the loading hopper 30 without using a medium such as air, the influence of noise generated by the manufacturing system 1 and the transport vehicle C on the detected vibration values ​​can be reduced.

[0036] (Image sensor) The image sensor 50 is a sensor (camera) that images ready-mixed concrete. In this disclosure, imaging ready-mixed concrete includes imaging ready-mixed concrete after it has been manufactured in the mixer 20, as well as imaging ready-mixed concrete while it is being manufactured in the mixer 20. Imaging ready-mixed concrete after it has been manufactured in the mixer 20 includes imaging ready-mixed concrete in the loading hopper 30 and imaging ready-mixed concrete being loaded from the loading hopper 30 into the transport vehicle C. The image data obtained by imaging with the image sensor 50 may be either still image data or video data. The image sensor 50 may perform imaging based on an operation instruction from the control device 60. The image sensor 50 outputs the image data obtained by imaging to the control device 60.

[0037] The image sensor 50 may be positioned to capture images of the ready-mixed concrete being manufactured in the mixer 20. The image sensor 50 may be positioned to capture images of the ready-mixed concrete contained in the space within the loading hopper 30. The image sensor 50 may be positioned in a location that is less susceptible to interference from splashing ready-mixed concrete in the mixer 20 or loading hopper 30. Figures 1 and 2(a) illustrate an example in which the image sensor 50 captures images of the ready-mixed concrete in the loading hopper 30 from an oblique angle above. The field of view of the image sensor 50 may include the entire bottom 32 and at least a portion of the four side walls. If the mixer 20 is a tilting mixer, it is difficult to capture images of the inside of the mixer. Therefore, by positioning the image sensor 50 to capture images of the inside of the loading hopper 30, it becomes possible to acquire image data of the ready-mixed concrete even if the mixer 20 is a tilting mixer.

[0038] The image sensor 50 may be positioned to capture images of the ready-mixed concrete being discharged from the loading hopper 30. The field of view of the image sensor 50 may include the area from the discharge port at the bottom of the loading hopper 30 to the input port of the transport vehicle C. The image sensor 50 may also be positioned to capture images of at least one of the ready-mixed concrete flowing down near the outlet of the loading hopper 30 and the ready-mixed concrete flowing down near the input port of the transport vehicle C. The image sensor 50 may include sensors capable of capturing images of two or more imaging targets from among the ready-mixed concrete in the mixer 20, the ready-mixed concrete in the loading hopper 30, and the ready-mixed concrete being discharged from the loading hopper 30. When imaging is performed near or inside the mixer 20 or the loading hopper 30, disturbances may occur in the image information from the image sensor 50 due to bounced ready-mixed concrete. As a result, this may affect the construction of the prediction model described later. In contrast, by incorporating images of the ready-mix concrete being loaded into the transport vehicle C into the image information from the image sensor 50, the effects of the above-mentioned disturbances can be reduced, making it possible to acquire image information of the ready-mix concrete more stably.

[0039] (Control device) The control device 60 is a device that controls the manufacturing apparatus 10. The control device 60 is composed of one or more control computers. An input / output device 62 may be connected to the control device 60 (see Figure 3). The input / output device 62 is a device that inputs information indicating instructions from an operator, etc., to the control device 60 and outputs information from the control device 60 to the operator, etc. The input / output device 62 may include a keyboard, operation panel, or mouse as an input device, and may include a monitor (e.g., a liquid crystal display) as an output device. The input / output device 62 may be a touch panel that integrates the input and output devices. The control device 60 and the input / output device 62 may be integrated.

[0040] The control device 60 may control the manufacturing apparatus 10 according to predetermined operating conditions. At least a portion of the operating conditions may be determined by instructions from an operator or the like. In addition to controlling the manufacturing apparatus 10, the control device 60 may be configured to predict the quality of the ready-mixed concrete produced by the manufacturing apparatus 10. In this case, the control device 60 constitutes a quality prediction device that predicts the quality of the ready-mixed concrete.

[0041] The control device 60 (quality prediction device) is configured to perform an acquisition step to acquire input information including at least vibration information indicating the magnitude of vibrations caused by the dropping or flow of ready-mixed concrete produced by the mixer 20, and image information of the ready-mixed concrete. The control device 60 is also configured to perform a prediction step to predict the quality of the ready-mixed concrete based on a prediction model pre-built by machine learning to output quality information indicating the quality of the ready-mixed concrete in response to the input information, and a construction step to build the prediction model before performing the acquisition step and the prediction step.

[0042] As shown in Figure 3, the control device 60 has the following functional components (hereinafter referred to as "functional modules"): an operating condition holding unit 72, an operating control unit 74, an input data acquisition unit 76, a quality prediction unit 78, a model holding unit 82, a model construction unit 84, and a notification unit 86. The processes performed by these functional modules correspond to the processes performed by the control device 60.

[0043] The operating condition holding unit 72 holds (stores) information indicating the operating conditions for operating the manufacturing apparatus 10. The operating conditions held by the operating condition holding unit 72 may be predetermined by the operator or the like. The operator or the like may adjust (change) the operating conditions according to the target quality of the ready-mixed concrete.

[0044] The operation control unit 74 controls the manufacturing apparatus 10 according to the operation conditions held by the operation condition holding unit 72. The operation conditions may include the concrete material mix conditions, the conditions of the materials used, and the conditions during the mixing of the concrete material in the mixer 20.

[0045] The input data acquisition unit 76 (information acquisition unit) acquires vibration information indicating the magnitude of vibrations caused by the falling or flowing of ready-mix concrete from the detection results of the vibration sensor 40, and acquires image information of the ready-mix concrete from the imaging results of the image sensor 50. In this way, the input data acquisition unit 76 acquires input information including vibration information and image information from the detection results of the vibration sensor 40 and the imaging results of the image sensor 50. The input information acquired by the input data acquisition unit 76 may also include operation information. The input data acquisition unit 76 may acquire input information for each batch process. The vibration information, image information, and operation information will be described below.

[0046] <Acquisition of vibration information> The input data acquisition unit 76 may acquire vibration information from the detection result of the vibration sensor 40 attached to the loading hopper 30. If the vibration sensor 40 is a sensor that detects acceleration, the input data acquisition unit 76 may acquire information indicating the acceleration of the loading hopper 30 as vibration information from the detection result of the vibration sensor 40. The following example illustrates the case where the vibration sensor 40 is a sensor that detects acceleration.

[0047] The input data acquisition unit 76 may acquire information indicating the acceleration of the loading hopper 30 in two or more directions as the magnitude of vibration of the loading hopper 30. If a vibration sensor 40 (accelerometer) is installed on the side wall of the loading hopper 30, the vibration information acquired by the input data acquisition unit 76 may include information indicating the acceleration of the loading hopper 30 in a direction intersecting the side wall, and information indicating the magnitude of the acceleration of the loading hopper 30 in a direction along the side wall. In one example, the vibration information acquired by the input data acquisition unit 76 may include information indicating the acceleration of the loading hopper 30 in the X-axis direction, the Y-axis direction, and the Z-axis direction, respectively.

[0048] The vibration information acquired by the input data acquisition unit 76 may also be information indicating the acceleration of the loading hopper 30 when ready-mixed concrete is discharged to the loading hopper 30. The input data acquisition unit 76 may acquire as vibration information the acceleration values ​​detected by the vibration sensor 40 during any period including the time when ready-mixed concrete is discharged from the mixer 20 to the loading hopper 30. For example, the input data acquisition unit 76 acquires a reference time corresponding to the time when ready-mixed concrete is discharged to the loading hopper 30. Then, after acquiring the reference time, the input data acquisition unit 76 acquires as vibration information time-series information indicating the magnitude of vibration of the loading hopper 30 during the period from a time before the reference time to a time after the reference time.

[0049] Here, we will explain an example of acquiring vibration information from the acceleration detection results of the vibration sensor 40. In the example below, the control device 60 (input data acquisition unit 76) acquires vibration information when it is unaware of the timing of the discharge of ready-mixed concrete from the mixer 20 to the loading hopper 30. The input data acquisition unit 76 acquires data showing the time change of acceleration obtained during a predetermined data acquisition period in the X, Y, and Z axis directions from the detection results of the vibration sensor 40. The predetermined data acquisition period is set in advance and includes the timing of one discharge of ready-mixed concrete from the mixer 20. In one example, the predetermined data acquisition period is set to the period from the start of one batch process (production of one batch of ready-mixed concrete) until the completion of the said batch process (loading onto the transport vehicle C).

[0050] Figure 4(a) shows a graph illustrating an example of the time change in acceleration in the Z-axis direction when the manufacturing apparatus 10 continues to operate and produces and loads dozens of batches of ready-mixed concrete. In the graph in Figure 4(a), an example of the data acquisition period is indicated by "Ta". As ready-mixed concrete is discharged from the mixer 20 to the loading hopper 30, the concrete falls into the loading hopper 30, causing a large change in the acceleration in the Z-axis direction in the negative direction (towards the outside of the loading hopper 30). The input data acquisition unit 76 calculates the time at which the acceleration in the Z-axis direction shows the minimum value (the absolute value of the acceleration in the Z-axis is maximum) in the acceleration data obtained during the data acquisition period as the reference time ts.

[0051] Next, the input data acquisition unit 76 sets the evaluation period to the time before the reference time ts and the time after the reference time ts within the data acquisition period. The evaluation period is the period during which the detected acceleration values ​​are acquired as vibration information. The evaluation period is set to include the time when the ready-mixed concrete is discharged from the mixer 20 to the loading hopper 30 (the time when it falls). The method for setting the evaluation period is predetermined.

[0052] The input data acquisition unit 76 extracts data for the evaluation period from the continuous detection values ​​by the vibration sensor 40 during the data acquisition period. Figure 4(b) shows a graph representing the time change of acceleration in the Z-axis direction during the evaluation period. In the example shown in Figure 4(b), the disclosure point of the evaluation period is set to 5 seconds before the reference time ts, and the end point of the evaluation period is set to 5 seconds after the reference time ts. 5 seconds is just an example; the evaluation period may be set to a period of several seconds to several tens of seconds around the reference time ts. The input data acquisition unit 76 may also set the evaluation period such that the time between the reference time ts and the disclosure point is different from the time between the reference time ts and the end point.

[0053] The input data acquisition unit 76 extracts (acquires) data as vibration information from the continuous detection values ​​by the vibration sensor 40 for each of the X, Y, and Z axis directions during the same evaluation period. In one example, if the evaluation period is set to a total of 10 seconds and the sampling period is 0.5 seconds, 63 acceleration values ​​(3 axes × 21) are acquired as vibration information.

[0054] <Acquiring image information> The input data acquisition unit 76 may acquire image data obtained by the image sensor 50 capturing images of the ready-mixed concrete in the loading hopper 30 as image information of the ready-mixed concrete. The input data acquisition unit 76 may also acquire video data obtained by capturing images of the inside of the loading hopper 30 from the image sensor 50. The input data acquisition unit 76 may cause the image sensor 50 to perform imaging when ready-mixed concrete is present in the loading hopper 30. For example, in a single batch process, the input data acquisition unit 76 causes the image sensor 50 to perform imaging for at least a portion of the period from when the discharge of ready-mixed concrete from the mixer 20 to the loading hopper 30 is completed until loading from the loading hopper 30 to the transport vehicle C begins.

[0055] The input data acquisition unit 76 may acquire as image information information information indicating still images (multiple still images) for each frame included in the video data generated by imaging by the image sensor 50. The input data acquisition unit 76 may acquire as image information information information indicating multiple still images captured at different timings in a single batch process. The input data acquisition unit 76 may acquire as image information information information indicating images obtained by imaging the ready-mixed concrete in the mixer 20 instead of the loading hopper 30.

[0056] <Acquisition of operation information> The input data acquisition unit 76 acquires operational information used to predict the quality of ready-mixed concrete. The operational information includes, for example, condition information indicating the operational conditions for operating the manufacturing apparatus 10, state information indicating the state of the apparatus or materials when the manufacturing apparatus 10 is actually operating, and environmental information indicating the environment when the manufacturing apparatus 10 is operating.

[0057] The operational information may include at least one of the following types of information: information regarding the mixing conditions of various materials, information regarding the materials used, information regarding mixing in the mixer 20, and information regarding the environment (external environment). Information regarding the mixing conditions of various materials may include, for example, guaranteed strength (nominal strength), mix strength, target slump, target slump flow, target air content, water-binder ratio, fine aggregate ratio, and unit amounts [kg / m³] of various materials (water, cement, admixtures, fine aggregate, coarse aggregate, and admixtures, etc.). 3 ] or unit volume [L / m³ 3 It includes at least one type of information from the following categories.

[0058] Information regarding the materials used includes, for example, the type, brand, manufacturing date, and density [g / cm³] of each material. 3], and also includes information on at least one of the following for aggregates (fine aggregates and coarse aggregates): water absorption rate, water content, surface moisture content, actual volume, coarseness ratio, particle size distribution, and maximum dimensions. Information regarding mixing in the mixer 20 includes, for example, information on the conditions or state of mixing the concrete materials in the mixer 20, and information on at least one of the temperature of the fresh concrete being produced in the mixer 20. Information regarding the conditions or state of mixing the concrete materials in the mixer 20 includes, for example, information on at least one of the following: the amount of each material to be mixed (total amount), the set value of the mixing time, the measured value of the mixing time, and the power load value of the mixer 20.

[0059] The operational information may include, in lieu of or in addition to the information described above, at least one of the following: information about cement and information about cement clinker. Information about cement may include, for example, at least one of the following: type of cement, chemical composition, mineral composition, wet f.CaO, loss on ignition, Blaine specific surface area, particle size distribution, sieve test residue, color, mineralogical and crystallographic properties of each mineral contained in the cement, and the hemihydrated ratio of gypsum contained in the cement.

[0060] Information regarding cement clinker includes, for example, the chemical composition, hydraulic coefficient, sieve test residue, Blaine specific surface area (powderness), and ignition loss of the raw materials used in the cement clinker formulation; the mineral composition, chemical composition, wet f.CaO (free lime), and volumetric gravity of the cement clinker; the crystallographic properties (lattice constant or crystallite size, etc.) of each mineral contained in the cement clinker; and the ratio of two or more mineral compositions contained in the cement clinker, with at least one type of information selected from this group.

[0061] The power load value of the mixer 20 may be measured in the mixer drive unit 22. The information indicating the power load value of the mixer 20 may be continuous time-series data while the mixer 20 is operating, statistical values ​​obtained from the time-series data, or power load values ​​at a few representative points. The information indicating the power load value may be the initial power load value at the start of operation of the mixer 20 (immediately after start), the maximum power load value during operation of the mixer 20, and the final power load value at the end of operation of the mixer 20 (immediately before end). The difference between the initial and maximum power load values ​​of the mixer 20, and the final power load value of the mixer 20, are thought to be affected by the hardness (softness) of the ready-mixed concrete, etc.

[0062] Information regarding the environment (external environment) includes, for example, at least one of the following: ambient temperature, humidity, temperature inside the mixer 20, temperature of various materials, and temperature of the containers storing the various materials. The various operational information described above is just an example, and any information that may affect the quality of the ready-mixed concrete may be included. The input information acquired by the input data acquisition unit 76 may include vibration information representing the time change of acceleration in the X, Y, and Z axes, image information of the ready-mixed concrete in the loading hopper 30, and the amount of mixing in the mixer 20.

[0063] The quality prediction unit 78 predicts the quality of the ready-mixed concrete based on input information including vibration information and image information. The quality of the ready-mixed concrete predicted by the quality prediction unit 78 includes, for example, at least one of the following: slump, slump flow, time to reach 50 cm flow, V-funnel flow time, air content, setting start time, setting end time, strength (guaranteed (nominal) strength, compressive strength, and flexural strength, etc.), Young's modulus, chloride content, durability, and color tone. The quality of the ready-mixed concrete may also include the quality of the hardened body formed when the ready-mixed concrete hardens.

[0064] The quality prediction unit 78 may predict both slump and slump flow as qualities of the ready-mixed concrete. The quality prediction unit 78 may predict both slump and slump flow when producing ready-mixed concrete with a slump of 21 cm or less, 23 cm or less, or 25 cm or less. In addition to predicting both slump and slump flow, or instead, the quality prediction unit 78 may predict the amount of air contained in the ready-mixed concrete as a quality of the ready-mixed concrete. The strength, slump, slump flow, air content, and chloride content of the ready-mixed concrete can be measured by the test methods specified in "JIS A 5308:2019 (Ready-Mixed Concrete)".

[0065] The quality prediction unit 78 predicts the quality of ready-mixed concrete based on a prediction model pre-built by machine learning to output quality information indicating the quality of ready-mixed concrete in response to input information including vibration information and image information, and input information acquired by the input data acquisition unit 76. The input information input to the prediction model may include multiple types of numerical data and image data, and the quality information output from the prediction model may be data (numerical value) indicating at least one type of quality. The quality prediction unit 78 may select a model corresponding to the type of quality to be predicted from multiple prediction models pre-built for multiple quality types, and then predict the quality of the ready-mixed concrete using the selected prediction model.

[0066] Machine learning is a method in which a machine (computer) autonomously discovers laws or rules by iteratively learning based on given information. The predictive model described above can be constructed using algorithms and data structures. For example, a predictive model can be realized using a neural network, which is an information processing model that mimics the structure of the human brain. The specific machine learning algorithm used when constructing a predictive model is not particularly limited.

[0067] A neural network has at least an input layer and an output layer. A neural network may also include one or more hidden layers. Including hidden layers allows for the construction of more complex predictive models and improves the accuracy of quality predictions. A single predictive model may include one or more hidden models constructed by the neural network. Vibration information may be input to any of the hidden models, and image information may be input to any of the hidden models. Output data from other hidden models may be input to a hidden model that outputs quality information.

[0068] One or more intermediate models may include at least one of the following: a regression prediction model (e.g., a first-order regression model), a neural network with an intermediate layer, a convolutional neural network, and a recurrent neural network. The input data for the convolutional neural network may be the image information described above. The input data for the recurrent neural network may be time-series information obtained during the evaluation period (time-series data of acceleration in three axes).

[0069] The model holding unit 82 stores the prediction model. The model building unit 84 builds the prediction model by performing machine learning based on test input information for building the prediction model. The model building unit 84 may build a prediction model for each quality type. The model building unit 84 acquires test input information, which includes test vibration information indicating the magnitude of vibration caused by the dropping or flow of test ready-mixed concrete produced by the mixer 20, and image information of the test ready-mixed concrete. The model building unit 84 then builds a prediction model based on the test input information and the measured quality values ​​of the test ready-mixed concrete associated with the test input information.

[0070] The model building unit 84 may autonomously construct a predictive model to predict quality by performing machine learning using, for example, data given as input to machine learning and ground truth data (measured quality values) from the output of machine learning. The input to machine learning is various datasets of input information. In the various datasets of input information, at least one piece of information contained in the input information is different from each other. The output of machine learning is data (numerical values) indicating the quality of ready-mixed concrete. The model building unit 84 iteratively learns a predictive model that outputs quality information indicating the quality of ready-mixed concrete using multiple combinations of the datasets of input information and measured quality values.

[0071] When the model building unit 84 constructs a prediction model, test input information is prepared, which includes test vibration information indicating the magnitude of vibrations caused by the dropping or flow of test ready-mix concrete produced by the mixer 20, and image information of the test ready-mix concrete. In addition, measured quality values ​​associated with the test input information are prepared. When the prediction model is constructed, for example, multiple combinations (multiple datasets) of test input information and measured quality values ​​are input to the control device 60. The measured quality values ​​are not values ​​that predict quality from certain measurement values ​​(predicted values), but rather measured values ​​of the quality itself. The model building unit 84 may also construct a prediction model using machine learning with a neural network based on the test input information and performance information.

[0072] The stage in which the above-mentioned predictive model is autonomously manufactured corresponds to the learning phase. The above-mentioned learning phase may be performed before the production phase in which ready-mixed concrete is produced, or it may be performed in the early stages of the production phase. The trained predictive model may be portable between computers. Therefore, the predictive model constructed in the control device 60 may be used in a manufacturing system other than the manufacturing system 1.

[0073] The quality prediction unit 78, during the production or evaluation phase, uses the prediction model held in the model holding unit 82 to predict the quality of the ready-mixed concrete produced when the input information was obtained, based on input information whose quality is unknown. The quality prediction unit 78 may also predict the quality of the ready-mixed concrete for each batch process based on the input information obtained in that batch process.

[0074] The notification unit 86 notifies the operator or others of the prediction results from the quality prediction unit 78. The notification unit 86 outputs quality information indicating the quality of the ready-mixed concrete predicted by the quality prediction unit 78 to the input / output device 62. The notification unit 86 may also display the above quality information on the monitor of the input / output device 62. After the operator or others confirm the quality information notified to the input / output device 62, they may decide whether or not the quality information meets the target quality. Alternatively, the control device 60 (for example, the quality prediction unit 78) may decide whether or not the predicted quality information meets the target quality, and then the notification unit 86 may output information indicating the result of that decision to the input / output device 62.

[0075] As shown in Figure 5, the control device 60 has a circuit 91. The circuit 91 includes one or more processors 92, a memory 93, a storage 94, an input / output port 95, and a timer 96. The storage 94 has a computer-readable storage medium, such as a non-volatile semiconductor memory. The storage 94 stores a program that causes the control device 60 to execute a control procedure to control the manufacturing equipment 10, the vibration sensor 40, the image sensor 50, and the input / output device 62. For example, the storage 94 stores a program to configure each of the above-mentioned functional modules.

[0076] Memory 93 temporarily stores the program loaded from the storage medium of storage 94 and the calculation results from processor 92. Processor 92 works in cooperation with memory 93 to execute the above program, thereby configuring each functional module of control device 60. Input / output ports 95 input and output electrical signals to and from the manufacturing equipment 10, vibration sensor 40, image sensor 50, and input / output devices 62, etc., according to commands from processor 92. Timer 96 measures elapsed time, for example, by counting reference pulses of a fixed period. Note that circuit 91 is not necessarily limited to configuring each function by program. For example, circuit 91 may configure at least some of its functions by dedicated logic circuits or an ASIC (Application Specific Integrated Circuit) that integrates them.

[0077] [Method of manufacturing ready-mixed concrete] Next, an example of a ready-mix concrete manufacturing method performed in manufacturing system 1 will be described. The ready-mix concrete manufacturing method includes a process for manufacturing ready-mix concrete and a process for predicting the quality of the ready-mix concrete. The process for manufacturing ready-mix concrete (hereinafter referred to as the "manufacturing process") includes, for example, a transport process, a weighing process, a input process, a mixing process, a discharge process, and a loading process. In the manufacturing process, these processes may be performed sequentially for each batch of ready-mix concrete to be manufactured. One batch process includes mixing in mixer 20, discharge from mixer 20 to loading hopper 30, and loading from loading hopper 30 to transport vehicle C. While the mixing process, discharge process, and loading process are being performed, the transport process and weighing process for the next batch may be performed.

[0078] In the conveying process, various concrete materials are transported to the storage bottle 12 by a conveying device such as a belt conveyor, and each material is supplied to the storage bottle 12 individually. In the weighing process, each material is supplied individually from the storage bottle 12 to the weighing bottle 14, and each material is weighed in the weighing bottle 14. In the weighing process, when the measured amount of each material reaches the set amount for one batch, that material is discharged to the collection hopper 16 or discharged directly to the mixer 20. If a collection hopper 16 is provided, in the input process, after all types of materials are collected in the collection hopper 16, one batch of material in the collection hopper 16 is input (supplied) to the mixer 20. If a collection hopper 16 is not provided, and one batch of various materials weighed in the weighing bottle 14 is directly input (supplied) to the mixer 20, the input process is performed after the weighing process.

[0079] In the mixing process, the concrete material is mixed in the mixer 20. During the mixing process, the control device 60 may control the mixer drive unit 22 to drive the two stirring members 21 according to predetermined operating conditions. In the discharge process, after the mixing of the concrete material in the mixer 20 is completed, the fresh concrete produced is discharged from the mixer 20 to the loading hopper 30. For example, the control device 60 determines that mixing in the mixer 20 is complete when it determines that the power load value of the mixer 20 has stabilized (in one example, when the time change of the power load value falls below a predetermined value). In the loading process, after one batch of fresh concrete is discharged from the mixer 20 to the loading hopper 30, the fresh concrete is loaded from the loading hopper 30 onto the transport vehicle C.

[0080] The process of predicting the quality of ready-mixed concrete (hereinafter referred to as the "quality prediction process") is performed during a period that overlaps with at least a portion of the execution period of the above-mentioned manufacturing process. The quality prediction process (quality prediction method) includes a model building process in the learning phase and a quality evaluation process in the production phase or evaluation phase. The quality evaluation process is performed while batch processing is being repeated. An example of the model building process and an example of the quality evaluation process will be described below. Note that the explanation will be based on the example where one quality evaluation is performed each time one batch of ready-mixed concrete is produced (for each batch processing).

[0081] (Model building process) Figure 6 is a flowchart showing an example of a series of processes performed in the model building process. This model building process is performed in parallel with the repeated production of test ready-mixed concrete in the manufacturing apparatus 10. The test ready-mixed concrete may be ready-mixed concrete that will not actually be shipped, or it may be ready-mixed concrete that is manufactured in the early stages of the production phase and will actually be shipped. While the model building process is being executed, the input data acquisition unit 76 of the control device 60 continues to acquire acceleration detection values ​​from the vibration sensor 40 at predetermined sampling periods.

[0082] In the model building process, the control device 60 first executes step S11. In step S11, for example, the control device 60 waits until one batch process is completed. The control device 60 may determine that one batch process is complete when the loading of the ready-mix concrete contained in the loading hopper 30 onto the transport vehicle C is completed.

[0083] Next, the control device 60 executes step S12. In step S12, for example, the input data acquisition unit 76 acquires test input information, which includes test vibration information and test image information. The test vibration information may be time-series data (one dataset) of acceleration during the evaluation period in the X, Y, and Z axis directions, respectively. The test image information may be image information of test ready-mixed concrete in the loading hopper 30. The test image information may also be image data (multiple datasets) of each of multiple frames included in the video data obtained from the image sensor 50. Each of the multiple datasets relating to images may be associated with one identical dataset relating to acceleration. The test input information obtained by the input data acquisition unit 76 may include test operation information.

[0084] Next, the control device 60 executes step S13. In step S13, for example, the model building unit 84 acquires measured values ​​of the quality of the ready-mixed concrete as ground truth data. In one example, after the ready-mixed concrete is loaded onto the transport vehicle C, some of the concrete is extracted by a worker or the like. The slump, slump flow, and air content of the extracted concrete are then measured by the worker or the like, and these measured values ​​are input to the control device 60. The control device 60 may associate the test input information with the measured quality values ​​and store this information. For each type of quality, multiple image datasets may be associated with one acceleration dataset and one measured quality value.

[0085] Next, the control device 60 executes step S14. In step S14, for example, the control device 60 determines whether a predetermined number of data sets have been acquired. If it is determined in step S14 that a predetermined number of data sets have not been acquired (step S14: NO), the control device 60 returns to step S11, and the control device 60 repeats steps S11 to S14. This process of repeating steps S11 to S14 continues until the number of data sets, which are associated with test input information and measured quality values, reaches a predetermined number.

[0086] If it is determined in step S14 that a predetermined number of datasets have been acquired (step S14: YES), the control device 60 proceeds to step S15. In step S15, for example, the model building unit 84 uses multiple datasets of test input information and actual quality values ​​to build a predictive model using machine learning with a neural network. The model storage unit 82 stores the predictive model built in step S15.

[0087] (Quality evaluation process) Figure 7 is a flowchart showing an example of a series of processes performed in the quality evaluation process. The quality evaluation process is performed in parallel with the production of ready-mixed concrete to be actually shipped, which is repeatedly carried out in the manufacturing apparatus 10, after the model construction process described above. While the production of ready-mixed concrete to be actually shipped continues, the input data acquisition unit 76 of the control device 60 may continue to acquire acceleration detection values ​​from the vibration sensor 40 at predetermined sampling periods.

[0088] In the quality evaluation process, the control device 60 executes steps S21 and S22. In step S21, for example, the control device 60 executes the same process as in step S11 described above. In step S22, for example, the input data acquisition unit 76 acquires evaluation input information, which includes evaluation vibration information obtained during the execution period of the batch process to be evaluated, and evaluation image information of the ready-mixed concrete. "For evaluation" means that the quality of the ready-mixed concrete is unknown. The vibration information may be time-series data (one dataset) of acceleration in the X, Y, and Z axis directions during the evaluation period. The evaluation image information may be image information of the ready-mixed concrete in the loading hopper 30. The image information of the ready-mixed concrete may be image data in a single frame, or it may be still image data captured at a certain timing.

[0089] Next, the control device 60 executes step S23. In step S23, for example, the quality prediction unit 78 inputs the evaluation input information obtained in step S22 into the prediction model held by the model holding unit 82, and predicts the quality of the ready-mixed concrete to be evaluated by acquiring the quality information output from that prediction model. Unlike the model construction process, the quality of the ready-mixed concrete is not measured in the quality evaluation process. The quality prediction unit 78 may predict the quality of the ready-mixed concrete for each quality type using a prediction model corresponding to that quality type (constructed based on measured values ​​of that quality type).

[0090] In one example, the quality prediction unit 78 uses the input information for evaluation and a prediction model for slump to predict the slump of the ready-mixed concrete to be evaluated, and uses the input information for evaluation and a prediction model for slump flow to predict the slump flow of the ready-mixed concrete to be evaluated. The quality prediction unit 78 also uses the input information for evaluation and a prediction model for air content to predict the air content of the ready-mixed concrete to be evaluated.

[0091] Next, the control device 60 executes step S24. In step S24, for example, the notification unit 86 outputs the quality prediction result obtained in step S23 to the input / output device 62. With this, one quality evaluation process performed for each batch processing is completed. The control device 60 may also execute the series of processes from steps S21 to S24 each time a batch of ready-mixed concrete is manufactured (for each batch processing). In this case, for each batch processing, an acquisition process to acquire input information and a prediction process to predict the quality of the ready-mixed concrete are executed.

[0092] [Differentiation] The series of processes shown in Figures 6 and 7 are examples and can be modified as appropriate. In the above series of processes, the control device 60 may execute one step and the next step in parallel, or it may execute each step in a different order than the example described above. The control device 60 may also execute steps with content different from the example described above.

[0093] Unlike the example described above, the control device 60 may predict the quality of the ready-mixed concrete for each batch of ready-mixed concrete produced (for each batch process). Alternatively, for each batch of ready-mixed concrete produced, the control device 60 may obtain measured values ​​of the ready-mixed concrete quality during the model construction process.

[0094] In the example described above, the time at which the absolute value of acceleration in the Z-axis is maximum is calculated as the reference time ts, but the method for calculating the reference time ts is not limited to this. As ready-mixed concrete is discharged from the mixer 20 to the loading hopper 30, the ready-mixed concrete falls into the loading hopper 30, causing large fluctuations in acceleration in the X and Y axes. The vibration data calculation unit 70 may calculate the time at which the absolute value of acceleration in the X-axis direction or the Y-axis direction is maximum in the time-series acceleration data obtained during the above data acquisition period as the reference time ts.

[0095] The timing at which the opening 24 of the mixer 20 is opened and the timing at which the loading hopper 30 vibrates significantly coincide. Therefore, if the control device 60 can detect the timing at which ready-mixed concrete is discharged from the mixer 20, the control device 60 may calculate the detected discharge timing, or the timing at which a predetermined time has elapsed since the discharge timing, as the reference time ts. The control device 60 may also acquire the timing at which a command to open the opening 24 of the mixer 20 is sent, or the timing at which a signal indicating that the opening 24 of the mixer 20 has been opened is received, as the discharge timing.

[0096] In the series of processes shown in Figures 6 and 7, the image information acquired by the input data acquisition unit 76 may include at least one of the following: image information of ready-mixed concrete in the loading hopper 30, image information of ready-mixed concrete in the mixer 20, and image information of ready-mixed concrete being discharged from the bottom of the loading hopper 30. For example, the image information acquired by the input data acquisition unit 76 may include image information of ready-mixed concrete in the mixer 20 but not image information of ready-mixed concrete in the loading hopper 30. The image information acquired by the input data acquisition unit 76 may include image information of ready-mixed concrete in the loading hopper 30 and image information of ready-mixed concrete in the mixer 20. The image information acquired by the input data acquisition unit 76 may include image information of ready-mixed concrete being discharged from the bottom of the loading hopper 30 but not image information of ready-mixed concrete in the loading hopper 30. The image information acquired by the input data acquisition unit 76 may include image information of ready-mixed concrete in the loading hopper 30 and image information of ready-mixed concrete being discharged from the bottom of the loading hopper 30.

[0097] The input data acquisition unit 76 may acquire time-series data of the moving average value of acceleration as vibration information instead of the acceleration value of the loading hopper 30 itself. The input data acquisition unit 76 may acquire one or more statistical values ​​for the moving average value of acceleration during the evaluation period as vibration information. One or more statistical values ​​may include the maximum value, minimum value, mean value, value based on standard deviation, median value, mode value, or number of peaks.

[0098] The input data acquisition unit 76 may calculate vibration information based on the frequency spectrum obtained by frequency transformation (e.g., Fast Fourier Transform) of the data acquired by the vibration sensor 40. The input data acquisition unit 76 may also calculate vibration information based on frequency information, feature quantities, or statistics obtained from the frequency spectrum. The input data acquisition unit 76 may also acquire image data of a graph (waveform) showing the measured values ​​acquired by the vibration sensor 40 as vibration information.

[0099] A separate computer (quality prediction device) from the control device 60 for controlling the manufacturing apparatus 10 may predict the quality of the ready-mixed concrete using a prediction model. In this case, instead of the control device 60, the separate computer may have a functional module that performs quality prediction processing, such as a quality prediction unit 78. The vibration information used for quality prediction may be information indicating vibrations caused by the dropping or flow of ready-mixed concrete in locations other than the loading hopper 30.

[0100] [Validation of the predictive model] To verify quality prediction using input information including image and vibration information, we compared the prediction results of a comparison model using only image information with the prediction results of the same prediction model using input information including both image and vibration information. Specifically, we prepared a training dataset of 24,200 to 28,200 data points and an evaluation dataset of 4,850 to 5,150 data points. In each dataset, the input information was matched with measured values ​​of slump, slump flow, and air content. Image information was obtained by imaging the ready-mixed concrete inside the loading hopper 30.

[0101] (Preparation of concrete for verification) Table 1 shows the manufacturing conditions (mix design, etc.) for ready-mixed concrete used to validate the prediction model, as well as the measured values ​​for each quality test item. [Table 1]

[0102] In Table 1, the cement types are as follows: "N" represents ordinary Portland cement, "BB" represents blast furnace cement type B, "M" represents moderate-heat Portland cement, "L" represents low-heat Portland cement, and "SFC" represents silica fume cement. Ready-mix concrete for verification by slump measurement was manufactured using multiple types of cement represented by N, BB, M, and L. The measured slump values ​​ranged from 11.5 cm to 25.0 cm. In the manufacture of ready-mix concrete for verification by slump measurement, manufacturing conditions such as guaranteed (nominal) strength were set within the range from the minimum to the maximum values ​​shown in Table 1. Ready-mix concrete for verification by slump flow and air content was manufactured using multiple types of cement represented by N, BB, M, L, and SFC. The measured slump flow values ​​ranged from 22.3 cm to 71.3 cm. The measured air content values ​​ranged from 1.2% to 7.0%. In the production of ready-mixed concrete for verification through measurements of slump flow and air content, the production conditions, such as guaranteed (nominal) strength, were set within the range from the minimum to the maximum values ​​shown in Table 1.

[0103] (Setting up the predictive model) In the verification of slump and slump flow, models for Comparative Example 1, Example 1, Example 2, Example 3, and Example 4 were constructed using the training dataset. In the verification of air content, models for Comparative Example 1, Example 1, Example 2, and Example 4 were constructed using the training dataset. In Comparative Example 1, a comparative model (a model including a convolutional neural network) was constructed to predict quality using only image information as input data. In Examples 1, 2, and 3, predictive models were constructed to predict quality using image information, vibration information, and mixing amount as input data. The predictive models were constructed to include one or more intermediate models, and one intermediate model included in the predictive models for Examples 1-3 was a convolutional neural network that receives image information as input.

[0104] In the prediction model according to Example 1, the vibration information was the detected value of acceleration itself, and the intermediate model into which the vibration information is input was a recurrent neural network. In the prediction model according to Example 2, the vibration information was the detected value of acceleration itself, and the intermediate model into which the vibration information is input was a neural network including an intermediate layer. In the prediction model according to Example 3, the vibration information was the moving average value of acceleration, and the intermediate model into which the vibration information is input was a recurrent neural network. In Example 4, a prediction model was constructed that predicts quality using image information, vibration information, mixing amount, blending conditions, and mixing conditions as input data. One intermediate model included in the prediction model according to Example 4 was a convolutional neural network into which image information is input. The vibration information was the detected value of acceleration itself, and the intermediate model into which the vibration information is input was a recurrent neural network.

[0105] In the construction of each model in Comparative Example 1 and Examples 1-4, the squared error between the output quality (e.g., slump) and the measured value was calculated when a certain value was input to the weights (coefficients) included in each intermediate model. The weight values ​​were then updated using the mini-batch gradient descent method to determine the weight values. After constructing each model in Comparative Example 1 and Examples 1-4, the predicted values ​​from the models were compared with the measured values ​​using an evaluation dataset. As evaluation metrics, the percentage of data points where the predicted value falls within a predetermined tolerance range from the measured value (hereinafter referred to as "accuracy") and the arithmetic mean of the absolute difference between the predicted value and the measured value (hereinafter referred to as "mean absolute difference") were calculated.

[0106] Table 2 shows the results of the verification regarding slumps. [Table 2]

[0107] In Table 2, the tolerance "±1.0cm" indicates that datasets where the model's predicted slump value falls within ±1.0cm of the measured slump value are judged as correct, and datasets where they fall outside this range are judged as incorrect. As shown in Table 2, compared to Comparative Example 1, which uses only image information, Examples 1 to 4 show improved accuracy and smaller absolute difference average values. In other words, the accuracy of slump prediction has improved compared to the case where only image information is used.

[0108] Table 3 shows the verification results regarding slump flow. [Table 3]

[0109] In Table 3, the tolerance "±3.0cm" indicates that datasets where the model's predicted slump flow value falls within ±3.0cm of the measured slump flow value are judged as correct, and datasets where they fall outside this range are judged as incorrect. As shown in Table 3, compared to Comparative Example 1, which uses only image information, Examples 1 to 4 show improved accuracy and smaller absolute difference average values. In other words, the prediction accuracy of slump flow is improved compared to the case where only image information is used.

[0110] Table 4 shows the verification results regarding the amount of air. [Table 4]

[0111] In Table 4, the tolerance "±1.0%" indicates that datasets where the model's predicted air volume falls within ±1.0% of the measured air volume are judged as correct, and datasets where they fall outside this range are judged as incorrect. As shown in Table 4, compared to Comparative Example 1, which uses only image information, Examples 1, 2, and 4 show improved accuracy and smaller absolute difference average values. In other words, the accuracy of air volume prediction has improved compared to using only image information.

[0112] The above verification results confirm that constructing a prediction model that uses vibration information in addition to image information improves the prediction accuracy of slump, slump flow, and air volume. The above verification results examine the effect of adding vibration information in addition to image information. It is believed that further improvement in prediction accuracy beyond the results shown in Tables 2 to 4 can be achieved by adding other types of motion information.

[0113] [Effects of the Embodiment] The ready-mixed concrete quality prediction method described above includes an acquisition step, a prediction step, and a construction step. In the acquisition step, input information is acquired, including vibration information indicating the magnitude of vibrations caused by the dropping or flow of ready-mixed concrete produced by the mixer 20 that mixes the concrete materials, and image information of the ready-mixed concrete. In the prediction step, the quality of the ready-mixed concrete is predicted based on a prediction model pre-built by machine learning to output quality information indicating the quality of the ready-mixed concrete in response to the input information, and the input information acquired in the acquisition step. In the construction step, the prediction model is constructed before the acquisition step and the prediction step are executed. The construction step includes acquiring test input information, including test vibration information indicating the magnitude of vibrations caused by the dropping or flow of test ready-mixed concrete produced by the mixer 20, and image information of the test ready-mixed concrete, and constructing a prediction model based on the test input information and measured values ​​of the quality of the test ready-mixed concrete associated with the test input information.

[0114] In this quality prediction method, vibration information is included in the input information in addition to image information that changes depending on the actual state of the ready-mixed concrete being manufactured, and the quality is predicted using a prediction model. Depending on the quality of the ready-mixed concrete, the state of vibration caused by the dropping or flow of the ready-mixed concrete may also change. Therefore, information not reflected in the image information can also be taken into consideration, and the quality can be predicted. Furthermore, since the prediction model is constructed based on actual measured values ​​obtained from the actual measurement of the ready-mixed concrete quality, the model is constructed so that the output value approaches the correct answer more closely. Thus, it is possible to improve the accuracy of predicting the quality of ready-mixed concrete.

[0115] In the quality prediction method described above, the acquisition step may include obtaining vibration information from the detection results of a vibration sensor 40 attached to a loading hopper 30, which is located below the mixer 20 and capable of containing ready-mixed concrete. The vibration information may also indicate the magnitude of vibration of the loading hopper 30 when ready-mixed concrete is discharged into the loading hopper 30. In this case, the vibration of the loading hopper 30 is directly detected by the sensor without the need for a medium such as air. Therefore, disturbances such as noise generated in the manufacturing system 1 or the transport vehicle C into which ready-mixed concrete is loaded from the manufacturing system 1 are less likely to be reflected in the vibration information. Consequently, it is possible to further improve the accuracy of predicting the quality of ready-mixed concrete.

[0116] In the quality prediction method described above, the acquisition process may include obtaining image information of the ready-mixed concrete by imaging the ready-mixed concrete inside the loading hopper 30. Depending on the type of mixer 20, it may be difficult to image the inside of the mixer 20. In contrast, the above method can acquire image information of the ready-mixed concrete regardless of the type of mixer 20. Furthermore, regarding the loading hopper 30, the same type of hopper is often used in multiple manufacturing systems 1. In this case, it is easy to use a prediction model constructed in one manufacturing system 1 in other manufacturing systems as well.

[0117] In the quality prediction method described above, the acquisition process may include acquiring a reference time ts corresponding to the time when the ready-mixed concrete is discharged from the loading hopper 30, and acquiring time-series information as vibration information indicating the magnitude of vibration of the loading hopper 30 during the period from before the reference time ts to after the reference time ts (evaluation period). When ready-mixed concrete falls into the loading hopper 30, the vibration of the loading hopper 30 tends to fluctuate significantly more than during other periods. Therefore, the vibration information obtained during the period including the time when the ready-mixed concrete falls contains a lot of information attributable to the quality of the ready-mixed concrete. Consequently, it is possible to further improve the accuracy of predicting the quality of the ready-mixed concrete.

[0118] In the quality prediction method described above, the vibration information may include information indicating the magnitude of vibration of the loading hopper 30 in a direction intersecting the side wall of the loading hopper 30, and information indicating the magnitude of vibration of the loading hopper 30 in a direction along the side wall of the loading hopper 30. At the side wall of the loading hopper 30 located below the mixer 20, the state of vibration of the side wall may fluctuate depending on the hardness or weight of the ready-mixed concrete produced. Furthermore, fluctuations in the state caused by the hardness or weight of the ready-mixed concrete may be reflected in only one direction. In the above method, information indicating vibrations in both the direction intersecting the side wall and the direction along the side wall can be obtained, making it possible to predict quality with higher accuracy.

[0119] In the quality prediction method described above, the acquisition process and the prediction process may be performed while batch processing, which includes mixing in the mixer 20, discharge from the mixer 20 to the loading hopper 30, and loading from the loading hopper 30 to the transport vehicle C, is being repeated. In the acquisition process, input information may be acquired for each batch processing. In the prediction process, the quality of the ready-mixed concrete may be predicted for each batch processing. In this case, there are more opportunities to grasp the quality prediction results compared to predicting the quality for each of the multiple batch processing steps. Also, since a prediction model is used, the amount of work required for quality prediction does not increase even if prediction is made for each batch processing step. Therefore, it is possible to ship ready-mixed concrete that is closer to the target quality while suppressing an increase in work hours.

[0120] In the quality prediction method described above, both the slump and slump flow of the ready-mixed concrete may be predicted in the prediction process. In this case, it is possible to easily determine whether both the slump and slump flow meet the target quality, and by looking at the relationship between the slump and slump flow, it is possible to evaluate the quality from different perspectives.

[0121] In the quality prediction method described above, the amount of air in the ready-mixed concrete may be predicted during the prediction process. In this case, it is possible to easily determine whether or not the amount of air meets the target quality. [Explanation of symbols]

[0122] 1... Ready-mix concrete manufacturing system, 10... Ready-mix concrete manufacturing equipment, 20... Mixer, 30... Loading hopper, 40... Vibration sensor, 50... Image sensor, 60... Control device, 76... Input data acquisition unit, 78... Quality prediction unit, 82... Model holding unit, 84... Model construction unit, C... Transport vehicle.

Claims

1. An acquisition step to acquire input information including vibration information indicating the magnitude of vibration caused by the dropping or flow of ready-mixed concrete produced by a mixer for mixing concrete materials, and image information of the ready-mixed concrete. A prediction model pre-built by machine learning to output quality information indicating the quality of the ready-mixed concrete in response to the input of the aforementioned input information, and a prediction step that predicts the quality of the ready-mixed concrete based on the input information acquired in the acquisition step, The process includes a construction step for constructing the prediction model before performing the acquisition step and the prediction step, The acquisition step includes obtaining the vibration information from the detection result of a sensor located below the mixer and attached to a hopper capable of containing the ready-mixed concrete, The sensor detects a physical quantity indicating the vibration of the hopper without the use of a medium, The vibration information is information indicating the magnitude of vibration of the hopper when the ready-mixed concrete is discharged from the hopper. The aforementioned construction process is, The method involves obtaining test vibration information indicating the magnitude of vibrations caused by the dropping or flow of test ready-mixed concrete produced by the mixer, and acquiring test input information including the test vibration information obtained from the sensor's detection results and image information of the test ready-mixed concrete. A method for predicting the quality of ready-mixed concrete, comprising constructing a prediction model based on the input information for the test and the measured values ​​of the quality of the ready-mixed concrete for the test, which are associated with the input information for the test.

2. The method for predicting the quality of ready-mixed concrete according to Claim 1, wherein the sensor detects the acceleration, velocity, angular velocity, position information, displacement, or both acceleration and angular velocity of the hopper.

3. The method for predicting the quality of ready-mixed concrete according to claim 1, wherein the acquisition step includes obtaining image information of the ready-mixed concrete by imaging the ready-mixed concrete in the hopper.

4. The acquisition process described above is: To obtain a reference time corresponding to the time when the ready-mixed concrete was discharged from the hopper, The method for predicting the quality of ready-mixed concrete according to claim 1, comprising acquiring time-series information indicating the magnitude of vibration of the hopper during the period from a time before the reference time to a time after the reference time, as vibration information.

5. The method for predicting the quality of ready-mixed concrete according to claim 1, wherein the vibration information includes information indicating the magnitude of vibration of the hopper in a direction intersecting the side wall of the hopper, and information indicating the magnitude of vibration of the hopper in a direction along the side wall of the hopper.

6. While the batch process, which includes mixing in the mixer, discharging from the mixer to the hopper, and loading from the hopper onto the transport vehicle, is being repeated, the acquisition process and the prediction process are being executed. In the acquisition step, the input information is acquired for each batch process. The method for predicting the quality of ready-mixed concrete according to any one of claims 1 to 5, wherein the prediction step predicts the quality of the ready-mixed concrete for each batch process.

7. The method for predicting the quality of ready-mixed concrete according to any one of claims 1 to 5, wherein the prediction step predicts both the slump and the slump flow of the ready-mixed concrete.

8. The method for predicting the quality of ready-mixed concrete according to any one of claims 1 to 5, wherein the prediction step predicts the amount of air in the ready-mixed concrete.

9. The mixing process involves mixing concrete materials in a mixer, An acquisition step to acquire input information including vibration information indicating the magnitude of vibration caused by the dropping or flow of the ready-mixed concrete produced in the mixing step, and image information of the ready-mixed concrete. A prediction model pre-built by machine learning to output quality information indicating the quality of the ready-mixed concrete in response to the input of the aforementioned input information, and a prediction step that predicts the quality of the ready-mixed concrete based on the input information acquired in the acquisition step, The process includes a construction step for constructing the prediction model before performing the acquisition step and the prediction step, The acquisition step includes obtaining the vibration information from the detection result of a sensor located below the mixer and attached to a hopper capable of containing the ready-mixed concrete, The sensor detects a physical quantity indicating the vibration of the hopper without the use of a medium, The vibration information is information indicating the magnitude of vibration of the hopper when the ready-mixed concrete is discharged from the hopper. The aforementioned construction process is, The method involves obtaining test vibration information indicating the magnitude of vibrations caused by the dropping or flow of test ready-mixed concrete produced by the mixer, and acquiring test input information including the test vibration information obtained from the sensor's detection results and image information of the test ready-mixed concrete. A method for producing ready-mixed concrete, comprising constructing the predictive model based on the input information for the test and the measured quality values ​​of the ready-mixed concrete for the test, which are associated with the input information for the test.

10. A mixer for mixing concrete materials, A hopper is positioned below the mixer and capable of containing the ready-mixed concrete, A vibration sensor attached to the hopper for detecting the magnitude of vibrations caused by the dropping or flow of ready-mixed concrete produced in the mixer, An image sensor for capturing images of the ready-mixed concrete, An information acquisition unit acquires input information including vibration information indicating the magnitude of vibration caused by the falling or flowing of the ready-mixed concrete and image information of the ready-mixed concrete, based on the detection results from the vibration sensor and the imaging results from the image sensor. A predictive model pre-built by machine learning to output quality information indicating the quality of the ready-mixed concrete in response to the input of the aforementioned input information, and a quality prediction unit that predicts the quality of the ready-mixed concrete based on the input information acquired by the information acquisition unit, The system includes a model building unit for constructing the aforementioned prediction model. The vibration sensor detects a physical quantity indicating the vibration of the hopper without the use of a medium, The vibration information is information indicating the magnitude of vibration of the hopper when the ready-mixed concrete is discharged from the hopper. The aforementioned model building unit, Test vibration information indicating the magnitude of vibration caused by the dropping or flow of test ready-mixed concrete produced by the mixer, and test input information including the test vibration information obtained from the detection results of the vibration sensor and image information of the test ready-mixed concrete, A ready-mix concrete manufacturing system that constructs a predictive model based on the input information for the test and the measured quality values ​​of the ready-mix concrete for the test, which are associated with the input information for the test.

Images