JP7326575B1 - Aggregate Quality Prediction Method, Ready-mixed Concrete Manufacturing Method, and Ready-mixed Concrete Manufacturing System - Google Patents

Abstract

An object of the present invention is to stabilize the quality of ready-mixed concrete.
Kind Code: A1 A method for predicting the quality of an aggregate is machine learning based on input information including image data obtained by imaging the aggregate and a correct value of the coarse grain ratio of the aggregate associated with the input information. includes a preparatory step of constructing a prediction model that outputs a value indicating the coarse grain ratio of the aggregate according to the input of the input information, and image data obtained by imaging the aggregate to be evaluated, the input information and a prediction step of inputting the evaluation information to the prediction model and obtaining a first prediction value output from the prediction model regarding the coarse grain ratio of the aggregate. include.
[Selection drawing] Fig. 3

Description

The present disclosure relates to an aggregate quality prediction method, a ready-mixed concrete manufacturing method, and a ready-mixed concrete manufacturing system.

Patent Document 1 discloses a method for estimating the particle size distribution of soil. Patent Literature 2 discloses an aggregate quality estimation method.

JP 2021-117625 A JP 2021-135199 A

The present disclosure provides an aggregate quality prediction method, a ready-mixed concrete production method, and a ready-mixed concrete production system that are useful for stabilizing the quality of ready-mixed concrete.

[1] By machine learning based on input information including image data obtained by imaging an aggregate and the correct value of the coarse grain ratio of the aggregate associated with the input information, according to the input information obtaining evaluation information corresponding to the input information, including image data obtained by imaging the aggregate to be evaluated; and a predicting step of inputting the evaluation information into the predictive model and obtaining a first predicted value output from the predictive model regarding the coarse grain rate of the aggregate.
[2] Aggregate quality prediction according to [1] above, wherein in the preparation step, the prediction model is constructed so as to output only a value indicating the coarse particle rate of the aggregate according to the input information. Method.
[3] The preparation step includes constructing the prediction model by machine learning based on the correct values of the mass ratio of each of the plurality of particle size categories of the aggregate, which are associated with the input information, and the preparation In the step, the prediction model is constructed so as to further output a value indicating a mass ratio for each class included in the plurality of particle size classes according to the input information, and the prediction step converts the evaluation information to the Inputting into a prediction model to further obtain a second prediction value output from the prediction model with respect to the mass ratio for each class, converting the second prediction value into a coarse grain rate of aggregate, The aggregate quality prediction method according to [1] above.
[4] The preparation step updates the intermediate model generated in the process of building the prediction model based on the first error regarding the coarse grain ratio of the aggregate and the second error regarding the mass ratio for each class. and in the preparing step, the intermediate model is updated using an error corrected to reduce the difference between the first error and the second error. Aggregate quality prediction method.
[5] The aggregate quality prediction method according to any one of [1] to [4] above, wherein the preparation step includes constructing the prediction model by machine learning using a convolutional neural network.
[6] In the acquisition step, in a state in which the aggregates to be evaluated are stacked, an image of a range including the aggregates to be evaluated is captured from a position distant from the aggregates to be evaluated, and the evaluation information is obtained. The aggregate quality prediction method according to any one of [1] to [5] above, including obtaining.
[7] The preparation step includes preparing aggregates for learning so as to include grains in two or more particle size categories, and the input information is an image obtained by imaging the aggregates for learning. The aggregate quality prediction method according to any one of [1] to [6] above, including data.
[8] A production step of kneading materials containing aggregates to produce ready-mixed concrete; In the quality prediction step, the input information is input by machine learning based on input information including image data obtained by imaging the aggregate and the correct value of the coarse grain ratio of the aggregate associated with the input information. Acquisition of evaluation information corresponding to the input information, including a preparation step of constructing a prediction model that outputs a value indicating the coarse particle ratio of the aggregate according to the above, and image data obtained by imaging the aggregate to be evaluated. and a prediction step of inputting the evaluation information to the prediction model and obtaining a first prediction value output from the prediction model regarding the coarse particle rate of the aggregate. .
[9] A production device for kneading materials containing aggregates to produce ready-mixed concrete; The prediction device uses machine learning based on input information including image data obtained by imaging aggregates and the correct value of the coarse grain ratio of aggregates associated with the input information to input the input information. Obtaining evaluation information corresponding to the input information, including a preparation step of constructing a prediction model that outputs a value indicating the coarse particle ratio of the aggregate according to the evaluation, and image data obtained by imaging the aggregate to be evaluated. and a predicting step of inputting the evaluation information to the predictive model and obtaining a first predicted value output from the predictive model regarding the coarse grain ratio of the aggregate. , ready-mixed concrete manufacturing system.

According to the present disclosure, there are provided an aggregate quality prediction method, a ready-mixed concrete production method, and a ready-mixed concrete production system that are useful for stabilizing the quality of ready-mixed concrete.

FIG. 1 is a schematic diagram showing an example of a ready-mixed concrete production system. FIG. 2 is a schematic diagram showing an example of the hardware configuration of the control device. FIG. 3(a) is a flowchart showing an example of processing executed in the learning phase. FIG. 3(b) is a flow chart showing an example of the process executed in the evaluation phase. FIG. 4A is a diagram schematically showing an example of prediction calculation using a prediction model. FIG. 4B is a cross-sectional view schematically showing an example of how image data for learning is acquired. FIG. 5 is a diagram showing an example of image data for learning. FIGS. 6(a) and 6(b) are graphs showing examples of comparison results between predicted values and correct values. FIGS. 7(a) and 7(b) are graphs showing examples of comparison results between predicted values and correct values. FIGS. 8(a) and 8(b) are graphs showing examples of comparison results between predicted values and correct values. FIGS. 9A and 9B are graphs showing examples of comparison results between predicted values and correct values. FIGS. 10(a) and 10(b) are graphs showing examples of comparison results between predicted values and correct values. FIGS. 11(a) and 11(b) are graphs showing examples of comparison results between predicted values and correct values. FIG. 12 is a graph showing an example of comparison results between predicted values and correct values. FIG. 13 is a graph showing an example of comparison results between predicted values and correct values. FIG. 14 is a graph showing an example of a comparison result between predicted values and correct values.

An embodiment will be described below with reference to the drawings. In the description, the same reference numerals are given to the same elements or elements having the same function, and overlapping descriptions are omitted.

FIG. 1 schematically shows a ready-mixed concrete manufacturing system according to one embodiment. A production system 1 shown in FIG. 1 is a system for producing ready-mixed concrete. The production system 1 kneads concrete materials to produce (generate) ready-mixed concrete. The manufacturing system 1 has a function of predicting the quality of concrete materials in addition to the function of manufacturing ready-mixed concrete.

Concrete materials used by the manufacturing system 1 include cement, admixtures, coarse aggregates, fine aggregates, water, admixtures, and the like. Coarse aggregates include, for example, gravel such as mountain gravel, land gravel, river gravel, and sea gravel; crushed stone; blast furnace slag aggregate, ferronickel slag aggregate, electric furnace oxidation slag aggregate, coal gasification slag aggregate slag coarse aggregate; lightweight coarse aggregate such as natural lightweight aggregate, by-product lightweight aggregate, and artificial lightweight aggregate; recycled coarse aggregate; Coarse aggregates may include, for example, crushed rock or crushed limestone. Examples of fine aggregates include sand such as mountain sand, land sand, river sand, and sea sand; crushed sand; Slag fine aggregate such as slag aggregate; lightweight fine aggregate such as natural lightweight aggregate, by-product lightweight aggregate, and artificial lightweight aggregate; recycled fine aggregate; be done. In addition, rock types of crushed stone and crushed sand include igneous rocks such as granite, diorite, gabbro, pyrite, diabase, rhyolite, andesite, basalt, and serpentinite; conglomerate, sandstone, shale, and slate; , sedimentary rocks such as tuff; metamorphic rocks such as gneiss and crystalline schist; others include silica, limestone, domarite, and peridotite. In the present disclosure, coarse aggregate and fine aggregate may be collectively referred to as "aggregate." In this case, "aggregate" means coarse aggregate, fine aggregate, or both coarse aggregate and fine aggregate.

The manufacturing system 1 loads the manufactured ready-mixed concrete onto the transport vehicle C. After the ready-mixed concrete is loaded, the transport vehicle C transports the ready-mixed concrete to a site (for example, a construction site) where the ready-mixed concrete is used. Examples of the transport vehicle C include an agitator vehicle (mixer vehicle) and a dump truck. The manufacturing system 1 may manufacture ready-mixed concrete from a concrete material so as to meet the target quality (required quality) set for each site. In one example, the target quality of ready-mixed concrete includes a target value for slump or slump flow. The manufacturing system 1 includes, for example, a material storage space 2, a transport device 8, a manufacturing device 10, an imaging device 50, and a control device 60.

The material storage space 2 is a place for storing concrete materials. Material storage 2 includes a plurality of silos 4 . The plurality of silos 4 are containers that store at least part of the concrete material for each type of material. The plurality of silos 4 includes a silo 4 storing coarse aggregate, a silo 4 storing fine aggregate, and a silo 4 containing cement.

The transporter 8 is a device that transports the concrete material stored in the plurality of silos 4 to the manufacturing device 10 . The conveying device 8 includes, for example, a belt conveyor for conveying concrete material. The width of the belt conveyor may be about 30 cm to 160 cm. The conveying device 8 may convey concrete materials for each type of material at different timings. In one example, based on an operation instruction from the control device 60 , a specific material among various concrete materials is transferred to the transport device 8 and transported to the manufacturing device 10 .

The production apparatus 10 is an apparatus for kneading concrete materials containing aggregates to produce ready-mixed concrete. The manufacturing apparatus 10 operates based on operation instructions from the control apparatus 60 . The manufacturing apparatus 10 includes, for example, a storage bottle 12, a weighing bottle 14, a collection hopper 16, a mixer 20, and a loading hopper 30.

The storage bottle 12 temporarily stores various concrete materials. Various concrete materials are transported (conveyed) from the material storage space 2 to the storage bottles 12 by the transport device 8 . The storage bins 12 are configured to individually store various concrete materials. Hereinafter, "concrete material" may be simply referred to as "material". Various materials stored in the storage bottle 12 are supplied to the weighing bottle 14 as required.

A weighing bottle 14 is arranged below the storage bottle 12 . The weighing bottle 14 operates based on operation instructions from the control device 60 and weighs various materials individually. When the weighing bottle 14 detects the target amount of material instructed by the control device 60 , the weighing bottle 14 supplies the material to the collecting hopper 16 . An admixture may be mixed with the water as it is supplied to the weighing bottle 14 . The collection hopper 16 is arranged below the weighing bottle 14 . The collecting hopper 16 collects various materials discharged from the weighing bottles 14 and supplies the collected various materials to the mixer 20 . If the collecting hopper 16 is not provided, various ingredients are supplied from the weighing bottle 14 to the mixer 20 .

The mixer 20 is arranged below the gathering hopper 16 . The mixer 20 is a device that mixes concrete materials. The mixer 20 manufactures ready-mixed concrete by kneading (kneading) aggregate, cement, water, an admixture, and the like. From the bottom of mixer 20 fresh concrete is discharged into loading hopper 30 . The loading hopper 30 is arranged below the mixer 20 and temporarily stores ready-mixed concrete. The loading hopper 30 supplies the transport vehicle C with ready-mixed concrete temporarily stored.

The manufacturing apparatus 10 described above is an example of a ready-mixed concrete manufacturing apparatus, and the ready-mixed concrete manufacturing apparatus may be configured in any manner as long as it is possible to mix concrete materials and manufacture ready-mixed concrete. good. For example, the manufacturing apparatus 10 may not include the gathering hopper 16 and various materials weighed in the weighing bottle 14 may be supplied from the weighing bottle 14 to the mixer 20 .

The imaging device 50 is a device (camera) capable of imaging aggregate, which is one of concrete materials. The imaging device 50 generates image data by imaging the aggregate. The imaging device 50 is, for example, a digital camera or a video camera that captures an imaging range based on visible light. The image data generated by the imaging device 50 is used to predict aggregate quality. Prediction of aggregate quality will be described later. Note that the imaging device 50 may generate a moving image of an imaging target, and still images included in the moving image may be acquired as image data.

The imaging device 50 is arranged so as to capture an image of a range including the aggregate to be imaged from a position distant from the aggregate to be imaged. The imaging device 50 may capture an image of a range including aggregates to be imaged from above. The imaging device 50 may capture an image of a range including aggregates in a piled state. Aggregates in a stacked state refer to aggregates in a state in which some grains contained in the aggregate overlap other grains when viewed vertically from above. The imaging device 50 may be arranged so as to be able to image the aggregate being transported by the transporting device 8 . The image capturing device 50 may capture an image of the aggregate being transported while the belt conveyor included in the transport device 8 is stopped or moving. Note that the imaging device 50 may capture an image of the aggregates loaded on the transport truck or the transport ship when receiving the aggregates. The imaging device 50 may be arranged in an imaging room in which a floor or a rubber sheet is prepared, and may image an aggregate placed on the floor or the like.

The control device 60 is a device that controls the manufacturing device 10 . The control device 60 is configured by one or more computers. When control device 60 is composed of a plurality of computers, these computers are communicatively connected to each other. The control device 60 controls the manufacturing device 10 according to the set operating conditions. At least part of the operating conditions may be determined by instructions from an operator such as a worker.

An input/output device 62 may be connected to the controller 60 (see FIG. 2). The input/output device 62 is a device for inputting information indicating an instruction from a worker or the like to the control device 60 and for outputting information from the control device 60 to the worker or the like. The input/output device 62 may include a keyboard, operation panel, or mouse as an input device, and may include a monitor (for example, a liquid crystal display) as an output device. The input/output device 62 may be a touch panel in which an input device and an output device are integrated. The control device 60 and the input/output device 62 may be integrated.

It is known that the quality of ready-mixed concrete produced by the production apparatus 10 is affected by the quality of the concrete material. Specifically, the particle size of the aggregates used to make the ready-mixed concrete affects the fluidity (eg, slump and slump flow) of the ready-mixed concrete. For example, if the unit water content is the same, the larger the grain size of the aggregate, the smaller the surface area of the aggregate, the smaller the water retention capacity, and the greater the fluidity of ready-mixed concrete. Conversely, the smaller the particle size of the aggregate, the greater the surface area of the aggregate, the greater the water retention capacity, and the less the fluidity of ready-mixed concrete. Thus, in order to stabilize the quality including fluidity of ready-mixed concrete, it is necessary to control the particle size of the aggregate. The particle size of the aggregate can be represented by the coarse particle ratio, and in order to stabilize the quality of ready-mixed concrete, for example, it is necessary to control whether the coarse particle ratio of the aggregate is within the desired range. .

In addition to controlling the production apparatus 10 , the control device 60 may be configured to predict the quality of the aggregate used in the production of ready-mixed concrete by the production apparatus 10 . In this case, the control device 60 constitutes a quality prediction device that predicts the quality of the aggregate. The control device 60 functioning as a quality prediction device predicts the coarse particle rate of at least a portion of the aggregate used by the manufacturing device 10 .

The control device 60 performs machine learning based on at least input information including image data obtained by imaging the aggregate and the correct value of the coarse particle ratio of the aggregate associated with the input information. It is configured to execute a preparatory step of constructing a predictive model that outputs a value indicating the coarseness of the aggregate according to the input. In addition, the control device 60 includes an acquisition step of acquiring evaluation information corresponding to the input information including image data obtained by imaging the aggregate to be evaluated, and inputting the evaluation information into the prediction model to obtain the aggregate obtaining a value output from the prediction model for the coarse grain fraction of the .

The control device 60 includes, for example, an operation control unit 68, a model building unit 72, a model holding unit 74, a quality prediction unit 76, and an output unit 78 as functional components (hereinafter referred to as "functional modules"). and The processes executed by these functional modules correspond to the processes executed by control device 60 . The operation control unit 68 is a functional module that controls the manufacturing apparatus 10 to manufacture ready-mixed concrete according to the operating conditions. A worker or the like may adjust (change) the operating conditions according to the target quality of ready-mixed concrete.

The model construction unit 72 is a functional module that constructs a model (hereinafter referred to as "prediction model M") for predicting the coarse particle ratio of aggregate. The prediction model M is a model that outputs at least a value indicating the coarse particle ratio of the aggregate according to input information including image data obtained by imaging the aggregate. The model construction unit 72 constructs the prediction model M by machine learning based on the input information and the correct answer value of the aggregate coarse particle ratio associated with the input information.

Machine learning is a method of autonomously discovering laws or rules by repeatedly learning a machine (computer) based on given information. The predictive model M can be constructed using algorithms and data structures. The prediction model M is realized, for example, using a neural network, which is an information processing model imitating the mechanism of human cranial nerves. A specific machine learning algorithm that is performed when constructing the prediction model M is not particularly limited. A neural network has an input layer, one or more intermediate layers, and an output layer. By including one or more intermediate layers, a more complicated prediction model M can be constructed and the prediction accuracy can be improved.

The model construction unit 72 may construct the prediction model M by machine learning using a convolutional neural network (CNN). The model construction unit 72 may construct the prediction model M by machine learning based on the correct value of the coarse particle ratio of the aggregate and the correct value of the mass ratio of each of the plurality of particle size categories of the aggregate. The prediction model M constructed by the model construction unit 72 outputs a value indicating the coarse grain ratio of the aggregate and a value indicating the mass ratio for each of the plurality of particle size divisions in accordance with the input information. may That is, the predictive model M may output one value indicating the coarse particle rate of the aggregate, and a plurality of values each indicating the mass ratio for a plurality of particle size categories.

The model building unit 72 predicts the coarse grain ratio by performing machine learning using, for example, data given as input for machine learning and correct data (correct values such as the coarse grain ratio) output from machine learning. You may construct|assemble the prediction model M for doing autonomously. Inputs for machine learning are various datasets of input information, including images of aggregates. The various data sets of the input information differ from each other at least in the image data included in the input information. The output of machine learning is data (numerical values) indicating the coarse particle ratio of the aggregate and the mass ratio for each particle size category. The model building unit 72 repeats the model that outputs the predicted value of the aggregate coarse particle rate and the mass ratio for each particle size classification using multiple combinations of input information data sets and correct values such as coarse particle rates. learn systematically.

The stage of autonomously constructing the prediction model M corresponds to the learning phase. The learning phase may be performed prior to the production phase in which ready-mixed concrete is produced, or may be performed in the early stages of the production phase. The predictive model M, which is a trained model, may be portable between computers. Therefore, the prediction model M constructed in the control device 60 may be used in another manufacturing system different from the manufacturing system 1 .

The model holding unit 74 is a functional module that holds (stores) the prediction model M. FIG. The quality prediction unit 76 is a functional module that predicts the coarse particle rate of aggregate. Hereinafter, the aggregate for which the coarse grain ratio is to be predicted is referred to as "evaluation target aggregate". The quality prediction unit 76 acquires information (hereinafter referred to as “evaluation information”) corresponding to the input information, including image data obtained by imaging aggregates to be evaluated. While the input information is associated with the correct value of the coarse particle ratio, the evaluation information is information in which the aggregate coarse particle ratio is unknown. The stage in which the quality prediction unit 76 predicts the coarse grain ratio of the aggregate using the prediction model M corresponds to the evaluation phase.

The quality prediction unit 76 inputs the evaluation information to the prediction model M, and obtains a prediction value (first prediction value) output from the prediction model M regarding the coarse particle rate of the aggregate. The quality prediction unit 76 also inputs the evaluation information to the prediction model M, and further acquires a prediction value (second prediction value) output from the prediction model M regarding the mass ratio for each particle size class. That is, the quality prediction unit 76 acquires, from the prediction model M, a plurality of prediction values (second prediction values) each indicating a mass ratio value for a plurality of particle size classes. The quality prediction unit 76 may convert the predicted value (a plurality of predicted values) regarding the mass ratio for each particle size division into the coarse particle ratio. By predicting the coarse grain ratio using the prediction model M by two different methods, the reliability of the predicted value of the coarse grain ratio can be evaluated.

The output unit 78 is a functional module that outputs at least the predicted value of the coarse particle ratio of the aggregate output from the prediction model M as the prediction result of the coarse particle ratio. The output unit 78 may further output a value obtained by converting the predicted value relating to the mass ratio for each particle size class, which is output from the prediction model M, into the coarse particle ratio as the prediction result of the coarse particle ratio. The output unit 78 may notify the operator or the like of the prediction result by outputting the prediction result of the coarse particle ratio to an output device (for example, a monitor) of the input/output device 62 .

As shown in FIG. 2, controller 60 has circuitry 91 . Circuitry 91 includes one or more processors 92 , memory 93 , storage 94 , input/output ports 95 and timers 96 . The storage 94 has a computer-readable storage medium such as a non-volatile semiconductor memory. The storage 94 controls the program for causing the control device 60 to control the manufacturing device 10 according to a preset control procedure, and predicts the coarse particle rate of the aggregate according to a preset prediction procedure. A program to be executed by the device 60 is stored. For example, the storage 94 stores programs for configuring each functional module described above.

The memory 93 temporarily stores the program loaded from the storage medium of the storage 94 and the calculation result by the processor 92 . The processor 92 configures each functional module of the control device 60 by executing the program in cooperation with the memory 93 . The input/output port 95 inputs/outputs electrical signals to/from the manufacturing apparatus 10, the imaging apparatus 50, the input/output device 62, and the like according to instructions from the processor 92. FIG. The timer 96 measures the elapsed time by, for example, counting reference pulses of a constant cycle. It should be noted that the circuit 91 is not necessarily limited to having each function configured by a program. For example, the circuit 91 may configure at least part of its functions by a dedicated logic circuit or an ASIC (Application Specific Integrated Circuit) integrated with this.

[Method for producing ready-mixed concrete]
Next, an example of a fresh concrete manufacturing method executed in the manufacturing system 1 will be described. A method for producing ready-mixed concrete includes a step of producing ready-mixed concrete and a step of predicting the quality of aggregate. The step of predicting the aggregate quality may be performed in a period that overlaps at least part of the period in which the step of producing ready-mixed concrete is performed.

The process of manufacturing ready-mixed concrete (hereinafter referred to as the "manufacturing process") includes, for example, a conveying process, a weighing process, a charging process, a kneading process, a discharging process, and a loading process. In the transporting process, various concrete materials are transported to the storage bottles 12 by the transporting device 8, and the various materials are supplied to the storage bottles 12 individually. In the weighing process, various materials are individually supplied from the storage bottle 12 to the weighing bottle 14 , and the various materials are weighed in the weighing bottle 14 . In the weighing process, each material is discharged to the collection hopper 16 when the measured amount reaches a predetermined set amount.

In the charging step, after all kinds of materials are gathered in the gathering hopper 16 , the materials in the gathering hopper 16 are charged (supplied) to the mixer 20 . In the kneading step, a plurality of types of concrete materials are kneaded in the mixer 20 . In the discharge step, the ready-mixed concrete is discharged from the mixer 20 to the loading hopper 30 after kneading of the concrete material in the mixer 20 is completed. In the loading process, the ready-mixed concrete discharged into the loading hopper 30 is loaded onto the carrier C. As shown in FIG.

The process of predicting the quality of aggregates (hereinafter referred to as "quality prediction process") includes a model building process in the learning phase and a quality evaluation process in the evaluation phase. In the quality prediction process (quality prediction method), the model building process is executed before the quality evaluation process. An example of the model construction process and an example of the quality evaluation process will be described below. An example will be described in which the coarse particle ratio of coarse aggregate is predicted as the quality of the aggregate, and the input information includes only image data of the aggregate. Further, a specific example in which the prediction model M outputs the predicted value of the coarse particle rate and the predicted value of the mass ratio for each particle size division will be described.

(Model construction process)
FIG. 3(a) is a flow chart showing an example of a series of processes executed in the model construction process. This model building process may be executed before the manufacturing process in the manufacturing apparatus 10 is executed, or at an initial stage when the manufacturing process is started. In this model building process, a part of the coarse aggregate used in the production of ready-mixed concrete in the production apparatus 10 is extracted, and part of the coarse aggregate may be used as a "coarse aggregate for learning". .

In the model construction process, step S11 is executed first. In step S11, for example, a worker or the like prepares learning data for performing machine learning. The learning data includes various image data (input information) obtained by imaging coarse aggregate for learning, and correct data associated with each of the various image data. FIG. 4(a) schematically shows a calculation method for predicting the coarse grain ratio from the image Im, which is the input information, using the prediction model M. As shown in FIG.

In the present disclosure, as shown in FIG. 4( a ), the prediction model M outputs two types of information, the mass ratio for each particle size division and the coarse particle ratio, which is referred to as “multitasking”. In addition, the output of either one of the mass ratio for each particle size division and the coarse particle ratio by the prediction model M is referred to as a "single task". FIG. 4(a) exemplifies calculation methods for predicting the coarse grain ratio using two types of methods using a prediction model M that performs multitasking.

In preparation of images for learning, for example, various coarse aggregates having known mass ratios and coarse particle ratios for each of a plurality of particle size categories are prepared by an operator or the like. The various coarse aggregates do not mean different types of coarse aggregates (for example, crushed stone), but coarse aggregates with different particle sizes. Then, as shown in FIG. 4B, for example, various coarse aggregates for learning are individually imaged by the camera 52 while being stacked on the sheet 8a for measurement. Thereby, image data for learning is obtained for each of various coarse aggregates. The camera 52 may be a digital camera, and is provided to capture an image of a range including the coarse aggregate from a predetermined position (for example, a predetermined position above) away from the coarse aggregate on the sheet 8a for measurement. .

The sheet 8a may be a sheet shaped like a belt conveyor of the conveying device 8, such as a rubber sheet. The width of the sheet 8a may be about half the width of the belt conveyor of the transport device 8. FIG. The amount of aggregate placed on the sheet 8a is not particularly limited, but may be about 1 kg to 10 kg. FIG. 4(b) schematically illustrates how the aggregate is deposited on the sheet 8a, and shows deposition states when the amounts of aggregate are 2 kg, 5 kg, and 8 kg, respectively.

In one example, various coarse aggregates are prepared under the conditions shown in Table 1 below, and images obtained by imaging the coarse aggregates are acquired. The mass ratio for each particle size category and the coarse particle rate in various coarse aggregates prepared under the conditions shown in Table 1 are known, and the known values are correct data in machine learning.

The coarse particle ratio (F.M.) is determined, for example, by the following formula (1) defined in the Concrete Standard Method. In addition, nine sieves excluding 80 mm in the formula (1) are used in the measurement prescribed in the "Building Standard Specification and Commentary JASS5 Reinforced Concrete Work". The coarse grain rate predicted in the present disclosure may be defined by any method.

In Table 1, it means that coarse aggregates with 10 levels of grain size (coarse grain ratio) are prepared. Category I represents the particle size category for grains smaller than 2.5 mm. Section I is divided into granules that pass through a 2.5 mm sieve. Category II represents the particle size category of grains greater than or equal to 2.5 mm and smaller than 5 mm. Section II separates the granules that remain on the 2.5 mm sieve but pass through the 5 mm sieve. Class III represents the particle size class of grains greater than or equal to 5 mm and smaller than 10 mm. Section III separates the grains that pass through the 10 mm sieve but remain on the 5 mm sieve. Class IV represents the particle size class of grains greater than or equal to 10 mm and smaller than 20 mm. Section IV separates the granules that pass through the 20 mm sieve but remain on the 10 mm sieve.

Regarding level 1, the coarse grain ratio is obtained as follows from the mass ratio for each particle size category (mass ratio for each of categories I to IV). For level 1 coarse aggregate, the percentage of sample mass that stays on the 80 mm, 40 mm, and 20 mm sieves is 0, respectively, and the percentage of sample mass that stays on the 10 mm sieve is 85, which stays on the 5 mm sieve. The mass percentage of the sample is 100 (=15+85). Also, since all grains remain on the 5 mm sieve, the percentage of the mass of the sample that remains on the 2.5 mm, 1.2 mm, 600 μm, 300 μm, and 150 μm sieves is 100 each. Therefore, the coarse grain ratio is calculated as in the following formula (2).

For Levels 2 to 10, the coarse grain ratio is obtained in the same manner as the above calculation for Level 1. The aggregate conditions for learning shown in Table 1 are only examples, and the number of levels, the coarse particle ratio at each level, the mass ratio for each particle size classification, and the method of setting the particle size classification etc., may be arbitrarily determined. Particle sizes in multiple categories, including the smallest category (<150 μm), may be combined into one particle size category, such as the multiple particle size categories shown in Table 1. Particle sizes in a certain range (for example, 20 to 40 mm) may not be included in multiple particle size ranges. Also, although different from the setting method in Table 1, particle sizes below a certain range (for example, 1.2 to 2.5 μm) may not be included in a plurality of particle size ranges.

FIG. 5 shows images T1 to T10 obtained by imaging learning aggregates prepared according to conditions of levels 1 to 10 in Table 1, respectively. Level n (n is an integer from 1 to 10) corresponds to image Tn. For example, an image T1 is obtained by imaging learning aggregates prepared according to Level 1. The aggregate for learning according to each level n contains grains of a plurality of grain size divisions. In this way, in step S11, aggregates for learning are prepared so as to include grains of two or more particle size divisions, and input information including image data obtained by imaging the aggregates for learning is prepared. may

For each level n, a plurality of images Tn having different states of learning aggregates (more specifically, deposition states) may be prepared. In one example, one image Tn is acquired by performing the first imaging in a state in which the aggregate for learning is placed on the sheet 8a for measurement. Then, the learning aggregate is temporarily removed from the sheet 8a, and the same learning aggregate is placed again on the sheet 8a so that the accumulated state of the many grains is different from that at the time of the first imaging. is placed. Then, the second imaging is performed. A plurality of images Tn may be prepared for each level n by repeating the above-described imaging by a worker or the like.

A worker or the like may input the prepared image data and correct answer data to the control device 60 via the input/output device 62 . One data set is composed of one image Tn and the correct data of the mass ratio and the coarse grain ratio at the level n associated with the image Tn. A worker or the like prepares a plurality of data sets as learning data.

Returning to FIG. 3A, after step S11 is executed, step S12 is executed. In step S12, for example, the model construction unit 72 of the control device 60 performs machine learning using the image data prepared in step S11 and the correct data linked to the image data, thereby creating the prediction model M. To construct. The predictive model M may be constructed to output four mass fraction values for the four grain size classes and one value for the coarse grain fraction (see also FIG. 4(a)). By converting the values of the four mass ratios output from the prediction model M as in the above equation (2), the predicted value of the coarse grain ratio is obtained.

Thus, the prediction model M may be a model that performs multitasking. If the prediction model M is a multi-tasking model, different approaches yield two coarse grain rate predictions. In this case, for example, it is possible to evaluate the reliability of the predicted value of the coarse grain ratio. In one example, when the difference between the two predicted values for the coarse grain ratio is small, it can be evaluated that the prediction result is highly reliable.

An example of the prediction model M will be described below using simplified formulas for easy understanding. The prediction model M constructed by the model construction unit 72 can be simply represented by, for example, the following equations (3) and (4).

In equation (4), Y represents one of y1 to y5, and prediction model M calculates equation (4) for each of y1 to y5. y1 to y5 are output data representing the mass ratio and the coarse particle ratio for each particle size division. Where y1, y2, y3, and y4 are the mass ratio output values for each of the four grain size classes, y5 is the coarse grain ratio output value. N is an integer of 2 or more and represents the number of input data. The input data includes at least the value of each pixel in the aggregate image. wi is a weight (coefficient) and b is a bias term (coefficient). f(U) represents an activation function. The activation function is a linear function (identity function) or polynomial, absolute value, step function, sigmoid function, hardsigmoid function, logsigmoid function, softmax function, logsoftmax function, softmin function, softplus function, softsign function, tanh function, tanhShrink function, hardtanh function, tanhexp function, ReLU function, ReLU6 function, Leaky-ReLU function, PReLU function, ELU function, SELU function, CELU function, Swith function, Mish function, ACON function, etc.

The model construction unit 72 constructs a prediction model M that outputs the values of y1 to y5 by determining the weight wi and the bias term b in Equation (4) for each of y1 to y5 using data for learning. . The model construction unit 72 determines weights wi for each of y1 to y5 by repeating the calculations represented by the following equations (5) and (6).

In equation (5), yj (j is any integer from 1 to 5) is the output of the model defined by the weight wi being updated, and tj is the correct value of the mass ratio or coarse grain ratio. be. E and Ej are errors representing the difference between the output yj and the correct value tj. In Equation (6), w' represents the updated weight w. By repeating the calculations of equations (5) and (6), the sum of the squared errors between the value of the output yj and the correct value tj when a certain value is input to the weight w is calculated, and by the gradient method, It means updating the value of the weight w so that the error is minimized. The number of repetitions of updating the weight w may be determined by a worker or the like. Gradient methods include gradient descent, stochastic gradient descent (SGD), mini-batch gradient descent, Adam, Nesterov's acceleration (NAG), AdaGrad, RMSprop, Adadelta, Adam, Adamax, AMSGRAD, and Momentum. be able to.

The explanation of the prediction model M using equations (3) to (6) is a general explanation of deep neural networks. Deep neural networks include, for example, convolutional neural networks, recurrent neural networks, and recurrent neural networks. In the aggregate quality prediction method according to the present embodiment, it is preferable to use, for example, a convolutional neural network. This is because convolutional neural networks have an advantage over conventional deep neural networks in updating weights when using input information including image data. In the case of a normal deep neural network, when a large amount of information is included, such as image data, the number of places where weights need to be calculated becomes extremely large. Convolutional neural networks, on the other hand, simplify weight updating because filters can be used to share weights. Thus, the use of a convolutional neural network is preferable in that it makes it possible to capture the features of an image more favorably.

After step S12 is executed, step S13 is executed. In step S13, for example, the model holding unit 74 stores the prediction model M constructed in step S12. With the above, the model construction process is completed.

(Quality evaluation process)
FIG. 3(b) is a flow chart showing an example of a series of processes executed in the quality evaluation process. This quality evaluation process is performed, for example, while the manufacturing process in the manufacturing apparatus 10 is being performed.

In this quality evaluation process, the controller 60 first executes step S21. In step S21, for example, the quality prediction unit 76 waits until the measurement timing. The measurement timing may be determined in advance at a certain time of the day, or may be a timing at which an instruction is received from an operator such as a worker. The control device 60 may perform control so that the coarse aggregate is placed on the transport device 8 from the material storage site 2 when the measurement timing comes. Note that the measurement timing may be set in accordance with the timing at which the coarse aggregate is being transported to the transporting device 8 .

Next, the control device 60 executes step S22. In step S<b>22 , for example, the quality prediction unit 76 causes the imaging device 50 to perform imaging of the evaluation target aggregate, and acquires image data from the imaging device 50 . As a result, the quality prediction unit 76 acquires evaluation information including image data obtained by imaging aggregates to be evaluated. In one example, the evaluation information includes only image data obtained by imaging aggregates to be evaluated. In step S22, for example, in a state in which the aggregates to be evaluated are stacked on the belt conveyor of the transport device 8, the imaging device 50 captures the aggregates to be evaluated from an upper position away from the aggregates to be evaluated. Take an image of the range.

Next, the control device 60 executes step S23. In step S23, the quality prediction unit 76, for example, based on the image data (evaluation information) acquired in step S22 and the prediction model M held by the model holding unit 74, the coarse grain ratio of the coarse aggregate to be evaluated to predict. In one example, the quality prediction unit 76 inputs the image data acquired in step S22 to the prediction model M, and acquires an output value relating to the coarse grain ratio that is directly output from the prediction model M. FIG. Also, the quality prediction unit 76 inputs the image data to the prediction model M, and obtains an output value relating to the mass ratio for each particle size classification, which is directly output from the prediction model M. FIG. Then, the quality prediction unit 76 indirectly acquires the predicted value of the coarse particle ratio by converting the output value related to the mass ratio for each particle size division into the coarse particle ratio.

Next, the control device 60 executes step S24. In step S24, for example, the output unit 78 outputs two prediction results regarding the coarse grain ratio obtained in step S23. In one example, the output unit 78 displays two prediction results regarding the coarse grain ratio on the monitor of the input/output device 62 . Thereby, an operator or the like can confirm the prediction result regarding the coarse particle rate. After confirming the prediction result of the coarse particle ratio, the workers, etc., take measures to maintain the quality of the ready-mixed concrete (for example, adjusting the particle size of the aggregate, correcting the composition, Temporary suspension of production, investigation of the cause, etc.) may be performed.

Thus, a series of processes for predicting the coarse grain ratio for one time is completed. After executing step S24, the control device 60 may repeat the series of processes of steps S21 to S24. In the above example, the coarse particle ratio is predicted for each measurement timing, but the aggregate being transported is continuously imaged by the transport device 8, and the aggregate used for the production of ready-mixed concrete is subjected to a 100% inspection. may be performed.

The above-mentioned quality prediction method can be used as a substitute for quality testing by sampling in addition to stabilizing quality by conducting 100% inspection of raw materials (materials) of ready-mixed concrete in the production process (manufacturing line) of ready-mixed concrete. is. Sampling quality tests include, for example, an acceptance inspection, which is an inspection when raw materials are received, and an in-process inspection, which is an inspection in each manufacturing process of ready-mixed concrete. Sampling quality tests are routinely performed to supply ready-mixed concrete of consistent quality. Conventionally, when the coarse aggregate is measured by a sampling quality test, for example, the coarse aggregate is sieved and the coarse aggregate is calculated from the mass ratio. If the aggregate quality prediction method according to the present embodiment is applied to routine quality tests, the time for quality prediction tests can be significantly reduced, and for example, the frequency of quality prediction tests per day can be increased. can be done. This makes it possible to supply ready-mixed concrete of more stable quality. When performing a quality test by sampling using the aggregate quality prediction method according to the present embodiment, for example, the sampled aggregate is deposited on a rubber sheet, and image data is generated therefrom by the imaging device 50. be able to. When the aggregate quality prediction method according to the present embodiment is used for acceptance inspection, for example, it is possible to attach the imaging device 50 to a transport truck or a transport ship and perform quality prediction during transportation.

[Verification of prediction method using prediction model]
Below, in order to verify the prediction method of the coarse grain rate using the prediction model M by machine learning using a convolutional neural network, using a data set with a known correct value, the prediction result by the prediction model M and the correct answer Comparison verification with the value was performed. As aggregates for learning, 10 levels of aggregates with different coarse particle rates were prepared according to the conditions in Table 1 above. For each level, 10 images were acquired by imaging 10 times in a room while the states of the aggregates on the sheet 8a for measurement were different from each other.

In addition, a total of 100 images (10 levels×10 times of imaging) were prepared according to the standard conditions shown in Table 2 below. As a result, 100 data sets in which images and correct values are associated in each data set were prepared as data for learning.

Using 80 data sets out of 100 data sets, the prediction model M was constructed, and 20 data sets were used for verification. Specifically, at each level, 8 data were used for model building and 2 data sets were used for validation. In the verification using 20 data sets, for each data set, two prediction results calculated using the image and the prediction model M, and the correct value such as the coarse grain ratio associated with the image. compared. The prediction model M was configured by a convolutional neural network in which six intermediate layers each including a convolutional layer and a pooling layer were provided, and four fully connected layers were provided. Also, the Huber function was used as the loss function. Note that in each dataset, the number of images was increased by the ImageAugmentation layer.

FIG. 6(a) is a graph showing the relationship between the predicted values of the mass ratio (%) and the corresponding correct values of the mass ratio (%) in Category III and Category IV in Table 1 above. In the graph shown in FIG. 6A, the horizontal axis is the correct value of the mass ratio, and the vertical axis is the predicted value of the mass ratio output from the prediction model M. As shown in FIG. The prediction results for Section III (5 mm to 10 mm) are plotted with Δ marks, and the prediction results for Section IV (10 mm to 20 mm) are plotted with ◯ marks. A solid line is drawn where the correct value and the predicted value match, and the closer the plotted result is to that line, the higher the prediction accuracy. From the results shown in FIG. 6(a), the prediction results are plotted in the vicinity of the above line, and it can be seen that the prediction model M can predict the mass ratio with good accuracy.

FIG. 6(b) is a graph showing the relationship between the predicted value of the coarse grain ratio output from the prediction model M and the correct value of the coarse grain ratio. In the graph shown in FIG. 6B, the horizontal axis is the correct value of the coarse grain ratio, and the vertical axis is the predicted value of the coarse grain ratio output from the prediction model M. As shown in FIG. Prediction results are plotted with squares. The solid line is the line where the correct value and the predicted value match, and the dashed upper limit line shifted vertically by +0.10 from the line and the lower limit line shifted vertically by −0.10. Drawn. A prediction result included in the region between the upper limit line and the lower limit line is defined as within the acceptable range. A range of ±0.10 was set as the allowable range with reference to the error of the coarse particle rate allowed in the existing ready-mixed concrete factories. All the prediction results using the prediction model M are within the allowable range, and it can be seen that the prediction model M can predict the coarse grain ratio with good accuracy.

In the following, the effects of various factors in preparing learning data on prediction by the prediction model M were verified. Specifically, at least one of the presence or absence of illumination, image size, amount of coarse aggregate, and moisture condition on the aggregate surface is changed from the standard conditions shown in Table 2, and the learning data is obtained. Got ready. Then, for each condition, a prediction model M was constructed from the corresponding learning data, and the prediction accuracy of the prediction model M was evaluated using the corresponding verification data.

The presence or absence of illumination is different between the graphs shown in FIGS. 7(a) and 7(b). Fig. 7(a) shows the comparison result of imaging with only room light (without lighting). ) are shown in FIG. 7(b). Furthermore, in any case with or without illumination, the amounts of coarse aggregate of 2 kg, 5 kg, and 8 kg were verified.

An absolute difference average value was calculated as an index showing how much the predicted value deviated from the correct value. The average absolute difference value is the arithmetic mean of the absolute values of the differences between the correct values and the predicted values in the 20 verification data sets. In the graphs shown in FIGS. 7(a) and 7(b), "direct prediction" represents the verification result of the output value of the coarse grain ratio directly output from the prediction model M. In the graphs shown in FIGS. "Indirect prediction" represents the verification result of the prediction value of the coarse grain ratio indirectly obtained by converting the output value of the mass ratio for each grain size classification output from the prediction model M.

The vertical axis represents the average absolute difference, and "Tg" represents the target value. When the average absolute difference value is equal to or less than the target value Tg (0.10), it can be evaluated that the prediction using the prediction model M is highly accurate. As shown in FIGS. 7(a) and 7(b), whether the amount of coarse aggregate is 2 kg, 5 kg, or 8 kg, or whether it is a direct prediction or an indirect prediction, , the average absolute difference was below the target value Tg regardless of the presence or absence of illumination. Therefore, it is considered that the presence or absence of illumination has little effect on the accuracy of prediction by the prediction model M.

The prediction methods are different between the graphs shown in FIGS. 8(a) and 8(b). The comparison result when using the direct prediction out of the predictions by the prediction model M is shown in FIG. (b). As shown in Figures 8(a) and 8(b), whether the amount of coarse aggregate is 2 kg, 5 kg, or 8 kg, and whether it is illuminated or not, the direct The mean absolute difference was less than or equal to the target Tg, regardless of whether prediction or indirect prediction was used. Therefore, it is considered that the selection result of direct prediction and indirect prediction has little influence on the accuracy of prediction by the prediction model M.

In each of the graphs shown in FIGS. 9(a) and 9(b), verification is performed by changing the amount of coarse aggregate. The comparison results when using direct prediction and changing the amount of coarse aggregate are shown in FIG. The results are shown in FIG. 9(b). As shown in FIGS. 9( a ) and 9 ( b ), regardless of the amount of coarse aggregate, whether direct prediction or indirect prediction, or with or without illumination, The average absolute difference was below the target value Tg. Therefore, it is considered that the amount of coarse aggregate has little effect on the accuracy of prediction by the prediction model M.

In each of the graphs shown in FIGS. 10A and 10B, verification is performed by changing the image size. FIG. 10(a) shows the comparison result when the direct prediction is used and the image size is changed, and FIG. 10(b) is the comparison result when the indirect prediction is used and the image size is changed. ). In each graph, the horizontal axis represents the image size ratio, with the image size under the reference conditions in Table 2 being 1. The lower limit of the image size is set within a range in which the aggregate can be visually recognized from the image. As shown in FIGS. 10(a) and 10(b), the image Regardless of the size ratio, the mean absolute difference was below the target value Tg. Therefore, it is considered that the image size (number of pixels or resolution) has little effect on the accuracy of prediction by the prediction model M.

The graphs shown in Figs. 11(a) and 11(b) show the verification results obtained by varying the moisture state of the surface of the aggregate. The "air-dried state" in this verification is a state in which the learning aggregates are left indoors for a predetermined period of time to dry (naturally dry). In addition, the "wet state" is a state in which water is added to the aggregate for learning in a predetermined ratio (for example, 1% to 5% in mass ratio) with respect to the mass of the aggregate and mixed in a container. is. From the results shown in FIGS. 11(a) and 11(b), it can be seen that the absolute difference average value is equal to or less than the target value Tg in the air-dried state, and the prediction model M can predict with high accuracy. In addition, in the image of the aggregate obtained in the wet state, when there was no illumination, compared with the image in the air-dried state, the overall color was uneven, and the edges of each particle of the aggregate appeared blurred. In addition, the image of the aggregate obtained in the wet state contained particles of the aggregate that looked whiter due to the reflection of the surface water compared to the image in the air-dried state with illumination.

[Modification]
As shown in FIG. 11(a), the mean absolute difference exceeds the target value Tg when the direct prediction is adopted and the amounts of coarse aggregate are 5 kg and 8 kg. As shown in FIG. 11(b), when indirect prediction is employed and the amount of coarse aggregate is 8 kg, the average absolute difference exceeds the target value Tg. Thus, it can be seen that in wet conditions, the amount of coarse aggregate can reduce the prediction accuracy. The prediction model M may be constructed differently than the example above to avoid a drop in prediction accuracy in wet conditions.

The prediction model M constructed by the model construction unit 72 may output only a value indicating the coarse particle ratio of the aggregate according to input information including image data obtained by imaging the aggregate for learning. . That is, the prediction model M may be a model that performs a single task. In this case, in the learning data for constructing the prediction model M, only the correct value of the coarse particle ratio of the aggregate is associated with the input information including the image data.

The model construction unit 72 may construct a prediction model M given by the following equation (7) instead of the above equation (4).

In the process of updating the weight w for generating the predictive model M, the model construction unit 72 may repeat the calculation represented by the following equation (8) instead of the above equation (5).

FIG. 12 shows the results of verification performed under the same conditions as the verification results shown in FIG. ing. As shown in FIG. 12, in both the air-dry state and the wet state, the absolute difference average value is equal to or less than the target value Tg. That is, it can be seen that even in a wet state, the coarse grain ratio can be predicted with high accuracy by using the prediction model M that performs a single task.

When the prediction model M performs multitasking, the calculation method in the process of updating the weight w may be different from the example described above. One of the factors that reduce the prediction accuracy in the wet state is the difference between the order of error relating to the mass ratio and the order of error relating to the coarse particle ratio in the process of updating the weight w. For example, of E, which is the sum of errors in the above formula (5), the order of E1, E2, E3, and E4, which are the errors related to the mass ratio (%), and the order of E5, which is the error related to the coarse grain ratio , there is a difference of about 10 to 100 times. In one example, the order of error for the mass ratio is about 1.0 to 10, and the order of error for the coarse grain ratio is about 0.001 to 0.10.

Repeating the calculations of the above equations (5) and (6) to update the weight w will result in an error (first error) regarding the coarse grain ratio of the aggregate and an error (second error) regarding the mass ratio for each particle size class error) and updating an intermediate model generated in the process of constructing a prediction model. If the order of the error related to the coarse grain ratio is smaller than the order of the error related to the mass ratio, the contribution of E5 in the above equation (5) becomes small, and in the process of updating the intermediate model, weighting is performed so as to make E5 small. Few operations that update w may be performed.

The model construction unit 72 uses the error corrected so as to reduce the difference between the coarse particle ratio error of the aggregate and the mass ratio error for each particle size division, and an intermediate model when constructing the prediction model M may be updated. For example, the model construction unit 72 multiplies each of the two types of errors by a coefficient so as to reduce the error difference. In one example, the model building unit 72 replaces the above formula (5) with the following formula (9) in the process of updating the weight w for generating the prediction model M (the process of updating the intermediate model). You may repeat the operations that

λ1 and λ2 in equation (9) are coefficients. The coefficient λ1 and the coefficient λ2 may be determined by an operator or the like based on the order of error of coarse grain ratio and the order of error of mass ratio. The coefficient λ2 may be about 5 to 200 times the coefficient λ1. The coefficient λ2 may be 10 times the coefficient λ1 or 100 times the coefficient λ1. In the present disclosure, the multitasking performed by the prediction model M constructed by updating the weight w by the calculation of Equation (9) is referred to as "coefficient adjustment multitasking." Further, the simple term "multitasking" means multitasking when error adjustment is not performed in the process of updating the weight w.

FIG. 13 shows verification results in a wet state for each of multitasking, coefficient adjustment multitasking, and single-tasking. The amount of coarse aggregate is 8 kg, and the image size ratio is 0.25. In multitasking and coefficient adjustment multitasking, direct prediction of the two types of prediction techniques predicts the coarse grain ratio. The coefficient λ1 is set to 1.00 and the coefficient λ2 is set to 100. From the results shown in FIG. 13, it can be seen that the prediction accuracy in the coefficient-adjusted multitasking and single-tasking is higher than the prediction accuracy in the multitasking in the wet state.

FIG. 14 shows the verification results when the values of the coefficients λ1 and λ2 are changed under the same conditions as the verification results shown in FIG. Specifically, the combination of coefficients (λ1, λ2) is defined as (1.0,10), (0.1,1.0), (1.0,100), and (0.01,1.0) ) was set for each of the above and evaluated. As shown in FIG. 14, regardless of whether the ratio of the coefficient λ2 to the coefficient λ1 was 10 or 100, the average absolute difference value was equal to or less than the target value of 0.1. From the results shown in FIG. 14, it can be seen that the average absolute difference tends to be smaller when the ratio of the coefficient λ2 to the coefficient λ1 is 100 than when the ratio is 100.

The various prediction models described above are examples, and the prediction model can be any model as long as it can output at least a prediction value for coarse aggregate from information including image data of aggregates. good. In the learning phase, images of the aggregates on the carrier 8 may be captured to prepare learning images (learning data). The input information and the evaluation information that are input to the prediction model may include, in addition to aggregate image data, information indicating physical quantities that can affect the coarse particle ratio.

The manufacturing system 1 may include a computer that controls the manufacturing apparatus 10 and a computer (quality prediction device) that predicts the coarse particle rate of the aggregate. A computer for predicting the coarse grain ratio of aggregate may have a model building section 72 , a model holding section 74 , a quality prediction section 76 and an output section 78 . At least part of the items described in the other examples may be applied to one of the various examples described above.

[Summary of this disclosure]
The aggregate quality prediction method described above is machine learning based on input information including image data obtained by imaging the aggregate and the correct value of the coarse grain ratio of the aggregate associated with the input information. According to the input information, including a preparatory step of constructing a prediction model M that outputs a value indicating the coarse particle ratio of the aggregate according to the input of the input information, and image data obtained by imaging the aggregate to be evaluated, an obtaining step of obtaining corresponding evaluation information; and a prediction step of inputting the evaluation information into the predictive model M to obtain a first predicted value output from the predictive model M regarding the aggregate coarseness.

Aggregate coarseness affects the quality of ready-mixed concrete produced by kneading aggregate-containing concrete materials. For example, in the ready-mixed concrete industry, among quality control operations, there is a circumstance that labor is used for process inspections such as fresh property tests. Therefore, in order to improve the efficiency of ready-mixed concrete production, it is desired to save labor in quality control. In the quality prediction method described above, the prediction model M constructed by machine learning is used to predict the coarse particle rate of the aggregate, so the work for inspecting the coarse particle rate of the aggregate used in the production of ready-mixed concrete is simplified. can be improved. As a result, the frequency of inspection can be increased, or labor can be used for quality control work different from the control of the coarse particle ratio, which is useful for stabilizing the quality of ready-mixed concrete.

In the quality prediction method described above, in the preparatory step, the prediction model M may be constructed so as to output only the value indicating the coarse particle ratio of the aggregate according to the input information. By the prediction model M performing the single task, as described above, even when the surface state of the aggregate is in a wet state, the prediction model M can be used to predict the coarse particle ratio with high accuracy. Aggregates in factories that manufacture ready-mixed concrete often contain more or less water, and improving the prediction accuracy in wet conditions is more beneficial for stabilizing the quality of ready-mixed concrete.

In the quality prediction method described above, the preparation step is to construct the prediction model M by machine learning based on the correct values of the mass ratios of the plurality of particle size categories of the aggregate, which are associated with the input information. may contain. In the preparatory step, the prediction model M may be constructed so as to further output a value indicating the mass ratio for each class included in the plurality of particle size classes according to the input information. The prediction step inputs the evaluation information to the prediction model M, further acquires a second prediction value output from the prediction model M regarding the mass ratio for each class, and uses the second prediction value as the coarse grain rate of the aggregate converting to . In this case, one predictive model M can be used to predict the aggregate coarseness in two different ways. Therefore, it is possible to evaluate the reliability of the prediction result of the coarse grain ratio while avoiding an increase in work for building the model.

In the quality prediction method described above, the preparation step is generated in the process of building the prediction model M based on the first error regarding the coarse grain ratio of the aggregate and the second error regarding the mass ratio for each category. The preliminary step includes updating the intermediate model, wherein the intermediate model is updated with the corrected error such that the difference between the first error and the second error is reduced. By performing the coefficient adjustment multitasking by the prediction model M, it is possible to accurately predict the coarse particle ratio using the prediction model M even when the surface state of the aggregate is in a wet state, as described above. Aggregates in factories that manufacture ready-mixed concrete often contain more or less water, and improving the prediction accuracy in wet conditions is more beneficial for stabilizing the quality of ready-mixed concrete.

In the quality prediction method described above, the preparation step may include constructing the prediction model M by machine learning using a convolutional neural network. In this case, a more complicated model can be constructed by performing machine learning using a convolutional neural network. Therefore, the predicted value of the coarse grain ratio can be brought closer to the correct value.

In the quality prediction method described above, the acquisition step is to capture an image of a range including the aggregates to be evaluated from a position distant from the aggregates to be evaluated while the aggregates to be evaluated are piled up, and obtain the evaluation information. may include obtaining Aggregates are often piled up in factories that manufacture ready-mixed concrete. For example, when receiving, aggregates are transported in transport trucks or transport vessels. Also, during production, the aggregates are transported in a piled state by a belt conveyor or the like. In the above configuration, imaging is performed in a state in which aggregates are piled up, so work for acquiring evaluation information can be simplified. It should be noted that the above-described quality prediction method can be used as a substitute for quality control work such as a quality sifting test, not only during transportation as described above. For example, at the time of receiving or manufacturing, a part of the aggregate may be collected, laid out on a floor or a sheet, and imaged.

In the quality prediction method described above, the preparation step may include preparing aggregates for learning so that grains in two or more grain size categories are included. The input information may include image data obtained by imaging the learning aggregate. In this case, the aggregates for learning are in a state similar to the actual aggregates used in a factory that manufactures ready-mixed concrete, and the accuracy of prediction by the prediction model M can be improved.

The method for producing ready-mixed concrete described above includes a production process for kneading materials containing aggregates to produce ready-mixed concrete, and a quality prediction for predicting the coarse particle rate of at least a part of the aggregates used in the production process. and In the quality prediction process, machine learning based on input information including image data obtained by imaging the aggregate and the correct value of the coarse grain ratio of the aggregate associated with the input information is performed. Acquisition of evaluation information corresponding to the input information, including a preparation step of building a prediction model M that outputs a value indicating the coarse particle ratio of the aggregate according to the evaluation, and image data obtained by imaging the aggregate to be evaluated. and a prediction step of inputting the evaluation information into the predictive model M to obtain a first predicted value output from the predictive model M regarding the aggregate coarseness. This manufacturing method is useful for stabilizing the quality of ready-mixed concrete, similarly to the quality prediction method described above.

The manufacturing system 1 described above includes a manufacturing apparatus 10 for kneading materials including aggregates to manufacture ready-mixed concrete, and a control apparatus 60 for predicting the coarse particle ratio of at least a portion of the aggregates used by the manufacturing apparatus 10. (quality prediction device); The control device 60 performs machine learning based on input information including image data obtained by imaging the aggregate and the correct value of the coarse grain ratio of the aggregate associated with the input information. Acquisition of evaluation information corresponding to the input information, including a preparation step of building a prediction model M that outputs a value indicating the coarse particle ratio of the aggregate according to the evaluation, and image data obtained by imaging the aggregate to be evaluated. and a prediction step of inputting the evaluation information to the prediction model M and obtaining a first prediction value output from the prediction model M regarding the coarse grain rate of the aggregate. This manufacturing system 1 is useful for stabilizing the quality of ready-mixed concrete, similarly to the quality prediction method described above.

DESCRIPTION OF SYMBOLS 1... Manufacturing system, 2... Material storage place, 4... Silo, 8... Conveyor, 10... Manufacturing apparatus, 50... Imaging device, 60... Control apparatus, 72... Model construction part, 74... Model holding part, 76... Quality prediction Part, 78... Output part, M... Prediction model.

Claims (9)

Machine learning based on input information including image data obtained by imaging the aggregate and the correct value of the coarse grain ratio of the aggregate associated with the input information, aggregates according to the input of the input information A preparatory step of building a prediction model that outputs only the value indicating the coarse grain rate of
an acquisition step of acquiring evaluation information corresponding to the input information, including image data obtained by imaging aggregates to be evaluated;
a predicting step of inputting the evaluation information into the predictive model and obtaining a first predicted value output from the predictive model with respect to the coarse grain rate of the aggregate. Input information including image data obtained by imaging an aggregate, a correct value of the coarse particle ratio of the aggregate associated with the input information, and a plurality of grain sizes of the aggregate associated with the input information. By machine learning based on the correct value of the mass ratio of each class, the value indicating the coarse grain ratio of the aggregate and the mass ratio of each class included in the plurality of grain size classes are calculated according to the input of the input information. A preparatory step of building a prediction model that outputs an indicated value;
an acquisition step of acquiring evaluation information corresponding to the input information, including image data obtained by imaging aggregates to be evaluated;
The evaluation information is input to the prediction model, and a first prediction value output from the prediction model regarding the coarse grain ratio of the aggregate and a second prediction value output from the prediction model regarding the mass ratio for each class and a prediction step of obtaining
The prediction step includes converting the second prediction value into a coarse grain rate of aggregate,
The preparation step updates the intermediate model generated in the process of building the prediction model based on a first error regarding the coarse grain ratio of the aggregate and a second error regarding the mass ratio for each class. including
The aggregate quality prediction method, wherein in the preparation step, the intermediate model is updated using an error corrected to reduce a difference between the first error and the second error. The aggregate quality prediction method according to claim 1 or 2 , wherein the preparation step includes constructing the prediction model by machine learning using a convolutional neural network. The acquisition step acquires the evaluation information by capturing an image of a range including the evaluation target aggregate from a position away from the evaluation target aggregate in a state where the evaluation target aggregate is stacked. The aggregate quality prediction method according to claim 1 or 2 , comprising: The preparation step includes preparing aggregates for learning so that grains of two or more grain sizes are included,
3. The aggregate quality prediction method according to claim 1 , wherein said input information includes image data obtained by imaging said learning aggregate. A manufacturing process of kneading materials including aggregate to manufacture ready-mixed concrete;
a quality prediction step of predicting the coarse particle rate of at least a portion of the aggregate used in the manufacturing process;
including
The quality prediction step includes
Machine learning based on input information including image data obtained by imaging the aggregate and the correct value of the coarse grain ratio of the aggregate associated with the input information, aggregates according to the input of the input information A preparatory step of building a prediction model that outputs only the value indicating the coarse grain rate of
an acquisition step of acquiring evaluation information corresponding to the input information, including image data obtained by imaging aggregates to be evaluated;
a prediction step of inputting the evaluation information into the prediction model and acquiring a first prediction value output from the prediction model regarding the coarse particle rate of the aggregate. A manufacturing process of kneading materials including aggregate to manufacture ready-mixed concrete;
a quality prediction step of predicting the coarse particle rate of at least a portion of the aggregate used in the manufacturing process;
including
The quality prediction step includes
Input information including image data obtained by imaging an aggregate, a correct value of the coarse particle ratio of the aggregate associated with the input information, and a plurality of grain sizes of the aggregate associated with the input information. By machine learning based on the correct value of the mass ratio of each class, the value indicating the coarse grain ratio of the aggregate and the mass ratio of each class included in the plurality of grain size classes are calculated according to the input of the input information. A preparatory step of building a prediction model that outputs an indicated value;
an acquisition step of acquiring evaluation information corresponding to the input information, including image data obtained by imaging aggregates to be evaluated;
The evaluation information is input to the prediction model, and a first prediction value output from the prediction model regarding the coarse grain ratio of the aggregate and a second prediction value output from the prediction model regarding the mass ratio for each class and a prediction step of obtaining
The prediction step includes converting the second prediction value into a coarse grain rate of aggregate,
The preparation step updates the intermediate model generated in the process of building the prediction model based on a first error regarding the coarse grain ratio of the aggregate and a second error regarding the mass ratio for each class. including
The ready-mixed concrete manufacturing method, wherein in the preparing step, the intermediate model is updated using an error corrected so as to reduce a difference between the first error and the second error. a manufacturing device for kneading materials including aggregate to manufacture ready-mixed concrete;
A quality prediction device that predicts the coarse particle rate of at least a portion of the aggregate used by the manufacturing device,
The quality prediction device is
Machine learning based on input information including image data obtained by imaging the aggregate and the correct value of the coarse grain ratio of the aggregate associated with the input information, aggregates according to the input of the input information A preparatory step of building a prediction model that outputs only the value indicating the coarse grain rate of
an acquisition step of acquiring evaluation information corresponding to the input information, including image data obtained by imaging aggregates to be evaluated;
and a prediction step of inputting the evaluation information into the prediction model and acquiring a first prediction value output from the prediction model regarding the coarse particle rate of the aggregate. manufacturing system. a manufacturing device for kneading materials including aggregate to manufacture ready-mixed concrete;
A quality prediction device that predicts the coarse particle rate of at least a portion of the aggregate used by the manufacturing device,
The quality prediction device is
Input information including image data obtained by imaging an aggregate, a correct value of the coarse particle ratio of the aggregate associated with the input information, and a plurality of grain sizes of the aggregate associated with the input information. By machine learning based on the correct value of the mass ratio of each class, the value indicating the coarse grain ratio of the aggregate and the mass ratio of each class included in the plurality of grain size classes are calculated according to the input of the input information. A preparatory step of building a prediction model that outputs an indicated value;
an acquisition step of acquiring evaluation information corresponding to the input information, including image data obtained by imaging aggregates to be evaluated;
The evaluation information is input to the prediction model, and a first prediction value output from the prediction model regarding the coarse grain ratio of the aggregate and a second prediction value output from the prediction model regarding the mass ratio for each class and a prediction step of obtaining and
The prediction step includes converting the second prediction value into a coarse grain rate of aggregate,
The preparation step updates the intermediate model generated in the process of building the prediction model based on a first error regarding the coarse grain ratio of the aggregate and a second error regarding the mass ratio for each class. including
The ready-mixed concrete manufacturing system, wherein in the preparing step, the intermediate model is updated using an error corrected to reduce a difference between the first error and the second error.

Images