Coal expansion pressure information prediction method and program

Abstract

This technology provides an improved prediction accuracy for coal expansion pressure information. [Solution] The coal expansion pressure information prediction method includes an acquisition step of obtaining the average maximum reflectance, inert ratio, and maximum fluidity of the coal to be predicted, and a prediction step of predicting expansion pressure clusters corresponding to the expansion pressure based on the acquired average maximum reflectance, inert ratio, and maximum fluidity of the coal to be predicted.

Description

Technical Field

[0001] This disclosure relates to a technique for predicting information on coal expansion pressure (coal expansion pressure information).

Background Art

[0002] In the process of producing coke by carbonizing coal, the heated coal expands and applies pressure to the coke oven. This pressure is called the expansion pressure. If the expansion pressure is high, it may have an adverse effect on the coke oven, and it is desirable to select coal that does not cause the expansion pressure to increase.

[0003] For example, Patent Document 1 discloses a technique for estimating the expansion pressure using the degree of coalification and the amount of inert components. Patent Document 2 discloses a technique for estimating the change over time of the expansion pressure using the amount of inert components and the maximum fluidity.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, it has been found that there is a type of coal in which a deviation occurs between the estimated expansion pressure and the actually measured value of the expansion pressure in the techniques described in Patent Documents 1 and 2.

[0006] This disclosure provides a technique for improving the prediction accuracy of coal expansion pressure information.

Means for Solving the Problems

[0007] The coal expansion pressure information prediction method of this disclosure includes an acquisition step of obtaining the average maximum reflectance, inert ratio, and maximum fluidity of the coal to be predicted, and a prediction step of predicting expansion pressure clusters corresponding to the expansion pressure based on the average maximum reflectance, inert ratio, and maximum fluidity of the coal to be predicted obtained. [Brief explanation of the drawing]

[0008] [Figure 1] This is a block diagram showing the training data generation system, the predictive model generation system, and the coal expansion pressure information prediction system according to the first embodiment. [Figure 2] This figure shows an example of coal data used in one embodiment of the present disclosure and an example of clustered coal data. [Figure 3] This figure shows a tree diagram generated by cluster analysis using four physical properties: average maximum reflectance, inert ratio, maximum fluidity, and total expansion coefficient, in one embodiment of the present disclosure. [Figure 4] This is a perspective view plotting multiple types of coal on a three-axis space representing average maximum reflectivity, inertness ratio, and maximum fluidity. [Figure 5] Figure 4 shows the space as viewed from a line of sight parallel to the axis of inertness. [Figure 6] Figure 4 shows the space as viewed from a line of sight parallel to the axis of maximum fluidity. [Figure 7] Figure 4 shows the space as viewed from a line of sight parallel to the axis of average maximum reflectance. [Figure 8] Figure 2 shows an example of a predictive model constructed using clustered coal data. [Figure 9] This flowchart shows the process performed by the training data generation system of the first embodiment. [Figure 10] This flowchart shows the process performed by the predictive model generation system of the first embodiment. [Figure 11] This flowchart shows the process performed by the coke quality prediction system of the first embodiment. [Figure 12]This figure shows a tree diagram generated by cluster analysis using three physical properties—average maximum reflectance, inert ratio, and maximum fluidity—for the same data as the example shown in Figure 3. [Figure 13] This is a perspective view plotting multiple types of coal on a three-axis space representing average maximum reflectivity, inertness ratio, and maximum fluidity. [Figure 14] Figure 13 shows the space as viewed from a line of sight parallel to the axis of inertness. [Figure 15] Figure 13 shows the space as viewed from a line of sight parallel to the axis of maximum fluidity. [Figure 16] Figure 13 shows the space as viewed from a line of sight parallel to the axis of average maximum reflectance. [Modes for carrying out the invention]

[0009] <Physical properties> The physical properties of coal used in this disclosure are described below. • Average maximum reflectance: This is one of the indicators of the degree of coalification. It is sometimes denoted as Ro. The unit is %. The ratio of the intensity of reflected light to incident light from the polished surface of a coal sample is measured using a microscope with polarized light. It is measured according to JIS M8816. • Inert ratio: One of the indicators of the amount of inert components. It is sometimes denoted as TI. The unit is %. Similar to the average maximum reflectance, it is measured by microscopy on the polished surface of a coal sample. It is measured according to JIS M8816. • Maximum fluidity: An indicator of coal softening and melting. Sometimes abbreviated as MF. Expressed in units of the rotational speed of the stirring rod of the Gieseler plastometer (rotations per minute: DDPM). Measured according to JIS M8801. The common logarithm of DDPM may also be used. When the common logarithm is used, it may be expressed as log MF or log DDPM. • Total expansion coefficient: This is the rate of volume change during the softening and melting of coal. It is sometimes abbreviated as TD. The unit is %. It is measured using a dilatometer in accordance with JIS M8801. · Expansion pressure: It is the pressure equivalent to the gas pressure of the gas generated during the softening and melting process of coal. It was measured using the device described in JP-A-2023-004004. The unit is [kPa]. The temperature was raised at a rate of 3°C per minute from the start of temperature increase until the coal core temperature reached 700°C or higher. During the temperature increase process, the pressure was measured with a load cell through a pressure receiving plate.

[0010] As described below, the inventors of the present application have found that by using at least three coal properties, namely the average maximum reflectance, inert ratio, and maximum fluidity of coal, an expansion pressure cluster corresponding to the expansion pressure can be accurately predicted. Furthermore, it has been found that by using four coal properties, namely the average maximum reflectance, inert ratio, maximum fluidity, and total expansion rate of coal, the prediction accuracy can be further improved. Therefore, in the first embodiment of the present disclosure described below, a method for constructing a prediction model 20 using machine learning and a method for predicting using the constructed prediction model 20 will be described. In the second embodiment, a method for constructing the prediction model 20 without using machine learning and a method for predicting using the constructed prediction model 20 will be described.

[0011] <First Embodiment> FIG. 1 is a block diagram showing a teacher data generation system 1, a prediction model generation system 2, and a coal expansion pressure information prediction system 3 according to the first embodiment.

[0012] As shown in Fig. 1, the coal expansion pressure information prediction system 3 predicts the expansion pressure cluster corresponding to the expansion pressure from the physical property values (average maximum reflectivity, inert ratio, maximum fluidity, total expansion rate) of the coal (raw coal, also expressed as a brand) to be predicted represented by the prediction target data D4. As shown in the cluster expansion pressure related data D6 shown in Fig. 1, the expansion pressure cluster is a cluster corresponding to the expansion pressure, and by collating the cluster expansion pressure related data D6, it is a cluster capable of specifying the range of the expansion pressure. The coal expansion pressure information prediction system 3 uses the prediction model 20 previously generated by machine learning by the prediction model generation system 2. The prediction model generation system 2 learns and constructs the prediction model 20 by machine learning using the teacher data D3. The teacher data generation system 1 generates the teacher data D3 for machine learning the prediction model 20 based on the coal data D1. The coal data D1 shows the physical property values (average maximum reflectivity, inert ratio, maximum fluidity, total expansion rate, expansion pressure) of each of a plurality of types of coal (raw coal). The numerical values shown in the coal data D1 are based on actual measurements. Hereinafter, each system will be described individually.

[0013] Each part constituting the teacher data generation system 1, the prediction model generation system 2, and the coal expansion pressure information prediction system 3 is realized by the CPU in a computer including a CPU, RAM, ROM, non-volatile memory, input / output interface, etc. executing information processing according to a program loaded from the ROM or non-volatile memory to the RAM.

[0014] <Teacher data generation system 1> As shown in Fig. 1, the teacher data generation system 1 includes a coal data acquisition unit 12, a cluster analysis unit 13, and a teacher data generation unit 11.

[0015] The coal data acquisition unit 12 acquires coal data D1 by input from the user, reading data from external storage, or reading data from internal storage. Coal data D1 includes multiple physical properties for each of several types of coal. The multiple physical properties (five physical properties in this embodiment) are average maximum reflectance (Ro, unit [%]), inert percentage (TI, unit [%]), maximum fluidity (MF, unit [log DDPM]), total expansion rate (TD, unit [%]), and expansion pressure (unit [kPa]). Figure 2 shows an example of coal data D1 and an example of clustered coal data D2 used in one embodiment of this disclosure. In the example coal data D1 shown in Figure 2, i=31 represents the physical properties of 31 types of coal (coking coal). In Figure 2, the coal brand names are converted to symbols for explanation purposes in this specification. The clustered coal data D2 shown in Figure 2 includes an expansion pressure cluster, but coal data D1 does not include data related to the cluster.

[0016] The cluster analysis unit 13 shown in Figure 1 performs dimensionality reduction processing, such as principal component analysis, on four data points in coal data D1: average maximum reflectance, inert ratio, maximum fluidity, and total expansion rate for multiple types of coal. By performing cluster analysis on the data reduced to three dimensions, multiple types of coal (i=31 types in this embodiment) are classified into one of several expansion pressure clusters. In this embodiment, one type of coal possessing average maximum reflectance, inert ratio, maximum fluidity, and total expansion rate is treated as one data point, the distance is set to Euclidean distance, and the merging method is the fully connected method (longest distance method) to perform hierarchical cluster analysis. The hierarchical cluster analysis generated the dendrogram shown in Figure 3. Figure 3 shows a dendrogram generated by cluster analysis using four physical properties: average maximum reflectance, inert ratio, maximum fluidity, and total expansion rate, in one embodiment of this disclosure. A threshold for Euclidean distance was determined so that there would be more clusters with similar expansion pressures for coals belonging to the same cluster, and multiple types of coal were classified into one of eight expansion pressure clusters (shown by circled numbers in Figure 3). The vertical axis in Figure 3 represents the symbols indicating the brands of the multiple types of coal, and the horizontal axis in Figure 3 represents the Euclidean distance. The expansion pressure clusters will be described later, but each expansion pressure cluster corresponds to an expansion pressure, and the value or range of the expansion pressure can be derived based on the expansion pressure cluster. On the other hand, it is not possible to identify an expansion pressure cluster from the value of the expansion pressure. The classification results for each of the multiple types of coal by the cluster analysis unit 13 are shown in the "Expansion Pressure Cluster" column of the table in Figure 2. Figure 2 is also a diagram showing clustered coal data D2, to which cluster information has been added to coal data D1. The clustered coal data D2 is stored in the storage unit 14 as shown in Figure 1. In this embodiment, the number of clusters (8) is smaller than the number of types of coal, but the number of clusters and the number of types of coal may be the same.

[0017] In this embodiment, hierarchical cluster analysis is used, but the method is not limited to this; any cluster analysis method will suffice. For example, DBSCAN, agglomerative clustering, or partitioning clustering may be used, or non-hierarchical cluster analysis methods such as x-means algorithm, k-means++, g-means, x-means, or xg-means may be used.

[0018] The training data generation unit 11 generates training data D3 based on the cluster analysis results (clustered coal data D2) from the cluster analysis unit 13. The training data D3, consisting of average maximum reflectance, inert ratio, maximum fluidity, and total expansion rate, becomes input to the prediction model 20, and the single correct output value from the prediction model 20 indicates that it is an expansion pressure cluster.

[0019] [Reasons for clustering based on four physical properties] This section explains why cluster analysis of four physical properties (average maximum reflectance, inert ratio, maximum fluidity, and total expansion rate) can accurately classify multiple types of coal into clusters corresponding to their expansion pressure. Figure 4 is a perspective view plotting multiple types of coal on a three-axis space with average maximum reflectance (Ro), inert ratio (TI), and maximum fluidity (log MF). Figure 5 is a view of the space shown in Figure 4 from a line of sight parallel to the inert ratio (TI) axis. Figure 6 is a view of the space shown in Figure 4 from a line of sight parallel to the maximum fluidity (log MF) axis. Figure 7 is a view of the space shown in Figure 4 from a line of sight parallel to the average maximum reflectance (Ro) axis. In Figures 4-7, each coal data point is shown as a circle, and the size of the circle corresponds to the expansion pressure. Larger expansion pressures are indicated by larger circles. The sign of the coal brand and the numerical value of the expansion pressure are shown near the circle of the data point. The total expansion rate (TD) is represented by the intensity of the gray color used to fill the circles representing the data points.

[0020] As shown in Figures 4-7, it can be seen that circles (data points) of similar size are clustered together in the spatial relationship between the three axes of average maximum reflectance (Ro), inert ratio (TI), and maximum fluidity (log MF). From this, it can be understood that by using the three physical properties of average maximum reflectance (Ro), inert ratio (TI), and maximum fluidity (log MF), it is possible to classify multiple types of coal into clusters of similar expansion pressure. Therefore, using the three physical properties of average maximum reflectance (Ro), inert ratio (TI), and maximum fluidity (log MF) makes it possible to classify coal into expansion pressure clusters compared to using only two of the three physical properties. Furthermore, even for data points that are close in distance on the three axes, the classification accuracy can be further improved by adding the total expansion rate (TD). Figure 12 shows a tree diagram generated by cluster analysis using three physical properties—average maximum reflectance, inert ratio, and maximum fluidity—for the same data as the example shown in Figure 3. For example, the expansion pressure of cluster (number 4) to which brand "R" in Figure 3 belongs is 48 kPa, and the expansion pressure of cluster (number 2) to which brands "V-1", "V-2", and "Y" in Figure 3 belong is 88-150 kPa. In Figure 12, these brands "R", "V-1", "V-2", and "Y" are close in distance. Also, the expansion pressure of cluster (number 5) to which brands "S-2" and "P" in Figure 3 belong is 4-10 kPa, and the expansion pressure of cluster (number 1) to which brand "Q" in Figure 3 belongs is 110-296 kPa. In Figure 12, these brands "S-2", "P", and "Q" are close in distance. In other words, the cluster analysis using four physical properties—average maximum reflectance, inert ratio, maximum fluidity, and total expansion coefficient—shown in Figure 3, results in a greater distance between brands with different expansion pressures compared to the cluster analysis using three physical properties—average maximum reflectance, inert ratio, and maximum fluidity—shown in Figure 12. Therefore, adding the total expansion coefficient can improve classification accuracy.

[0021] <Predictive Model Generation System 2> As shown in Figure 1, the prediction model generation system 2 has a learning unit 21 that trains and constructs a prediction model 20 using machine learning with training data D3. The prediction model 20 in this embodiment uses a decision tree. The decision tree algorithm used is CART (Classification and Regression Trees), but is not limited to this. Figure 8 shows an example of a prediction model 20 constructed using the clustered coal data D2 shown in Figure 2. As shown in Figure 8, CART is a decision tree that makes predictions based on Yes or No conditions for the values ​​of multiple physical properties (average maximum reflectance (Ro), inert ratio (TI), maximum fluidity (log MF), total expansion rate (TD)), which are explanatory variables. Decision trees are preferable to use because they clearly provide the basis for the prediction model 20's decisions. In this embodiment, the prediction model 20 is a decision tree, but it is not limited to this. For example, the prediction model 20 can be any supervised machine learning model, such as a random forest, support vector machine, neural network, or ensemble.

[0022] <Coal Expansion Pressure Information Prediction System 3> The coal expansion pressure information prediction system 3 shown in Figure 1 comprises a prediction target data acquisition unit 30, a prediction unit 31, an alert notification unit 33, and an expansion pressure notification unit 34.

[0023] The prediction target data acquisition unit 30 acquires prediction target data D4, which shows four physical properties of coal for which expansion pressure information is to be predicted (average maximum reflectance, inert ratio, maximum fluidity, and total expansion rate). If the prediction model 20 has been machine-trained using three physical properties (average maximum reflectance, inert ratio, and maximum fluidity) as input, the prediction target data D4 may show these three physical properties (average maximum reflectance, inert ratio, and maximum fluidity).

[0024] The prediction unit 31 uses the prediction model 20 to predict expansion pressure clusters corresponding to the data to be predicted D4. As described above, the prediction model 20 is pre-built by machine learning using the learning unit 21 of the prediction model generation system 2 to output expansion pressure clusters from four physical properties of coal (average maximum reflectance, inert ratio, maximum fluidity, and total expansion rate). By inputting the data to be predicted D4 into the prediction model 20, the prediction model 20 outputs expansion pressure clusters.

[0025] The storage unit 32 of the coal expansion pressure information prediction system 3 stores cluster expansion pressure-related data D6 as information related to expansion pressure. As shown in Figure 1, cluster expansion pressure-related data D6 is data that associates expansion pressure clusters with expansion pressure values ​​or ranges of expansion pressure values. As illustrated in Figure 1, the expansion pressure of coal belonging to cluster "1" is approximately 110-300 kPa, the expansion pressure of coal belonging to cluster "2" is approximately 88-150 kPa, the expansion pressure of coal belonging to cluster "3" is approximately 23-56 kPa, and the expansion pressure of coal belonging to cluster "8" is approximately 220-330 kPa. Cluster expansion pressure-related data D6 is generated based on clustered coal data D2 (see Figure 2). The expansion pressure value or range can be obtained by aggregating the expansion pressure values ​​in the clustered coal data D2 by expansion pressure cluster. The expansion pressure range can be obtained from the maximum and minimum expansion pressure values ​​belonging to the cluster. The value of the expansion pressure can be obtained from the statistical values ​​(mean, median, etc.) of multiple expansion pressures belonging to the cluster.

[0026] The expansion pressure notification unit 34 is configured to notify externally of expansion pressure information (the value or range of the expansion pressure) corresponding to the expansion pressure cluster predicted by the prediction unit 31, based on the cluster expansion pressure related data D6 and the expansion pressure cluster predicted by the prediction unit 31. Examples of notification methods include displaying the information on a display or transmitting the data to a predetermined notification terminal. For example, if the expansion pressure cluster is "8", the notification will state that the expansion pressure is approximately 220 to 323 kPa.

[0027] The storage unit 32 of the coal expansion pressure information prediction system 3 stores alert information D5. Alert information D5 is information about expansion pressure clusters that will issue an alert, and it associates expansion pressure clusters with alert notification rules. The alert notification rules include information such as whether or not to issue an alert, and if an alert is issued, the content of the notification. In this embodiment, clusters containing expansion pressure values ​​that are evaluated as high are targeted for notification. Specifically, expansion pressure clusters "1" and "8" are targeted for notification, and the notification content is "high expansion pressure".

[0028] The alert notification unit 33 is configured to notify alerts based on alert information D5 and the expansion pressure cluster predicted by the prediction unit 31. Examples of notification methods include displaying the information on a screen and transmitting data to a predetermined notification terminal. For example, if the prediction unit 31 predicts that the expansion pressure cluster is "1" or "8", it can notify an alert stating, "The coal being predicted has high expansion pressure and requires attention."

[0029] [Method for generating training data] The method for generating training data will be explained using Figure 9. The training data generation method (training data generation process) is executed by one or more processors that constitute the training data generation system 1. As shown in Figure 9, in step ST100, the training data generation system 1 acquires coal data D1 representing the average maximum reflectance, inert ratio, maximum fluidity, total expansion rate, and expansion pressure for each of several types of coal. In the next step, ST101, the training data generation system 1 performs cluster analysis on the average maximum reflectance, inert ratio, maximum fluidity, and total expansion rate of multiple types of coal in the coal data D1, and classifies the multiple types of coal into one of several expansion pressure clusters corresponding to the expansion pressure. In the next step, ST102, the training data generation system 1 takes the average maximum reflectance, inert ratio, maximum fluidity, and total expansion rate as inputs and generates training data D3 with expansion pressure clusters as output.

[0030] [Method for generating a coal expansion pressure information prediction model] The method for generating a coal expansion pressure information prediction model will be explained using Figure 10. The coal expansion pressure information prediction model generation method is executed by one or more processors that constitute the prediction model generation system 2. As shown in Figure 10, in step ST200, the above training data generation process is executed to generate training data D3. In the next step, ST201, the prediction model generation system 2 generates a prediction model 20 that predicts expansion pressure clusters by using training data D3 for machine learning.

[0031] [Method for predicting coal expansion pressure information] The coal expansion pressure information method will be explained using Figure 11. The coal expansion pressure information prediction method is performed by one or more processors that constitute the coal expansion pressure information prediction system 3. As shown in Figure 11, in step ST300, the coal expansion pressure information prediction system 3 obtains the average maximum reflectivity, inert ratio, maximum fluidity, and total expansion rate of the coal to be predicted. In the next step, ST301, the coal expansion pressure information prediction system 3 predicts expansion pressure clusters from the average maximum reflectivity, inert ratio, maximum fluidity, and total expansion ratio of the coal, using a prediction model 20 constructed by machine learning to output expansion pressure clusters corresponding to the expansion pressure from the average maximum reflectivity, inert ratio, maximum fluidity, and total expansion ratio of the coal to be predicted. In the next step, ST302, the coal expansion pressure information prediction system 3 notifies the predicted expansion pressure cluster. It also notifies an alert based on the predicted expansion pressure cluster and alert information D5. Furthermore, it notifies the expansion pressure information (expansion pressure value or range of value) corresponding to the predicted expansion pressure cluster based on the predicted expansion pressure cluster and cluster expansion pressure related data D6.

[0032] <Second Embodiment> In the second embodiment, a method for constructing a predictive model 20 without using machine learning techniques will be described. In the second embodiment, the predictive model 20 is constructed by creating classification lines in a space with three axes: average maximum reflectance, inert ratio, and maximum fluidity. Figure 13 is a perspective view plotting each of several types of coal on a space with three axes: average maximum reflectance, inert ratio, and maximum fluidity. Figure 14 is a view of the space shown in Figure 13 from a line of sight parallel to the axis of inert ratio. Figure 15 is a view of the space shown in Figure 13 from a line of sight parallel to the axis of maximum fluidity. Figure 16 is a view of the space shown in Figure 13 from a line of sight parallel to the axis of average maximum reflectance. As shown in Figures 14 to 16, the coals were classified into three stages based on the maximum fluidity (MF) values ​​of 1.5 and 3. As shown in Figures 14 to 15, the coals were classified into three stages based on the inert ratio (TI) values ​​of 15 and 27. In Figure 14, the expansion pressure group is classified into those with an MF of 1.5 to 3. However, since there are differences in expansion pressure even within this group, it is further classified into two groups based on an average maximum reflectance (Ro) value of 1.0. The same line type is used for the same classification line (solid line, dashed line, large dashed line, small dashed line) in each figure. Coals (S and P) included in the MF 1.5 to 3 classification in Figure 14 can also be classified by considering the inert ratio (TI) as shown in Figure 15. As can be seen from Figures 13 to 16, it can be understood that a certain degree of expansion pressure clusters can be classified by classifying them based on the magnitude relationship of the three axes, even without using machine learning methods. The number of classification lines and classification points shown in Figures 13 to 16 can be changed to further refine the classification or to further coarse the classification. Note that the method for constructing the prediction model 20 described in the first and second embodiments is merely an example, and it may be constructed using other methods.

[0033] [1] As in the embodiment described above, the coal expansion pressure information prediction method may include an acquisition step of acquiring the average maximum reflectance, inert ratio, and maximum fluidity of the coal to be predicted, and a prediction step of predicting expansion pressure clusters corresponding to the expansion pressure based on the acquired average maximum reflectance, inert ratio, and maximum fluidity of the coal to be predicted. In this way, by using at least three coal properties—average maximum reflectivity, inert ratio, and maximum fluidity—it becomes possible to predict expansion pressure clusters corresponding to the expansion pressure of coal.

[0034] [2] The coal expansion pressure information prediction method described in [1] above may also be configured such that, in the acquisition step, the total expansion rate of the coal to be predicted is acquired, and in the prediction step, the expansion pressure cluster is predicted based on the acquired average maximum reflectance, inert ratio, maximum fluidity, and total expansion rate of the coal to be predicted. Thus, by using four coal properties—average maximum reflectance, inert ratio, maximum fluidity, and total expansion rate—it becomes possible to accurately predict the expansion pressure cluster corresponding to the expansion pressure of the coal.

[0035] [3] The coal expansion pressure information prediction method described in [1] or [2] above may also be used in the prediction step, wherein a prediction model 20 constructed to predict expansion pressure clusters based on the average maximum reflectivity, inertness ratio, and maximum fluidity of coal is used.

[0036] [4] The coal expansion pressure information prediction method described in [3] above may be such that the prediction model 20 is constructed using coal data D1 representing the average maximum reflectance, inertness ratio, maximum fluidity, and expansion pressure of each of several types of coal. In this way, it becomes possible to construct a predictive model 20 based on the properties of multiple types of coal.

[0037] [5] The coal expansion pressure information prediction method described in [4] above may be constructed by machine learning using training data D3, which takes the average maximum reflectance, inert ratio, and maximum fluidity of multiple types of coal in coal data D1 as input and outputs expansion pressure clusters obtained by cluster analysis of the average maximum reflectance, inert ratio, and maximum fluidity of multiple types of coal in coal data D1. In this way, it becomes possible to construct a predictive model 20 using machine learning techniques.

[0038] [6] A coal expansion pressure information prediction method described in any of [1] to [5] above, which may include the steps of: providing information about an expansion pressure cluster for which an alert is to be issued; and issuing an alert based on the predicted expansion pressure cluster. It will be possible to send alerts based on predicted expansion pressure clusters. For example, it will be possible to send alerts for coal that is under high expansion pressure and requires attention.

[0039] [7] A coal expansion pressure information prediction method according to any of the above [1] to [6] may include the step of reporting the expansion pressure value or range based on related data that associates the expansion pressure value or range with an expansion pressure cluster, and the predicted expansion pressure cluster. It becomes possible to notify the expansion pressure value or range of value corresponding to the predicted expansion pressure cluster.

[0040] [8] As in the above embodiment, the program may cause one or more processors to execute the method described in any of [1] to [7] above.

[0041] The computer-readable temporary recording medium according to this embodiment stores the above-mentioned program.

[0042] Although embodiments of this disclosure have been described above with reference to the drawings, it should be understood that the specific configurations are not limited to these embodiments. The scope of this disclosure is indicated not only by the description of the embodiments above but also by the claims, and further includes all modifications within the meaning and scope equivalent to the claims.

[0043] The structures adopted in each of the above embodiments can be adopted in any other embodiment. The specific configuration of each part is not limited to the embodiments described above, and various modifications are possible without departing from the spirit of this disclosure.

[0044] (A) In the above embodiment, an alert notification unit 33 and an expansion pressure notification unit 34 are provided, but the embodiment is not limited thereto. The alert notification unit 33 may be omitted, the expansion pressure notification unit 34 may be omitted, or both the alert notification unit 33 and the expansion pressure notification unit 34 may be omitted.

[0045] (B) In the above embodiment, cluster analysis is performed based on four physical properties (average maximum reflectance of coal, inert ratio, maximum fluidity, and total expansion rate), and the prediction model 20 is input to these four physical properties, but is not limited to this. Even if the four physical properties are reduced to three physical properties (average maximum reflectance of coal, inert ratio, and maximum fluidity), as described above, it is still possible to classify the expansion pressure clusters to some extent, which is useful. [Explanation of Symbols]

[0046] 20: Predictive Models D1: Coal data D3: Training data

Claims

1. Acquisition steps to obtain the average maximum reflectivity, inert percentage, and maximum fluidity of the coal to be predicted, A prediction step to predict expansion pressure clusters corresponding to expansion pressure based on the average maximum reflectivity and inertness ratio and maximum fluidity of the coal to be predicted obtained, A method for predicting coal expansion pressure information, including the above.

2. In the acquisition step, the total expansion rate of the coal to be predicted is acquired. The coal expansion pressure information prediction method according to claim 1, wherein in the prediction step, the expansion pressure cluster is predicted based on the average maximum reflectance, inert ratio, maximum fluidity, and total expansion rate of the coal to be predicted obtained.

3. The coal expansion pressure information prediction method according to claim 1, wherein in the prediction step, a prediction model constructed to predict the expansion pressure cluster based on the average maximum reflectivity, inertness ratio, and maximum fluidity of the coal is used.

4. The coal expansion pressure information prediction method according to claim 3, wherein the prediction model is constructed using coal data representing the average maximum reflectance, inertness ratio, maximum fluidity, and expansion pressure for each of several types of coal.

5. The coal expansion pressure information prediction method according to claim 4, wherein the prediction model is constructed by machine learning using training data in which the average maximum reflectance, inert ratio, and maximum fluidity of the multiple types of coal in the coal data are inputs, and expansion pressure clusters obtained by cluster analysis of the average maximum reflectance, inert ratio, and maximum fluidity of the multiple types of coal in the coal data are output.

6. A coal expansion pressure information prediction method according to any one of claims 1 to 5, comprising the steps of: providing information about the expansion pressure cluster for which an alert is to be provided; and providing an alert based on the predicted expansion pressure cluster.

7. A coal expansion pressure information prediction method according to any one of claims 1 to 5, comprising the steps of: providing related data that associates an expansion pressure value or range of values ​​with the expansion pressure cluster; and providing information on the expansion pressure value or range of values ​​based on the predicted expansion pressure cluster.

8. A program that causes one or more processors to execute the method according to any one of claims 1 to 5.

Images