Operation plan creation system

Abstract

An operation plan creation system disclosed herein includes: a storage unit for storing feature amount data on each of a plurality of products manufactured in the past, the feature amount data including data on the power consumption when manufacturing each of the plurality of products; a reception unit for receiving requests for creating an operation plan; an extraction unit for extracting data on a product to be manufactured from a production plan including data on the product when a request for creating an operation plan is received by the reception unit; a first determination unit for determining whether the product extracted by the extraction unit matches one of the plurality of products stored in the storage unit; a second determination unit for determining whether a corresponding physical model can be generated using the product as the specific product when it is determined that the product does not match; a machine learning model generation unit that, when it is determined that a corresponding physical model cannot be generated, generates a machine learning model corresponding to the specific product using at least a portion of the feature amount data on the plurality of products stored in the storage unit; a power consumption prediction unit that predicts the power consumption of the specific product using the machine learning model of the specific product generated by the machine learning model generation unit; an operation plan creation unit that uses data on the power consumption of the specific product, predicted by the power consumption prediction unit, to create an operation plan so that power is within a desired power range in a predetermined period; and an output unit that outputs the operation plan created by the operation plan creation unit.

Description

Operation plan creation system

[0001] The present disclosure relates to an operation plan creation system.

[0002] 2. Description of the Related Art A technology relating to an operation plan creation support device that supports the creation of an operation plan for manufacturing equipment has been disclosed (see, for example, Patent Document 1).

[0003] Japanese Patent Application Laid-Open No. 2019-21211

[0004] Conventionally, when creating an operation plan for planning the operating status of equipment in a production plan for manufacturing multiple types of products, the operation plan has been created using historical data, such as power consumption data for products that have been manufactured in the past, stored in a database. Power consumption data for a certain product refers to the total power consumption trend of the equipment used to manufacture that product from the start of production to the end of production. However, with this method, if the production plan includes the manufacture of a product that has never been manufactured in the past, it is difficult to create an appropriate operation plan because the database does not contain historical data. In this case, a method of creating a physical model for the product without historical data is considered. However, if the manufacture of a product without historical data includes a process, such as firing, in which the composition changes in a complex manner or the atmosphere during firing changes, it is difficult to create an appropriate physical model. As a result, it is difficult to accurately predict power consumption, and an appropriate operation plan cannot be created. It is therefore necessary to create a more appropriate equipment operation plan even when a production plan includes a product that has never been manufactured in the past.

[0005] Therefore, one of the objects is to provide an operation plan creation system that can create a more appropriate equipment operation plan even when a production plan includes a product that has not been manufactured in the past.

[0006] An operation plan creation system according to the present disclosure creates operation plans for a plurality of pieces of equipment so that power consumption falls within a desired range within a predetermined period of time when a plurality of products are manufactured using a plurality of pieces of equipment. the operation plan creation system includes: a memory unit that stores feature data for each of a plurality of products that have been manufactured in the past, including data on the power consumption of each of the plurality of products when manufacturing each of the plurality of products; a reception unit that receives a request for creation of an operation plan; an extraction unit that, when the reception unit receives a request for creation of an operation plan, extracts product data from a production plan including data on the product to be manufactured; a first judgment unit that determines whether the product extracted by the extraction unit matches one of the plurality of products stored in the memory unit; a second judgment unit that, when the first judgment unit determines that there is no match, determines whether a corresponding physical model can be generated for the product as a specific product; a machine learning model generation unit that, when the second judgment unit determines that a corresponding physical model cannot be generated, generates a machine learning model corresponding to the specific product using at least a portion of the feature data for the plurality of products stored in the memory unit; a power consumption prediction unit that predicts the power consumption of the specific product using the machine learning model of the specific product generated by the machine learning model generation unit; an operation plan creation unit that creates an operation plan using the power consumption data of the specific product predicted by the power consumption prediction unit to keep the power consumption within a desired power range within a predetermined period; and an output unit that outputs the operation plan created by the operation plan creation unit.

[0007] According to such an operation plan creation system, even when a production plan includes a product that has not been manufactured in the past, it is possible to create a more appropriate equipment operation plan.

[0008] Fig. 1 is a block diagram showing a schematic configuration of an operation plan creation system in embodiment 1. Fig. 2 is a flowchart showing the processing flow when creating an operation plan using the operation plan creation system. Fig. 3 is a graph showing the relationship between the total power of all equipment and the power in an electric kiln and the elapsed time when a product is manufactured in a conventional manner. Fig. 4 is a graph showing the relationship between the total power of all equipment and the power in an electric kiln and the elapsed time when a product is manufactured in accordance with an operation plan created by the operation plan creation system. Fig. 5 is a graph showing the difference in power consumption of an electric kiln before and after degradation.

[0009] [Outline of the embodiment] First, embodiments of the present disclosure will be described. When a plurality of products are manufactured using a plurality of pieces of equipment, an operation plan creation system according to the present disclosure creates operation plans for the plurality of pieces of equipment so that the power consumption falls within a desired range within a predetermined period of time. the operation plan creation system includes: a memory unit that stores feature data for each of a plurality of products that have been manufactured in the past, including data on the power consumption of each of the plurality of products when manufacturing each of the plurality of products; a reception unit that receives a request for creation of an operation plan; an extraction unit that, when the reception unit receives a request for creation of an operation plan, extracts product data from a production plan including data on the product to be manufactured; a first judgment unit that determines whether the product extracted by the extraction unit matches one of the plurality of products stored in the memory unit; a second judgment unit that, when the first judgment unit determines that there is no match, determines whether a corresponding physical model can be generated for the product as a specific product; a machine learning model generation unit that, when the second judgment unit determines that a corresponding physical model cannot be generated, generates a machine learning model corresponding to the specific product using at least a portion of the feature data for the plurality of products stored in the memory unit; a power consumption prediction unit that predicts the power consumption of the specific product using the machine learning model of the specific product generated by the machine learning model generation unit; an operation plan creation unit that creates an operation plan using the power consumption data of the specific product predicted by the power consumption prediction unit to keep the power consumption within a desired power range within a predetermined period; and an output unit that outputs the operation plan created by the operation plan creation unit.

[0010] According to the operation plan creation system disclosed herein, when a production plan includes a characteristic product that has never been manufactured in the past and a physical model cannot be generated, a machine learning model corresponding to the specific product is generated using at least a portion of the feature data of a product that has been manufactured in the past. The generated machine learning model can then be used to predict power consumption. This makes it possible to more accurately predict power consumption and create an operation plan that falls within a desired power range, even for a product that has never been manufactured in the past and for which a physical model cannot be generated. Therefore, with this operation plan creation system, a more appropriate operation plan can be created even when a production plan includes a product that has never been manufactured in the past.

[0011] In the operation plan creation system of the above aspect, the product feature data may include at least one of data on temperature transitions during product manufacturing, data on pressure transitions during product manufacturing, and data on time during product manufacturing. This allows more effective data to be used as product feature data for machine learning of power consumption. This allows for more accurate machine learning models to be generated, and more appropriate operation plans to be created.

[0012] In the operation plan creation system of the above aspect, the operation plan creation unit may create an operation plan that reflects the degree of deterioration of the equipment. The equipment deteriorates with use, and power consumption changes along with the deterioration. With this configuration, the operation plan can be created taking into account changes in power consumption that change depending on the degree of deterioration of the equipment, thereby creating a more appropriate operation plan.

[0013] In the operation plan creation system according to any one of the above aspects, the machine learning in the machine learning model generation unit may be performed using a random forest. Such machine learning is more suitable for creating an operation plan.

[0014] In the operation plan creation system according to any one of the above aspects, the facility may include an electric kiln having a kiln formed using carbon. With such an electric kiln, the amount of carbon decreases with continued use, and power consumption changes and is not constant. Therefore, the operation plan creation system having the above configuration makes it possible to create an operation plan that responds to changes in power consumption.

[0015] In the operation plan creation system according to any one of the above aspects, the feature data may include heat curve data during product manufacturing. Heat curve data during product manufacturing is very important as feature data accompanying changes in power consumption during product manufacturing. Therefore, by adopting such a configuration, a more appropriate operation plan can be created.

[0016] [Specific Example of Embodiment] Next, specific examples of the operation plan creation system of the present disclosure will be described with reference to the drawings. In the following drawings, the same or corresponding parts are designated by the same reference characters, and description thereof will not be repeated.

[0017] First Embodiment A production planning system according to the present disclosure will be described. FIG. 1 is a block diagram showing a schematic configuration of an operation planning system according to the first embodiment. The operation planning system 10 according to the first embodiment of the present disclosure creates operation plans for multiple pieces of equipment so that a desired power range is maintained within a predetermined period when multiple products are manufactured using multiple pieces of equipment. Here, "within the desired power range" refers to controlling power within the desired range. One example of an operation plan is adjusting the operation timing of multiple pieces of equipment to minimize peak power. Examples of the multiple pieces of equipment include manufacturing equipment directly used in manufacturing ceramic products and lighting that is not directly used in the manufacture of products but is used indirectly. An example of manufacturing equipment is an electric kiln that heats raw materials. For example, multiple electric kilns are used in parallel in a factory. An electric kiln may be used in a batch format, where one batch is processed from the start to the end of heat treatment. In this embodiment, the multiple pieces of equipment are multiple electric kilns in a single factory. Furthermore, in this embodiment, the desired power range is defined as a range in which peak power is minimized. In this case, the peak power is controlled to be minimized within the range of the contracted power.

[0018] Referring to FIG. 1 , an operation plan creation system 10 according to the first embodiment of the present disclosure is provided in a control unit of a single computer, such as a server or a laptop PC. The control unit includes a CPU, a volatile memory, and the like. Of course, each component may be included across two or more computers. The operation plan creation system 10 includes a server 11 and multiple, specifically, three, electric kilns 12a, 12b, and 12c. Note that wireless communication is possible between the server 11 and each of the electric kilns 12a, 12b, and 12c. The server 11 and each of the electric kilns 12a, 12b, and 12c may also be configured to communicate via wires or via other components. The server 11 is installed, for example, in an office within a factory, and the electric kilns 12a, 12b, and 12c are installed at a manufacturing site within the factory.

[0019] The power consumption of each electric kiln 12a, 12b, 12c during operation is not always constant but varies depending on the temperature, elapsed time, etc. Because the manufacturing conditions differ for each of the multiple types of products being manufactured, the power consumption of the equipment varies even when the same heat treatment is performed. For example, because the manufacturing conditions differ for each product, the heat curve, which is the progression of the baking temperature, differs, and the power consumption also varies accordingly. The electric kilns 12a, 12b, 12c have kilns made of carbon. The carbon-made portion is located on the inner wall side of the kiln.

[0020] The server 11 includes a database 21 as a storage unit, a reception unit 22, an extraction unit 23, a first determination unit 24, a second determination unit 25, a machine learning model generation unit 26, a power consumption prediction unit 27, an operation plan creation unit 28, and an output unit 29. That is, in the present embodiment, the server 11 includes all of the components included in the operation plan creation system 10.

[0021] The database 21 stores various types of data. The data stored in the database 21 includes historical data on products that have been manufactured in the past, such as data on the power consumption, heating temperature, heating time, and heat curve of products manufactured in the past. The heating temperature, heating time, and heat curve data may be used as feature data that characterizes the manufacturing of the product. In this embodiment, the product feature data includes data on temperature transitions during product manufacturing, data on pressure transitions during product manufacturing, and data on the time required for manufacturing the product. The database 21 stores feature data for each of multiple products that have been manufactured in the past, including data on the power consumption during the manufacturing of each of the multiple products. Each piece of feature data is stored in association with each product. Even if the same product is manufactured at different times, each piece of feature data is stored at the respective timing. In other words, when manufacturing a product under a new production plan, feature data from the most recent manufacturing run can be selected and used.

[0022] The receiving unit 22 receives a request for creating an operation plan. When the receiving unit 22 receives the request for creating an operation plan, the extraction unit 23 extracts product data from a production plan including data on the products to be manufactured. The first determination unit 24 determines whether the product extracted by the extraction unit 23 matches one of the multiple products stored in the database 21. If the first determination unit 24 determines that there is no match, the second determination unit 25 determines whether a corresponding physical model can be generated for the product as a specific product. If the second determination unit 25 determines that a corresponding physical model cannot be generated, the machine learning model generation unit 26 generates a machine learning model corresponding to the specific product using at least a portion of the feature data of the multiple products stored in the database 21. In this embodiment, the machine learning in the machine learning model generation unit 26 is performed using a random forest. The power consumption prediction unit 27 predicts the power consumption of the specific product using the machine learning model of the specific product generated by the machine learning model generation unit 26. The operation plan creation unit 28 uses the power consumption data of the specific product predicted by the power consumption prediction unit 27 to create an operation plan such that the power consumption is within a desired power range within a predetermined period. The output unit 29 outputs the operation plan created by the operation plan creation unit 28. These components will be described in detail later.

[0023] Next, a description will be given of a case where an operation plan is created using the operation plan creation system 10. Fig. 2 is a flowchart showing a processing flow when an operation plan is created using the operation plan creation system 10.

[0024] Referring also to FIG. 2 , the server 11 first receives a request for creating an operation plan through the receiving unit 22 (YES in step S11; hereinafter, the term "step" will be omitted). The receiving unit 22 receives the request for creating an operation plan, for example, based on input information input to the server 11. At this time, the receiving unit 22 also receives input of production plan data including the operation plan. That is, the receiving unit 22 also receives input of data on products to be manufactured in the production plan and data on equipment that is the target of the operation plan. The receiving unit 22 also receives requests for the operation plan, for example, a request to set a desired power range within which peak power should be minimized. The receiving unit 22 then receives data on a predetermined period, for example, one week. That is, the receiving unit 22 requests that an operation plan be created that minimizes peak power over a one-week period.

[0025] Next, the extraction unit 23 extracts product data from the production plan including data on the products to be manufactured when the request for creating an operation plan is received by the receiving unit 22. Specifically, for example, when the request for creating an operation plan is received by the receiving unit 22, the extraction unit 23 first accesses the database 21 (S12) and extracts product data included in the production plan (S13).

[0026] The first determination unit 24 then determines whether the products extracted by the extraction unit 23 include any products stored in the database 21 that have not been manufactured in the past (S14). That is, it determines whether the data of products included in the production plan is included in the data of products that have been manufactured in the past stored in the database 21. In this case, the determination is made for all products included in the production plan. If it is determined that no products that have not been manufactured in the past are included (NO in S14), the operation plan creation unit 28 creates an operation plan based on past power consumption (S20). In this case, an operation plan is created that minimizes peak power within a desired power range based on the data of products that have been manufactured in the database 21 and the feature data corresponding to the products. The created operation plan is then output, and the process ends (S19).

[0027] On the other hand, if the first judgment unit 24 determines that there is no match (YES in S14), the second judgment unit 25 determines whether a corresponding physical model can be generated for the product as a specific product (S15). That is, a product that has not been manufactured in the past is defined as a specific product, and if the first judgment unit 24 determines that a product that has not been manufactured in the past is included, the second judgment unit 25 determines whether a corresponding physical model can be generated for the product as a specific product. Examples of criteria for determining whether a physical model can be generated include whether the manufacturing of the product does not include a firing process or the manufacturing of the product does not involve complex changes in composition.

[0028] If it is determined that a physical model can be created (YES in S15), the operation plan creation unit 28 creates a physical model for the specific product (S21) and predicts power consumption using the created physical model (S22). Then, an operation plan is created based on the predicted power consumption of the specific product and the power consumption of past products (S22). In this case, an operation plan is created that minimizes peak power within a desired power range based on power consumption data of products that have been manufactured in the past and have a track record in the database 21, power consumption data predicted for the specific product for which a physical model has been created, and feature data corresponding to each product. The created operation plan is then output, and the process ends (S19).

[0029] Next, if the second judgment unit 25 determines that a physical model cannot be created (NO in S15), the machine learning model generation unit 26 generates a machine learning model for the specific product (S16). In this case, the machine learning model generation unit 26 performs machine learning using random forest, and therefore the generated machine learning model is a machine learning model based on random forest. Here, an example of generating a machine learning model is described. First, data on the power consumption, temperature, pressure inside the electric kiln, etc. of various products baked in the past is obtained from the database 21. Then, feature data such as the amount of temperature change, the state of the electric kiln (how the temperature is rising, how the temperature is maintained, and how the temperature is falling), and the elapsed time in the state of the electric kiln are generated from the temperature and pressure inside the electric kiln. Then, a model for predicting power consumption is generated using machine learning using the generated feature data.

[0030] Next, the power consumption prediction unit 27 predicts the power consumption of the specific product using the machine learning model generated for the specific product by the machine learning model generation unit 26 (S17). To predict the power consumption of the specific product, the set temperature and set pressure inside the electric baking furnace of the specific product that is being manufactured for the first time are input into the prediction model described above to predict the power consumption.

[0031] Next, the operation plan creation unit 28 uses the power consumption data of the specific product predicted by the power consumption prediction unit to create an operation plan so that the power consumption falls within a desired power range within a predetermined period (S18). In this embodiment, the predetermined period is one week, and the operation plan is created so that the desired power range falls within a range that minimizes peak power. As a specific example, mathematical optimization is used to create an operation plan that minimizes power peaks based on the types of products manufactured within one week, the number of processing times, and the power consumption. Furthermore, in this embodiment, the operation plan creation unit 28 creates an operation plan that reflects the degree of deterioration of the equipment, specifically the electric kilns 12a, 12b, and 12c.

[0032] Finally, the output unit 29 outputs the operation plan created by the operation plan creation unit 28 (S19). The output by the output unit 29 is performed, for example, by displaying the data of the operation plan on a display connected to the server 11, or by transmitting the data of the operation plan to a mobile terminal owned by the planner who created the production plan.

[0033] According to the operation plan creation system 10, when a production plan includes a characteristic product that has not been manufactured in the past and a physical model cannot be generated, a machine learning model corresponding to the specific product is generated using at least a portion of the feature data of a product that has been manufactured in the past. The generated machine learning model can then be used to predict power consumption. This makes it possible to more accurately predict power consumption even for a product that has not been manufactured in the past, and to create an operation plan that falls within a desired power range. Therefore, with this operation plan creation system, a more appropriate operation plan can be created even when a production plan includes a product that has not been manufactured in the past.

[0034] FIG. 3 is a graph showing the relationship between the total power consumption of all the equipment and the power consumption of the electric kilns and the elapsed time when a product is manufactured in a conventional manner. In FIG. 3, the upper part shows a graph of the total power consumption of all the equipment, and the lower part shows a graph of the power consumption of each electric kiln. FIG. 4 is a graph showing the relationship between the total power consumption of all the equipment and the power consumption of each electric kiln and the elapsed time when a product is manufactured in accordance with the operation plan created by the operation plan creation system 10. FIG. 4 is a graph of a simulation. In FIG. 4, the upper part also shows a graph of the total power consumption of all the equipment, and the lower part shows a graph of the power consumption of each electric kiln. In FIGS. 3 and 4, the vertical axis represents power, and the horizontal axis represents the elapsed time. In the graphs at the bottom of each of Figures 3 and 4, in addition to the electric kilns 12a, 12b, and 12c, graphs for the other electric kilns 12d, 12e, 12f, 12g, 12h, 12i, 12j, 12k, 12l, and 12m included in the entire facility are also shown (the total number of electric kilns is 13).

[0035] 3 and 4, the power consumption of the electric kiln is not so different from the peak power value of each electric kiln when a product is manufactured in accordance with the operation plan created by the operation plan creation system 10. However, the peak power value X of the total power of all the equipment when a product is manufactured in the conventional manner is not so different from the peak power value of each electric kiln when a product is manufactured in accordance with the operation plan created by the operation plan creation system 10. １ In comparison with the peak power value X of the total power of all the facilities when manufacturing products in accordance with the operation plan created by the operation plan creation system 10, ２ is greatly reduced.

[0036] In this embodiment, the product feature data includes data on temperature transitions during product manufacturing, data on pressure transitions during product manufacturing, and data on time during product manufacturing. Therefore, more effective data can be used as product feature data for machine learning of power consumption. This allows for more accurate machine learning models to be generated, and more appropriate operation plans to be created.

[0037] In this embodiment, the operation plan creation unit 28 creates an operation plan that reflects the degree of deterioration of the equipment. Equipment deteriorates with use, and power consumption changes along with this deterioration. With this configuration, an operation plan can be created that takes into account changes in power consumption that change depending on the degree of deterioration of the equipment, thereby creating a more appropriate operation plan.

[0038] 5 is a graph showing the difference in power consumption before and after deterioration of an electric kiln. In FIG. 5, the vertical axis represents power consumption, and the horizontal axis represents elapsed time. Line 41 shows the change in power consumption before deterioration of the electric kiln, and line 42 shows the change in power consumption after deterioration of the electric kiln.

[0039] 5, even if the temperature of the electric kiln is increased at the same rate, the power consumption varies depending on the degree of deterioration, as shown by lines 41 and 42. In this embodiment, the operation plan creation unit 28 creates an operation plan that reflects the degree of deterioration of the electric kiln. Such an operation plan is more appropriate to the actual situation.

[0040] In this embodiment, the machine learning is performed by a random forest in the machine learning model generation unit 26. Such machine learning is more suitable for creating an operation plan.

[0041] In this embodiment, the facility includes an electric kiln having a kiln formed using carbon. With such an electric kiln, the amount of carbon decreases with continued use, and the power consumption varies and is not constant. Therefore, the operation plan creation system 10 configured as described above makes it possible to create an operation plan that corresponds to changes in power consumption.

[0042] In the present embodiment, in the operation plan creation system 10, the feature data includes data on heat curves during product manufacturing. The heat curve data during product manufacturing is very important as feature data accompanying changes in power consumption during product manufacturing. Therefore, by adopting such a configuration, a more appropriate operation plan can be created.

[0043] (Other Embodiments) In the above embodiment, the desired power range is a range in which peak power is minimized, but this is not limiting. The desired power range may be, for example, a range of a certain fixed power value, or may be set to several percent of the peak power, or the desired power may be different between the first half and second half of a week or between daytime and nighttime on a single day. Furthermore, the predetermined period is not limited to one week, and may be, for example, two weeks or one month.

[0044] Furthermore, in the above embodiment, the machine learning in the machine learning model generation unit is performed using a random forest, but this is not limited to this, and the machine learning in the machine learning model generation unit may use other methods such as a support vector machine or a neural network.

[0045] In the above embodiment, the product feature data includes all of the data on temperature transitions during product manufacturing, data on pressure transitions during product manufacturing, and data on the time taken to manufacture the product. However, the data is not limited to this and may include at least one of the data on temperature transitions during product manufacturing, data on pressure transitions during product manufacturing, and data on the time taken to manufacture the product. This allows more effective data to be used as product feature data for machine learning of power consumption. This allows for more accurate machine learning models to be generated, and more appropriate operation plans to be created.

[0046] It should be understood that the embodiments disclosed herein are illustrative in all respects and are not limiting in any respect. The scope of the present invention is defined not by the above description but by the claims, and it is intended to include all modifications within the meaning and scope of the claims.

[0047] 10 Operation plan creation system, 11 Server, 12a, 12b, 12c, 12d, 12e, 12f, 12g, 12h, 12i, 12j, 12k, 12l, 12m Electric kiln, 21 Database, 22 Reception unit, 23 Extraction unit, 24 First judgment unit, 25 Second judgment unit, 26 Machine learning model generation unit, 27 Power consumption prediction unit, 28 Operation plan creation unit, 29 Output unit, 41, 42 Lines.

Claims

1. An operation plan creation system that, when manufacturing a plurality of products using a plurality of pieces of equipment, creates an operation plan for the plurality of pieces of equipment so that the power consumption falls within a desired range within a predetermined period of time, comprising: a memory unit that stores feature data for each of the plurality of products that have been manufactured in the past, including data on the power consumption of each of the plurality of products when manufacturing the respective products; a reception unit that receives a request for creating an operation plan; an extraction unit that, when the reception unit receives the request for creating an operation plan, extracts data for the product from a production plan including data on the product to be manufactured; a first determination unit that determines whether the product extracted by the extraction unit matches one of the plurality of products stored in the memory unit; a second determination unit that, when the first determination unit determines that there is no match, determines whether a corresponding physical model can be generated for the product as a specific product; a machine learning model generation unit that, when the second determination unit determines that a corresponding model cannot be generated, generates a machine learning model corresponding to the specific product using at least a portion of the feature data for the plurality of products stored in the memory unit; and a power consumption prediction unit that predicts the power consumption of the specific product using the machine learning model of the specific product generated by the machine learning model generation unit. an operation plan creation unit that creates the operation plan using data on the power consumption of the specific product predicted by the power consumption prediction unit so that the power consumption falls within the desired power range within the specified period; and an output unit that outputs the operation plan created by the operation plan creation unit.

2. An operation plan creation system as described in claim 1, wherein the product feature data includes at least one of data on temperature changes during the manufacture of the product, data on pressure changes during the manufacture of the product, and data on time during the manufacture of the product.

3. An operation plan creation system according to claim 1 or claim 2, wherein the operation plan creation unit creates the operation plan by reflecting the degree of deterioration of the equipment.

4. An operation plan creation system according to any one of claims 1 to 3, wherein machine learning in the machine learning model generation unit is performed using a random forest.

5. An operation plan creation system according to any one of claims 1 to 4, wherein the facility includes an electric kiln having a kiln formed using carbon.

6. An operation plan creation system according to any one of claims 1 to 5, wherein the feature data includes heat curve data when manufacturing the product.

Images