Gas amount prediction method, plant operation method, and gas amount prediction device

Abstract

A gas amount prediction method is a method of predicting an amount of gas generated in a plant, including: a generation amount calculation step in which a generation amount of gas that is actually usable is calculated using a learning model that learns a relationship between a generation amount of gas and a usage amount of gas in past operation data.

Description

TECHNICAL FIELD

[0001] The present application relates to a gas amount prediction method, an operation method of a plant, and a gas amount prediction device. BACKGROUND

[0002] In the energy management process, for example, steam generated in a plant is used for a wide range of uses such as heating in a production process, heating / humidification of air conditioning, and the like. The steam generated in the plant is transported to a power plant, another plant, or the like through a pipe or the like, but becomes liquid due to heat dissipation. Therefore, the amount of steam that can actually be used becomes small with respect to the amount of steam generated in the plant. Therefore, in Patent Literatures 1 and 2, a method for grasping the amount of steam that can actually be used is proposed.

[0003] In Patent Literature 1, a method is disclosed in which a correction coefficient is calculated based on a pipe internal temperature of a steam pipe calculated from a pressure of steam, and a value related to a measured evaporation amount of steam is corrected using the correction coefficient, whereby a loss of the steam pipe is determined. In addition, in Patent Literature 2, a method is disclosed in which a correction coefficient is calculated based on a pipe internal heat transfer rate corresponding to a liquid film thickness of the steam pipe that varies due to a flow rate of steam, and a measured value of the loss of the steam pipe is corrected.

[0004] PRIOR ART DOCUMENTS

[0005] PATENT LITERATURE

[0006] Patent Literature 1: Japanese Patent No. 6303473

[0007] Patent Literature 2: Japanese Patent No. 6264898 SUMMARY

[0008] PROBLEMS TO BE SOLVED BY THE INVENTION

[0009] However, the method disclosed in Patent Literature 1 has a problem that it cannot be used in a situation where it is difficult to measure the pressure of steam. In addition, the method disclosed in Patent Literature 2 also has a problem that it cannot be used in a situation where it is difficult to measure the flow rate of steam.

[0010] The present application was made in view of the above problems, and aims to provide a gas amount prediction method, an operation method of a plant, and a gas amount prediction device that can easily predict the amount of steam that can actually be used in the energy management process of a plant.

[0011] METHOD FOR SOLVING THE PROBLEM

[0012] To solve the above problems and achieve the object, the gas amount prediction method of the present application is a gas amount prediction method of predicting the amount of gas generated in a factory, which includes: a generation amount calculation step of calculating the generation amount of the gas that can actually be used using a learning model that learns the relationship between the generation amount of the gas and the usage amount of the gas in past operation data.

[0013] Further, in the gas amount prediction method of the present application, in the above invention, in the generation amount calculation step, the usage rate of the gas with respect to the generation amount of the gas is calculated using the learning model, and the generation amount of the gas that can actually be used is calculated by multiplying the generation amount of the gas obtained at the time of operation by the usage rate of the gas.

[0014] Further, in the gas amount prediction method of the present application, in the above invention, in the generation amount calculation step, the learning of the learning model is performed using operation data under a climate condition similar to the present climate condition in the past operation data.

[0015] Further, in the gas amount prediction method of the present application, in the above invention, the learning model is learned by regression analysis or autoregressive moving average.

[0016] Further, in the gas amount prediction method of the present application, in the above invention, a deviation calculation step is included after the generation amount calculation step, in which, for the generation amount of the gas calculated in the generation amount calculation step, a deviation depending on a characteristic amount related to the gas is calculated, and the deviation is prompted.

[0017] Further, in the gas amount prediction method of the present application, in the above invention, the characteristic amount is sea level pressure.

[0018] Further, in the gas amount prediction method of the present application, in the above invention, the gas is steam, and in the generation amount calculation step, the generation amount of the steam that can actually be used is calculated taking into account the loss of steam due to heat dissipation.

[0019] To solve the above problems and achieve the object, the operation method of the factory of the present application changes the operation plan of the factory based on the generation amount of the gas that can actually be used predicted by the gas amount prediction method.

[0020] Further, in the operation method of the factory of the present application, in the above invention, the operation plan of the factory is changed based on a characteristic amount related to the gas, a deviation depending on the characteristic amount, and the generation amount of the gas that can actually be used predicted by the gas amount prediction method.

[0021] To solve the above problems and achieve the object, the gas amount prediction device of the present application is a gas amount prediction device that predicts the amount of gas generated in a factory, and has: a generation amount calculation section that calculates the generation amount of the gas that is actually usable using a learning model that learns the relationship between the generation amount of the gas and the usage amount of the gas in past operation data.

[0022] Effects of Invention

[0023] According to the present application, without the need for measurement of the pressure, flow rate, etc. of the gas and complicated calculation, the amount of gas that is actually usable in a factory can be simply predicted using the generation amount and usage amount of the gas in past operation data. BRIEF DESCRIPTION OF DRAWINGS

[0024] Figure 1 is a block diagram showing the schematic configuration of a gas amount prediction device of an embodiment of the present application.

[0025] Figure 2 is a flowchart showing the flow of a gas amount prediction method of an embodiment of the present application.

[0026] Figure 3 is a graph showing the relationship between the number of operation data (teacher data) used for prediction and the prediction error (RMSE) in the gas amount prediction method of an embodiment of the present application.

[0027] Figure 4 is a scatter plot showing the relationship between the steam usage amount and the steam generation amount before correction in an example of the gas amount prediction method of an embodiment of the present application.

[0028] Figure 5 is a scatter plot showing the relationship between the steam usage amount and the steam generation amount after correction in an example of the gas amount prediction method of an embodiment of the present application.

[0029] Figure 6 is a histogram showing a comparison of the error of the steam generation amount before and after correction in an example of the gas amount prediction method of an embodiment of the present application.

[0030] Figure 7 is a scatter plot showing the relationship between the sea level atmospheric pressure and the correction error of the steam generation amount in an example of the gas amount prediction method of an embodiment of the present application.

[0031] Figure 8 is a graph showing the correction error of the steam generation amount according to the size of the sea level atmospheric pressure in an example of the gas amount prediction method of an embodiment of the present application. DETAILED DESCRIPTION

[0032] A gas amount prediction method, an operation method of a plant, and a gas amount prediction device according to an embodiment of the present application will be described with reference to the drawings.

[0033] (Gas amount prediction device)

[0034] Reference Signs Figure 1 A configuration of a gas amount prediction device 1 according to an embodiment of the present application will be described. The gas amount prediction device 1 is a device that predicts an amount of gas generated in a plant, specifically, an amount of gas that can actually be used in the plant, in an energy management process of the plant. As the gas, steam generated in the plant, generated gas, or the like can be listed. In the present embodiment, an example in which the gas is steam will be described.

[0035] The gas amount prediction device 1 is realized by a general-purpose information processing device such as a personal computer or a workstation, and has an input section 10, a database (DB) 20, an arithmetic section 30, and a display section 40.

[0036] The input section 10 is an input unit for the arithmetic section 30, and is realized by an input device such as a keyboard, a mouse pointer, a number key, or the like. The past operation data (performance data) is stored in the database 20.

[0037] The arithmetic section 30 is realized by, for example, a processor constituted by a CPU (Central Processing Unit) or the like and a memory (main storage section) constituted by a RAM (Random Access Memory), a ROM (Read Only Memory), or the like. The arithmetic section 30 executes a program loaded to a work area of the main storage section, and controls each configuration section and the like by the execution of the program, thereby realizing a function in accordance with a predetermined purpose. The arithmetic section 30 functions as a generation amount calculation section 31 and a deviation calculation section 32 by the execution of the program.

[0038] The generation amount calculation section 31 calculates an actually usable steam generation amount (an amount of usable steam) using a learning model that learns a relationship between a steam generation amount (hereinafter referred to as "steam generation amount") and a steam usage amount (hereinafter referred to as "steam usage amount") in the past operation data. Here, the "actually usable steam generation amount" refers to a steam generation amount that takes into account a loss (steam loss) caused by, for example, liquefaction of steam due to heat dissipation. In addition, the "actually usable steam generation amount" indicates an amount of steam obtained by subtracting the above steam loss from an amount of steam generated in the plant.

[0039] Specifically, the generation amount calculation section 31 calculates a usage rate of steam with respect to the steam generation amount (hereinafter referred to as "steam usage rate") using a learning model that learns a relationship between the steam generation amount and the steam usage amount in past operation data. The steam usage rate can be expressed by "steam usage amount / steam generation amount", for example. In addition, learning of the learning model can be performed using a method such as regression analysis or autoregressive moving average. In addition, as a specific method of regression analysis, for example, the least squares method, which is one type of linear regression, or the like can be cited, but a mechanical learning model such as a neural network or a decision tree can also be used.

[0040] The generation amount calculation section 31 uses learning data under a climate condition similar to the current climate condition as learning data (teacher data) for learning. That is, the generation amount calculation section 31 uses the steam generation amount and the steam usage amount contained in operation data under a climate condition similar to the current climate condition among past operation data as teacher data, and performs learning of the learning model. The "operation data under a climate condition similar to the current climate condition" indicates, for example, recent operation data that is not too far in time.

[0041] The generation amount calculation section 31 can use, for example, the steam generation amount and the steam usage amount contained in operation data from "1 hour ago to 40 hours ago" as teacher data. In this way, by performing learning of the learning model using as much recent operation data as possible, the prediction accuracy of the steam generation amount is improved. Then, the generation amount calculation section 31 calculates the actual usable steam generation amount by multiplying the steam generation amount obtained (predicted or measured) at the time of operation by the steam usage rate.

[0042] The deviation calculation section 32 calculates a deviation depending on a characteristic amount related to steam with respect to the steam generation amount calculated by the generation amount calculation section 31. Then, the deviation calculation section 32 prompts the calculated deviation to the operator (user) through the display section 40. As the "characteristic amount related to steam", for example, sea level atmospheric pressure, air temperature, humidity, and the like can be cited. Here, the "deviation" is a statistical amount that evaluates the standard deviation (σ), 2σ, or the variance, RMSE, or the like, which is called a deviation, of the difference between the steam generation amount and the steam usage amount calculated using the learning model of the steam usage rate. Here, as an example, a case where 1σ is used will be described.

[0043] (Gas amount prediction method)

[0044] Reference Figure 2 The gas amount prediction method of the present embodiment will be described. Note that the learning method of the learning model is implemented mainly by the generation amount calculation section 31 and the deviation calculation section 32 of the arithmetic section 30.

[0045] First, the generation amount calculation section 31 reads past operation data (steam generation amount and steam usage amount) required for the processing from the database 20 (step S1). Next, the generation amount calculation section 31 calculates a steam usage rate using a regression analysis or an autoregressive moving average or the like (step S2).

[0046] Next, the generation amount calculation section 31 calculates an actually usable steam generation amount by multiplying the steam generation amount obtained at the time of the operation by the steam usage rate (step S3). Next, the deviation calculation section 32 calculates a deviation depending on a characteristic amount (for example, sea level atmospheric pressure, air temperature, humidity, or the like) related to the steam with respect to the steam generation amount calculated by step S3. Then, the deviation calculation section 32 prompts the operator of the deviation through the display section 40 (step S4).

[0047] The gas amount prediction device 1 and the gas amount prediction method according to the present embodiment can simply predict the amount of the gas actually usable in the plant using the generation amount and the usage amount of the gas in the past operation data without requiring measurement of the pressure, the flow rate, or the like of the gas and complicated calculation.

[0048] In addition, the gas amount prediction device 1 and the gas amount prediction method according to the present embodiment calculate and prompt the deviation depending on the characteristic amount (for example, sea level atmospheric pressure, air temperature, humidity, or the like) related to the steam in addition to the prediction of the steam generation amount actually usable. Thereby, the operator can make a more accurate operation plan of the steam, and the reliability of the energy management is improved.

[0049] (Operation method of plant)

[0050] The above gas amount prediction method can also be applied to the operation method of the plant. In this case, the operation plan of the plant is changed based on the generation amount of the gas (steam) actually usable predicted by the above gas amount prediction method, whereby the operation method of the gas in the plant is changed.

[0051] In addition, in the operation method of the plant, the operation plan of the plant can also be changed based on the generation amount of the gas actually usable predicted by the above gas amount prediction method and the characteristic amount (for example, sea level atmospheric pressure, air temperature, humidity, or the like) related to the above gas and the deviation depending on the characteristic amount.

[0052] Embodiment

[0053] Reference Figures 3-8 An embodiment of the gas amount prediction method according to the present embodiment is described. In the present embodiment, steps S1 to S4 of the above gas amount prediction method are implemented, and the effects or the like thereof are verified.

[0054] First, linear regression analysis (least square method) is performed using the operation data from 1 hour ago to 40 hours ago every 24 hours, and the steam usage rate is calculated. The explanatory variable at the time of linear regression analysis is set to the steam generation amount, and the target variable is set to the steam usage amount.

[0055] Here, Figure 3 is a graph showing the relationship between the number of operation data used for prediction (teacher data) and the prediction error (Root Mean Square Error: RMSE). As shown in the graph, the prediction accuracy of the steam generation amount depends on the number of operation data, and the prediction accuracy varies depending on the operation data used up to which time. Therefore, as shown in the graph, the operation data used for prediction is evaluated in advance, and the number of operation data used for prediction is determined. In the present embodiment, the operation data from 1 hour ago to 40 hours ago, which has the highest accuracy, is used. Note that the verification period of the operation data is 3 months.

[0056] Next, the actually usable steam generation amount is calculated by multiplying the steam generation amount obtained at the time of operation by the steam usage rate. Note that the actually usable steam generation amount is actually calculated by multiplying the predicted value of the steam generation amount calculated in advance by the steam usage rate, and in the present embodiment, the actual performance value of the steam generation amount is used assuming an ideal situation in which the prediction of the steam generation amount is 100%.

[0057] Here, Figure 4 is a scatter plot showing the relationship between the steam usage amount and the steam generation amount before correction using the steam usage rate. As shown in the graph, before correction of the steam generation amount using the steam usage rate, a distribution is formed in which the steam usage amount is low with respect to the steam generation amount.

[0058] On the other hand, Figure 5 is a scatter plot showing the relationship between the steam usage amount and the steam generation amount (usable steam amount) after correction using the steam usage rate. As shown in the graph, after correction of the steam generation amount using the steam usage rate, the steam generation amount is correlated with the steam usage amount.

[0059] In addition, Figure 6 is a bar graph comparing the error of the steam generation amount before and after correction using the steam usage rate with RMSE. As shown in the graph, it is known that by correcting the steam generation amount using the steam usage rate, the accuracy is improved by 80%. In this way, according to the present embodiment, for example, at the time of 15 o'clock the previous day, the steam generation amount (usable steam amount) of the next day can be obtained, and therefore, the operator can make a more accurate steam use plan.

[0060] Figure 7is a scatter diagram showing the relationship between the sea level atmospheric pressure and the correction error of the steam generation amount. Note that the "correction error of the steam generation amount" indicates a value obtained by subtracting the corrected steam generation amount from the steam usage amount. As shown in the diagram, it is known that the deviation of the correction error is small when the sea level atmospheric pressure is low, and the deviation of the correction error becomes large as the sea level atmospheric pressure becomes high.

[0061] Figure 8 is a diagram showing the correction error of the steam generation amount depending on the magnitude of the sea level atmospheric pressure in the form of a probability distribution. As shown in the diagram, in the case of A (sea level atmospheric pressure < 1000 hpa) where the sea level atmospheric pressure is low, the deviation of the correction error is small, and thus a steep probability distribution is formed. On the other hand, in the case of B (sea level atmospheric pressure > 1020 hpa) where the sea level atmospheric pressure is high, the deviation of the correction error of the steam generation amount is large, and thus a gentle probability distribution is formed.

[0062] In the gas amount prediction method of the present embodiment, by concentrating the deviations shown in the above-described Figure 4 , Figure 5 , Figure 7 and Figure 8 on one screen and displaying them on the display section 40, it is possible to guide the operator on the reliability of the corrected steam generation amount (usable steam amount). Then, the operator who has received such guidance will, for example, in the case where the sea level atmospheric pressure is high, adjust (delay or advance) the start time of the RH (Ruhrstahl Heraeus) process in which steam is used in large amounts in the secondary refining of steelmaking, due to the deviation of the correction error of the steam generation amount becoming large. Thereby, it is possible to achieve stabilization of the steam use. Note that in the RH process, in addition to backflowing the molten steel with Ar gas to perform degassing, oxygen is injected into the molten steel to perform decarburization, and in order to promote this reaction, steam is discharged to the outside of the device to depressurize the vacuum tank.

[0063] In this way, the deviation of the correction error of the steam generation amount is evaluated with the sea level atmospheric pressure, which is a specific characteristic quantity, and based on the value of this characteristic quantity, the operation conditions (for example, the operation start time, the operation end time, the steam usage amount, and the like) of the specific process in which steam is used in large amounts are changed. Thereby, it is possible to make the steam usage amount in the entire plant not exceed the generation amount, and to make the steam use efficient and stable.

[0064] The gas amount prediction method, the plant operation method, and the gas amount prediction device of the present application have been described in detail above through the specific embodiments and examples, but the gist of the present application is not limited to these descriptions, and needs to be broadly interpreted based on the descriptions of the claims. In addition, solutions after various modifications, changes, and the like based on these descriptions are also included in the gist of the present application.

[0065] For example, in the gas amount prediction method, the plant operation method, and the gas amount prediction device of the present embodiment, the description is made on the premise that the gas is steam, but it can also be applied to a case where the gas is a generated gas.

[0066] Symbol explanation

[0067] 1 Gas amount prediction device

[0068] 10 Input unit

[0069] 20 Database (DB)

[0070] 30 Calculation unit

[0071] 31 Generation amount calculation unit

[0072] 32 Deviation calculation unit

[0073] 40 Display unit

Claims

1. A gas amount prediction method that is a gas amount prediction method that predicts an amount of steam generated in a plant, comprising: a generation amount calculation step in which a generation amount of steam that is actually usable, which takes into account a loss of steam due to heat dissipation, is calculated using a learning model that learns a relationship between the generation amount of steam and a usage amount of steam in past operation data.

2. The gas amount prediction method according to claim 1, wherein In the generation amount calculation step, the usage rate of steam with respect to the generation amount of steam is calculated using the learning model, and the generation amount of steam that is actually usable is calculated by multiplying the generation amount of steam obtained at the time of operation by the usage rate of steam.

3. The gas amount prediction method according to claim 1 or claim 2, wherein, In the generation amount calculation step, learning of the learning model is performed using operation data under a climate condition similar to a present climate condition in the past operation data.

4. The gas amount prediction method according to claim 1 or claim 2, wherein, The learning model is learned by regression analysis or autoregressive moving average.

5. The gas amount prediction method according to claim 1 or claim 2, wherein, a deviation calculation step is included after the generation amount calculation step, in which, for the generation amount of steam calculated in the generation amount calculation step, a deviation that depends on a characteristic amount related to the steam is calculated, and the deviation is prompted.

6. The gas amount prediction method according to claim 5, wherein The characteristic amount is sea level atmospheric pressure.

7. A method of operating a factory, wherein, Based on the generation amount of steam that is actually usable predicted by the gas amount prediction method according to any one of claims 1 to 6, a method of using the steam in the plant is changed, whereby an operation plan of the plant is changed.

8. The method of operating a factory of claim 7, wherein, Based on a characteristic amount related to steam, a deviation that depends on the characteristic amount, and the predicted generation amount of steam that is actually usable, an operation plan of a plant is changed.

9. A gas amount prediction device that is a gas amount prediction device that predicts an amount of steam generated in a plant, comprising: a generation amount calculation section that calculates a generation amount of steam that is actually usable, which takes into account a loss of steam due to heat dissipation, using a learning model that learns a relationship between the generation amount of steam and a usage amount of steam in past operation data.

Images