Prediction device, prediction method, control program, manufacturing system, and manufacturing method

Abstract

Predict the amount of impurities contained in the propylene oxide produced. [Solution] The prediction device (10) includes a reception unit (11) that receives input of the amount of raw material to be supplied to the manufacturing device (20), and a prediction unit (12) that predicts the amount of impurities from the amount of raw material to be supplied using a prediction model (14) generated by machine learning using training data in which the amount of raw material to be supplied is the explanatory variable and the amount of impurities is the objective variable.

Description

Technical Field

[0004] ,

[0006] , , , ,

[0005] , , , , , ,

[0001] The present invention relates to a prediction device for predicting the amount of impurities contained in produced propylene oxide and the like.

Background Art

[0002] Propylene oxide (PO) is an organic compound often used as a raw material for various chemical products. Since propylene oxide containing a large amount of impurities cannot be used as a product, it is important to control the impurities contained in propylene oxide. For example, as described in Patent Documents 1 and 2, there are various methods for producing propylene oxide. However, these documents do not mention properly controlling the impurities.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Patent Document 2

[0007] To solve the aforementioned problems, a manufacturing apparatus according to one aspect of the present invention includes the prediction device and a manufacturing apparatus for producing propylene oxide, wherein the manufacturing apparatus includes an adjustment unit that adjusts the raw material supply amount based on the prediction result of the prediction device.

[0008] To solve the aforementioned problems, a prediction method according to one aspect of the present invention is a prediction method for predicting the amount of impurities contained in propylene oxide produced by the chlorohydrin method, comprising: a reception step of receiving input of the amount of raw material to be supplied to a production apparatus; and a prediction step of using a prediction model generated by machine learning with learning data in which the amount of raw material to be supplied to the production apparatus is the explanatory variable and the amount of impurities contained in the propylene oxide produced after a predetermined time has elapsed with the raw material to be supplied is the objective variable, to predict the amount of impurities contained in the propylene oxide produced after a predetermined time has elapsed from the amount of raw material received in the reception step.

[0009] To solve the aforementioned problems, a manufacturing method according to one aspect of the present invention includes a determination step of determining whether the amount of impurities predicted by the prediction method is less than or equal to a predetermined value, and an adjustment step of adjusting the raw material supply amount in the determination step so that the amount of impurities is less than or equal to a predetermined value. [Effects of the Invention]

[0010] According to one aspect of the present invention, the amount of impurities contained in the propylene oxide after manufacturing can be made known to managers or other personnel in advance. [Brief explanation of the drawing]

[0011] [Figure 1] This figure shows the manufacturing process in a manufacturing apparatus according to an embodiment of the present invention. [Figure 2] This is a functional block diagram showing the main components of a prediction device according to an embodiment of the present invention. [Figure 3] This figure shows an example of a learning device that generates predictive models. [Figure 4] This is a diagram illustrating an example of a prediction method in a prediction model. [Figure 5] This figure shows an example of the display on the display unit of the prediction device. [Figure 6] This is a flowchart showing the processing flow in the prediction device. [Figure 7] This is a diagram showing the main components of the manufacturing system. [Figure 8] This is a flowchart showing the processing flow in a manufacturing system. [Modes for carrying out the invention]

[0012] [Embodiment 1] The following describes in detail one embodiment of the present invention. The prediction device 10 according to this embodiment predicts the amount of impurities contained in propylene oxide produced by the chlorohydrin method from the amount of raw material supplied. The chlorohydrin method is a method of synthesis in which propylene is reacted with chlorine gas and water to produce propylene chlorohydrin (PCH), and hydrogen chloride is removed with a base such as sodium hydroxide, potassium hydroxide, or calcium hydroxide.

[0013] By being able to predict the amount of impurities contained in the produced propylene oxide from the raw material supply amount, the raw material supply amount can be adjusted so that the amount of impurities becomes below a predetermined value. Thereby, it is possible to suppress the production of propylene oxide with a large amount of impurities. Hereinafter, as the raw material supply amount, the supply water amount which is the supply amount of water will be described as an example. Note that it is also possible to use the supply amount of chlorine gas instead of water as the raw material supply amount.

[0014] 〔Process for producing propylene oxide〕 First, referring to FIG. 1, the outline of the process for producing propylene oxide will be described. FIG. 1 is a diagram showing the outline of a production apparatus 20 for producing propylene oxide. As shown in FIG. 1, in the production apparatus 20, propylene oxide is produced through four processes: a reaction process 21, a saponification process 22, a low-boiling process 23, and a rectification process 24.

[0015] The reaction process 21 includes one or more reactors (the first reactor 21-1, the second reactor 21-2, the third reactor 21-3,..., the nth reactor 21-n). The water supplied to the first reactor 21-1 is sequentially supplied to the second reactor 21-2, the third reactor 21-3,..., the nth reactor 21-n. Chlorine gas and propylene gas are supplied to each reactor, whereby PCH is generated.

[0016] In the saponification process 22, saponification of PCH generated in the reactor is performed in the saponification tank 22A. Specifically, PCH and a base are supplied from the upper part of the saponification tank 22A, and steam is supplied from the lower part of the saponification tank 22A, and they are brought into countercurrent contact in the saponification tank 22A to generate propylene oxide.

[0017] In the low-boiling process 23, in the low-boiling tower 23A, organic compounds having a boiling point lower than that of propylene oxide are removed from the propylene oxide generated in the saponification process 22.

[0018] In the rectification process 24, rectification of the propylene oxide from which organic compounds have been removed in the low-boiling process 23 is performed in the rectification tower 24A.

[0019] The propylene oxide that has gone through the rectification process 24 becomes the final product manufactured by the manufacturing apparatus 20.

[0020] [Prediction device] Next, the prediction device 10 according to this embodiment will be described with reference to Figure 2. Figure 2 is a functional block diagram showing the main components of the prediction device 10. As shown in Figure 2, the prediction device 10 includes a reception unit 11, a prediction unit 12, a display unit 13, and a prediction model 14.

[0021] The reception unit 11 receives input from the user regarding the amount of water to be supplied to the manufacturing device 20.

[0022] The prediction unit 12 uses the prediction model 14 to predict the amount of impurities contained in the propylene oxide produced during the time it passes through the reaction process 21 (after a predetermined time has elapsed), based on the amount of water supplied by the receiving unit 11. Impurities are chlorinated organic substances produced by a direct gas-phase reaction between the raw material chlorine, propylene, and impurities contained in propylene, and examples include 1-chloropropane.

[0023] In this embodiment, since it takes a certain amount of time (a predetermined time) to produce propylene oxide in the manufacturing apparatus 20, the amount of impurities contained in the propylene oxide produced after the predetermined time has elapsed is predicted based on the current supply water volume. However, the prediction by the prediction device 10 is not limited to the amount of impurities contained in the propylene oxide produced after the predetermined time has elapsed. It may be appropriately changed according to the manufacturing time in the manufacturing apparatus 20, or it may predict the amount of impurities at a desired time regardless of the manufacturing time in the manufacturing apparatus 20.

[0024] The display unit 13 is a display device that displays the prediction results from the prediction unit 12.

[0025] [Predictive Model] Next, the prediction model 14 will be described with reference to Figures 3 and 4. Figure 3 is a diagram showing an example of a learning device 30 that generates the prediction model 14. As shown in Figure 3, the learning device 30 includes a learning unit 31 that performs machine learning using a dataset 40. The dataset 40 is, as an example, data that associates the amount of supplied water with the amount of impurities contained in the propylene oxide produced after a predetermined time has elapsed.

[0026] The learning unit 31 performs machine learning with the supply water volume included in the dataset 40 as the explanatory variable and the amount of impurities as the dependent variable, and generates a predictive model 14. The predictive model 14 is a predictive model that takes the supply water volume as input and outputs the amount of impurities contained in the propylene oxide produced after a predetermined time has elapsed.

[0027] In the prediction model 14, the temperature of the next reactor is predicted based on the temperature of the previous reactor, in the order of the multiple reactors, for example, the temperature of the next reactor is predicted based on the temperature of the previous reactor, for example, in the order of the reaction process 21, and the amount of impurities contained in the propylene oxide produced is predicted.

[0028] Since reactor temperature has a significant impact on the amount of impurities, this method allows us to predict the amount of impurities based on the temperature of each reactor, thereby improving the accuracy of the prediction.

[0029] Although only the supply water volume is listed as an explanatory variable here, various measurement values ​​that can be measured in the manufacturing process by the manufacturing apparatus 20 may be added as explanatory variables. For example, raw material purity and supply volume can be appropriately selected and used as explanatory variables.

[0030] By using a machine learning-based predictive model that incorporates measured data to predict impurity levels, the accuracy of the prediction can be improved.

[0031] [Example Display] Next, an example of the display in the display unit 13 will be explained with reference to Figure 5. Figure 5 is a diagram showing an example of the display in the display unit 13. As shown in 501 of Figure 5, the display unit 13 shows the current total chlorine value and the current supply water volume in column 511, and the trend of the total chlorine value up to the present in a graph in column 514. The total chlorine value indicates the amount of impurities mentioned above.

[0032] [Processing flow] Next, the processing flow in the prediction device 10 will be explained with reference to Figure 6. Figure 6 is a flowchart showing the processing flow in the prediction device 10. As shown in Figure 6, in the prediction device 10, when the reception unit 11 receives the input of the water supply amount (S101, reception step), the prediction unit 12 uses the prediction model 14 to predict the total chlorine value after a predetermined time has elapsed (S102, prediction step). Then, the prediction result from the prediction unit 12 is displayed on the display unit 13 (S103). This concludes the processing flow in the prediction device 10.

[0033] [Manufacturing System] Next, the manufacturing system 1, including the prediction device 10, will be described with reference to Figures 7 and 8. Figure 7 is a diagram showing the main components of the manufacturing system 1. As shown in Figure 7, the manufacturing system 1 includes the prediction device 10 and the manufacturing device 20A.

[0034] The manufacturing apparatus 20A further includes an adjustment unit 25 in addition to the manufacturing apparatus 20 described above. The adjustment unit 25 adjusts the amount of water supplied to the manufacturing apparatus 20A based on the prediction results of the prediction device 10. For example, the adjustment unit 25 adjusts the amount of water supplied to the manufacturing apparatus 20A so that the amount of impurities predicted by the prediction device 10 is below a predetermined value.

[0035] Referring to Figure 8, the processing flow in manufacturing system 1 will be explained. Figure 8 is a flowchart of the processing flow in manufacturing system 1. As shown in Figure 8, in manufacturing system 1, when the prediction device 10 receives input for the amount of water to be supplied (S201), the prediction model 14 is used to predict the amount of impurities after a predetermined time has elapsed (S202). The adjustment unit 25 of manufacturing apparatus 20A determines whether the amount of impurities predicted by the prediction device 10 is less than or equal to a predetermined value (S203, determination step). If it exceeds the predetermined value (NO in S203), the amount of water to be supplied is changed (for example, increased by a specified amount) (S204, adjustment step), and the changed amount of water to be supplied is input to the prediction device 10.

[0036] Then, steps S201 to S204 are repeated, and in step S203, if the amount of impurities falls below a predetermined value (YES in S203), the adjustment unit 25 sets the amount of water supplied to the manufacturing apparatus 20A to the amount of water supplied when the amount of impurities falls below a predetermined value (S205). The above is the processing flow in manufacturing system 1.

[0037] [Examples of implementation using software] The function of the prediction device 10 (hereinafter referred to as "the device") is a program that causes the device to function as a computer, and can be realized by a program that causes the computer to function as each control block of the device (particularly the prediction unit 12).

[0038] In this case, the device includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., memory) as hardware for executing the program. By executing the program using this control device and storage device, the functions described in each of the embodiments above are realized.

[0039] The above program may be recorded on one or more computer-readable recording media, not temporary ones. These recording media may or may not be provided by the above device. In the latter case, the program may be supplied to the above device via any wired or wireless transmission medium.

[0040] Furthermore, some or all of the functions of each of the above control blocks can also be realized by logic circuits. For example, an integrated circuit in which logic circuits functioning as each of the above control blocks are formed is also included in the scope of the present invention. In addition, it is also possible to realize the functions of each of the above control blocks by, for example, a quantum computer.

[0041] Furthermore, each process described in the above embodiments may be performed by AI (Artificial Intelligence). In this case, the AI ​​may operate on the control device described above, or it may operate on other devices (for example, an edge computer or a cloud server).

[0042] The present invention is not limited to the embodiments described above, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present invention.

[0043] 〔summary〕 A prediction device according to Embodiment 1 of the present invention is a prediction device for predicting the amount of impurities contained in propylene oxide produced by the chlorohydrin method, comprising: a reception unit that receives input of the amount of raw material supplied to a production device; and a prediction unit that uses a prediction model generated by machine learning using training data in which the amount of raw material supplied to the production device is the explanatory variable and the amount of impurities contained in the propylene oxide produced after a predetermined time has elapsed from the raw material supply amount received by the reception unit, to predict the amount of impurities contained in the propylene oxide produced after a predetermined time has elapsed from the raw material supply amount. The inventors of the present invention have discovered that there is a correlation between the amount of raw material supplied to a production device and the amount of impurities contained in the propylene oxide after production. With the above configuration, the amount of impurities contained in the propylene oxide produced after a predetermined time has elapsed from the raw material supply amount can be predicted using a prediction model obtained by machine learning using the amount of raw material supplied and the amount of impurities as training data. Therefore, the amount of impurities contained in the propylene oxide after production can be made known to managers, etc., in advance.

[0044] In the prediction device according to embodiment 2 of the present invention, in embodiment 1, the amount of raw material supplied may be the amount of water supplied to the manufacturing device. The inventors of the present invention have discovered that there is a correlation between the amount of water, in particular, of the amount of raw material supplied to the manufacturing device and the amount of impurities contained in the propylene oxide after manufacturing. With the above configuration, it is possible to predict the amount of impurities contained in the propylene oxide produced after a predetermined time has elapsed from the amount of water supplied to the manufacturing device using a prediction model obtained by machine learning with the amount of supplied water and the amount of impurities as training data. This makes it possible to predict the amount of impurities contained in the propylene oxide after manufacturing with high accuracy.

[0045] In the prediction device according to aspect 3 of the present invention, in aspect 1 or 2, the prediction model may be machine-trained using training data that includes one or more measured values ​​measured during the production of propylene oxide in the production device as explanatory variables. The inventors of the present invention have discovered that the accuracy of the prediction model can be improved by including measured values ​​measured during the production of propylene oxide, which change with the instrumental variable, in the training data when generating the prediction model. With the above configuration, the amount of impurities can be predicted using a prediction model that has been machine-trained using training data including measured values, thus improving the accuracy of the prediction.

[0046] The prediction device according to aspect 4 of the present invention may, in any of aspects 1 to 3, include a plurality of reactors, and in the prediction model, in the process sequence of the plurality of reactors, predict the temperature of the first reactor, which is the first reactor, based on the amount of water supplied to the first reactor, and thereafter, in the process sequence, predict the temperature of the next reactor based on the temperature of the previous reactor, and predict the amount of impurities based on the temperature of each reactor. Since the temperature of the reactor has a large influence on the amount of impurities, the above configuration allows for prediction of the amount of impurities based on the temperature of each reactor, thereby improving the accuracy of the prediction.

[0047] The prediction device according to aspect 5 of the present invention may include a display unit that displays the prediction results from the prediction unit, in any of aspects 1 to 4. With the above configuration, administrators and others can visually recognize the prediction results.

[0048] A manufacturing system according to embodiment 6 of the present invention includes the prediction device and a manufacturing apparatus for producing propylene oxide, wherein the manufacturing apparatus includes an adjustment unit that adjusts the raw material supply amount based on the prediction result of the prediction device. With this configuration, the raw material supply amount can be adjusted based on the prediction result of the prediction device, so that propylene oxide with a desired amount of impurities can be produced.

[0049] A prediction method according to aspect 7 of the present invention is a prediction method for predicting the amount of impurities contained in propylene oxide produced by the chlorohydrin method, comprising: a reception step of receiving input of the amount of raw material to be supplied to a production apparatus; and a prediction step of predicting the amount of impurities contained in the propylene oxide produced after a predetermined time has elapsed from the amount of raw material received in the reception step, using a prediction model generated by machine learning with learning data in which the amount of raw material to be supplied to the production apparatus is the explanatory variable and the amount of impurities contained in the propylene oxide produced after a predetermined time has elapsed from the raw material supply amount is the objective variable.

[0050] A manufacturing method according to embodiment 8 of the present invention includes a determination step of determining whether the amount of impurities predicted by the prediction method described in embodiment 7 is less than or equal to a predetermined value, and an adjustment step of adjusting the raw material supply amount in the determination step so that the amount of impurities is less than or equal to a predetermined value.

[0051] Each aspect of the present invention may be implemented by a computer, in which case a control program for the prediction device that enables the computer to implement the prediction device by operating the computer as each part (software element) of the prediction device, and a computer-readable recording medium on which the program is recorded, also fall within the scope of the present invention. [Explanation of Symbols]

[0052] 1. Manufacturing System 10 Prediction device 11 Reception Department 12 Prediction Section 13 Display section 14 Predictive Models 20, 20A manufacturing equipment 21 Reaction Steps 21-1 Reactor No. 1 21-2 Reactor No. 2 21-3 Reactor No. 3 21-n reactor 22 Saponification process 22A Saponification tank 23 Low boiling process 23A Low boiling tower 24 Rectification process 24A Rectification tower 25 Adjustment part 30 Learning device 31. Learning Department 40 datasets

Claims

1. A predictive device for predicting the amount of impurities contained in propylene oxide produced by the chlorohydrin method, A receiving unit that receives input for the amount of raw material to be supplied to the manufacturing equipment, A prediction device comprising: a prediction unit that predicts the amount of impurities contained in the propylene oxide produced after a predetermined time has elapsed, based on the amount of raw material supplied to the manufacturing apparatus, using a prediction model generated by machine learning with training data in which the amount of raw material supplied is an explanatory variable and the amount of impurities contained in the propylene oxide produced after a predetermined time has elapsed with the raw material supplied is an objective variable, based on the amount of raw material supplied received by the receiving unit.

2. The prediction device according to claim 1, wherein the amount of raw material supplied is the amount of water supplied to the manufacturing apparatus.

3. The prediction apparatus according to claim 1, wherein the prediction model is machine-trained using training data in which the explanatory variables include one or more measured values ​​measured during the production of the propylene oxide in the production apparatus.

4. The manufacturing apparatus includes a plurality of reactors, The prediction device according to claim 2, wherein the prediction model predicts the temperature of the first reactor based on the amount of water supplied to the first reactor, which is the first reactor, in the order of the processes of the plurality of reactors, and thereafter predicts the temperature of the next reactor based on the temperature of the previous reactor in the order of the processes, and predicts the amount of impurities based on the temperature of each reactor.

5. The prediction device according to claim 1, further comprising a display unit for displaying the prediction results from the prediction unit.

6. The prediction device according to claim 1, A manufacturing apparatus for producing propylene oxide, The manufacturing apparatus is a manufacturing system comprising an adjustment unit that adjusts the raw material supply amount based on the prediction results of the prediction device.

7. A method for predicting the amount of impurities contained in propylene oxide produced by the chlorohydrin method, A reception step that accepts input for the amount of raw material to be supplied to the manufacturing equipment, A prediction method comprising: a prediction step in which a prediction is made by machine learning using training data in which the amount of raw material supplied to the manufacturing apparatus is an explanatory variable and the amount of impurities contained in the propylene oxide manufactured after a predetermined time has elapsed with the raw material supply is an objective variable, and the amount of impurities contained in the propylene oxide manufactured after a predetermined time has elapsed with the raw material supply is used to predict the amount of impurities contained in the propylene oxide manufactured after a predetermined time has elapsed from the raw material supply amount received in the acceptance step.

8. A determination step of determining whether the amount of impurity predicted by the prediction method described in claim 7 is less than or equal to a predetermined value, A method for producing propylene oxide, comprising: an adjustment step in the determination step, which adjusts the amount of raw material supplied so that the amount of impurities is less than or equal to a predetermined value.

9. A control program for causing a computer to function as a prediction device according to claim 1, wherein the control program causes the computer to function as the prediction unit.

Images