Chemical plant management device, chemical plant management system, and chemical plant management method

Abstract

To specify a relation between a raw material or an operating condition and a production result in a chemical plant production.SOLUTION: A chemical plant management device comprises: a storage unit for storing information about a first product produced by charging a first raw material into a first reactor and using a first operating condition in a chemical plant; an input unit for inputting a second product produced in a second reactor; and a control unit for generating a plurality of property groups each containing a plurality of first products having a predetermined range of properties based on the information about the first product, wherein a predetermined property group including the second product out of the plurality of property groups is identified based on the property of the second product and a range of properties of the plurality of property groups, and a second raw material and a second operating condition are specified for producing the input second product in the second reactor.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a chemical plant management device, a chemical plant management system, and a chemical plant management method. [Background technology]

[0002] Conventionally, there is a technology for managing plants described in Japanese Patent Application Laid-Open No. 2020-187616 (Patent Document 1). This publication states that "The plant monitoring model creation device of the present application includes a weight coefficient calculation unit that calculates weight coefficients for a plurality of trained models, an average feature model calculation unit that calculates an average feature model using the weight coefficients calculated by the weight coefficient calculation unit, and a trained model creation unit that generates a trained model for the monitored plant using differential learning between the average feature model calculated by the average feature model calculation unit and plant data of the monitored plant, thereby making it possible to create a plant monitoring model in a short period of time." [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2020-187616 Summary of the Invention [Problem to be solved by the invention]

[0004] However, conventional technologies were not designed for chemical plants and could not accommodate multi-product production. Multi-product production in a chemical plant involves the production of a wide variety of items using different raw materials and operating conditions, resulting in lower production volumes for each item compared to conventional production. While identifying the relationship between raw materials, operating conditions, and production results for a specific product is important for production management, conventional technologies have made it difficult to identify such relationships before production begins or when production volumes are low.

[0005] Therefore, an object of the present invention is to identify the relationship between raw materials and operating conditions and production results in production at a chemical plant. [Means for solving the problem]

[0006] In order to achieve the above-mentioned object, one of the representative chemical plant management devices and chemical plant management systems of the present invention is characterized by having, in a chemical plant, a memory unit that stores information on a first raw material to be fed into a first reactor, information on first operating conditions, and information on a first product that is produced by feeding the first raw material into the first reactor and using the first operating conditions; an input unit that inputs a second product to be produced in a second reactor; and a control unit that identifies the second raw material and second operating conditions that will produce the input second product based on the first raw material, the first operating conditions, and the information on the first product. Furthermore, one representative chemical plant management method of the present invention is characterized by including the steps of: storing in a memory unit, in a chemical plant, information on a first raw material to be fed into a first reactor, information on first operating conditions, and information on a first product produced by feeding the first raw material into the first reactor and using the first operating conditions; receiving an input of a second product to be produced in a second reactor; and identifying, based on the first raw material, the first operating conditions, and the information on the first product, the second raw material and second operating conditions for producing the input second product. [Effects of the Invention]

[0007] According to the present invention, when producing a product in a chemical plant, it is possible to identify the relationship between raw materials, operating conditions, and production results. Furthermore, when producing a product under new conditions, it is possible to efficiently determine new production conditions based on information about raw materials and operating conditions. Objects, configurations, and advantages other than those described above will become clear from the following description of the embodiment. [Brief explanation of the drawings]

[0008] [Figure 1] An explanatory diagram of management of a chemical plant in an embodiment. [Figure 2] 1 is a configuration diagram of a chemical plant management system according to an embodiment; [Figure 3] A diagram showing the configuration of a chemical plant management device. [Figure 4] Example of a product table [Figure 5] Example of an ingredient table [Figure 6] Operating Condition Table [Figure 7] Illustration of spectroscopic analysis [Figure 8] Explaining the predicted reaction model [Figure 9] Diagram of creating a model from theory [Figure 10] An explanatory diagram of creating a model using machine learning [Figure 11] Diagram of the use of predictive response models (part 1) [Figure 12] Diagram of the use of predictive response models (part 2) [Figure 13] Flowchart showing the processing procedure of the chemical plant management device [Figure 14] Specific examples of display output by the display unit DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, a representative embodiment for carrying out the present invention will be described with reference to the drawings as appropriate. FIG. 1 is an explanatory diagram of management of a chemical plant according to an embodiment. In this embodiment, a chemical plant that produces a wide variety of products is the subject of management. Therefore, data on multiple types of products can be obtained from the chemical plant. The multiple types of products are, for example, products with different product numbers, reactors, raw materials, operating conditions, etc.

[0010] The product number is identification information for identifying the product to the extent that it can be considered identical for subsequent processing or supply as a product. If the product number is the same, the basic composition, etc. can be specified, but the details of the raw materials and operating conditions, and the quality are not necessarily the same. In other words, even when producing a substance with the same product number, the quality varies depending on differences in the raw materials, operating conditions, reactor, environmental factors, etc. Therefore, in order to efficiently produce a substance of the desired quality, it is necessary to fine-tune the raw materials, operating conditions, etc.

[0011] The chemical plant management system disclosed in this embodiment generates a predictive reaction model for multiple types of products from existing product data. The existing product data includes product number, reactor, raw materials, operating conditions, etc. By generating a predictive reaction model using this data and the generation results, it is possible to determine the raw materials and operating conditions suitable for generating the product.

[0012] Predictive reaction models are generally based on models derived from theory, and their accuracy gradually improves by feeding back production results. However, when producing a new product, sufficient accuracy cannot be expected as is because there are no existing production results. Furthermore, if the required production volume is small, production will end before accuracy improves sufficiently.

[0013] Therefore, the disclosed chemical plant management system uses existing predictive reaction models of other products when creating a predictive reaction model for a new product. Specifically, the disclosed chemical plant management system stores a plurality of existing predictive reaction models obtained from data on existing products, and groups the plurality of predictive reaction models according to their similarity. Here, we will explain the grouping of multiple predictive reaction models. Grouping may be performed based on the type of product, such as product number, chemical properties, physical properties, raw materials, reactor, or specific process steps included in the operating conditions, etc. Furthermore, for products that cannot be grouped based on these criteria, grouping may be performed based on important explanatory factors of the predictive reaction model. In this way, by combining feature-based clustering with the criteria described above, reliable and efficient grouping can be performed, contributing to the management of chemical plants that handle a wide variety of products. When the disclosed system receives input about data on a new product, it searches for existing products similar to the new product and identifies the group to which the new product belongs as a similar group (1).Then, it creates a predictive response model for the new product by ensemble learning of multiple predictive response models included in the similar group (2). The data for the new product can include the product number and reactor. Existing products that are similar to the new product may be determined by matching product numbers or reactors. For example, if groups Gr1 and Gr2 are classified by the reactors used, similar groups will be identified by the reactor specified in the data for the new product, and a predictive reaction model for the new product will be created from multiple predictive reaction models created for the same reactor. Furthermore, while the grouping has been described in terms of creating a set of groups for the multidimensional parameters of a product, it is also possible to create a group for each of the multiple characteristic types related to the product. That is, as an example, when focusing on two characteristic types A and B, grouping for characteristic type A (pattern 1) and grouping for characteristic type B (pattern 2) are performed separately. In this case, when data on a new product is input, similar groups are identified for each type, and a predictive response model for the new product is created by ensemble learning of multiple predictive response models included in the identified multiple similar groups.

[0014] By using a predictive reaction model of the new product thus produced and a group of similar raw materials and operating conditions, it is possible to identify the raw materials and operating conditions that should be applied to produce the new product. Furthermore, by utilizing the raw materials and operating conditions used in the production and a predictive reaction model, the results of the production can be predicted.

[0015] Fig. 2 is a configuration diagram of a chemical plant management system according to an embodiment. As shown in Fig. 2, the chemical plant management system includes a chemical plant management device 10 and a data integration platform 20. The data integration platform 20 collects data from various devices installed in the chemical plant and outputs the data to the chemical plant management device 10.

[0016] Specifically, raw material data 31, control performance data 32, quality data 33, etc. are obtained from the chemical plant. The raw material data 31 is data indicating the raw materials charged into the reactor. The control performance data 32 is data indicating the performance of reactor control acquired from the controller of the device related to reactor control. The quality data 33 is data indicating the quality of the product. Any index can be used to evaluate the quality of the product, but in this embodiment, spectroscopic analysis is used as an example. As another example, the quality data 33 may be a prediction result obtained by the chemical plant management device 10 predicting the production result during operation. The quality data 33 may be acquired by communicating with a system that integrates and manages quality control data of products in the chemical plant.

[0017] The chemical plant management device 10 is a device that manages a chemical plant using raw material data 31, control performance data 32, quality data 33, etc. The chemical plant management device 10 can create and store reaction prediction models for various products and output prediction results using the reaction prediction models to a manager. Based on the output of the chemical plant management device 10, the manager can determine raw materials, operating conditions, etc., and operate the chemical plant.

[0018] Furthermore, the chemical plant management apparatus 10 can receive data on a new product, identify similar groups, and create a predictive reaction model for the new product from the predictive reaction models of the similar groups.

[0019] In addition, the chemical plant management device 10 may be configured to determine raw materials and operating conditions using a reaction prediction model, generate information regarding control instructions for equipment related to reactor control, and transmit the information to the equipment controller to automatically control the chemical plant. The chemical plant management device 10 may be configured to automatically verify the validity of the current operating conditions by evaluating the current production results and operating conditions, and predicting and evaluating future production results, based on quality data 33 of the newly produced product (e.g., information on the emission spectrum inside the reactor that produces the product) before or during operation. Furthermore, the chemical plant management device 10 may be configured to automatically verify the validity of the current operating conditions by evaluating the current production results and operating conditions and predicting and evaluating future production results based on the control performance data 32 and the quality data 33 before or during operation. In this case, the chemical plant management device 10 may determine whether the current operating conditions need to be changed, identify new operating conditions after the change, generate information regarding control instructions for equipment related to reactor control so that the operating conditions are changed to the new operating conditions, and automatically send the information to the equipment controller. This configuration can contribute to the automatic operation of chemical plants.

[0020] 3 is a configuration diagram showing the configuration of the chemical plant management device 10. As shown in FIG. 3, the chemical plant management device 10 includes a calculation device 11, a main memory device 12, an input unit 13, a display unit 14, a communication interface 15, and an auxiliary memory device 16.

[0021] The arithmetic device 11 is, for example, a CPU (Central Processing Unit), and operates as a control unit as defined in the claims. The main storage device 12 is, for example, a RAM (Random Access Memory). The arithmetic unit 11 loads various programs and data into the main storage device 12 and sequentially executes the programs to control the operation of the chemical plant management device 10.

[0022] The input unit 13 is, for example, a keyboard, and receives operational input from the administrator. Specifically, the input unit 13 is used to input data of a new product. The display unit 14 is, for example, a liquid crystal display, and is used to display and output various types of information to the administrator.

[0023] The communication interface 15 acquires various data (raw material data 31, control performance data 32, quality data 33, etc.) from the chemical plant via the data integration platform 20. The communication interface 15 acquires the control performance data 32, thereby functioning as an acquisition unit as recited in the claims. Furthermore, the communication interface 15 acquires the spectroscopic analysis result (first emission spectrum) of the existing product as the quality data 33, thereby functioning as a first spectroscopic input unit as claimed. Furthermore, the communication interface 15 acquires the spectroscopic analysis result (second emission spectrum) of the new product as the quality data 33, thereby functioning as a second spectroscopic input unit as claimed.

[0024] The auxiliary storage device 16 is a storage unit that stores various data, such as a product table 41, a raw material table 42, an operating condition table 43, and a predicted reaction model 44.

[0025] The product table 41 is a table of information about existing products. If the reactor used to produce the existing product is a first reactor, the raw material input to the first reactor is a first raw material, and the operating conditions for the existing product are first operating conditions, the information about the existing product is "information about the first product that is produced by inputting the first raw material into the first reactor and using the first operating conditions." The raw material table 42 is a table of information on the first raw material to be fed into the first reactor. The operating condition table 43 is a table of information on operating conditions of existing products. The predicted reaction model 44 is information about a predicted reaction model related to a reaction that produces a first product in the first reactor. The predicted reaction model is a type of information related to the first product, and therefore may be stored in the product table 41, but for convenience, it is treated as separate data.

[0026] 4 is a specific example of a product table 41. The product table 41 shown in FIG. The product entry indicates the general name of the product, such as "Polymer A" or "Polymer B." The item number indicates identification information such as "A-0001" or "B-0002" for identifying products within a range that allows them to be considered identical for subsequent processing or supply as products. Quality control value 1 and quality control value 2 indicate the range of evaluation values ​​that indicate the quality of the product.

[0027] Fig. 5 is a specific example of a raw material table. The raw material table 42 shown in Fig. 5 has the items of product, product number, raw material 1 to raw material 5. The product and product number items are the same as in the product table 41. Raw materials 1 to 5 indicate the raw materials (first raw materials) used to produce the product. The raw materials may include solvents, initiators, monomers, catalysts, etc. The raw material table may also include any data related to raw materials other than the examples shown in FIG. 5. For example, the raw material table may include data related to the amount of each raw material.

[0028] Fig. 6 is a specific example of an operating condition table. The operating condition table 43 shown in Fig. 6 has items such as product, product number, reactor, temperature rise time, reaction temperature 1, reaction temperature 2, and hold time as operating conditions (second operating conditions). The product and product number items are the same as those in the product table 41. The reactor is information that identifies the reactor (first reactor) used to produce the product. The temperature rise time, reaction temperature 1, reaction temperature 2, hold time, etc. are conditions related to the temperature of the reactor. The operating condition table may also include any operating conditions other than the examples given in Fig. 6. For example, it may include the timing of feeding each raw material.

[0029] Figure 7 is an explanatory diagram of spectroscopic analysis. As shown in Figure 7, the liquid is sent from the inside of the reactor to the production section and spectroscopic analysis is performed. By acquiring the spectral data of the product using a spectrophotometer and constructing a regression equation, it is possible to create a regression equation that calculates the quality value from the spectral data. The spectral data acquired from an existing product is acquired as a first emission spectrum. The first emission spectrum or the quality value calculated from the first emission spectrum is stored in a product table 41 and thereby stored in the storage unit. The spectral data obtained from the new product is obtained as a second emission spectrum. The second emission spectrum or a quality value calculated from the second emission spectrum is used to output to a manager or to control the operation of the chemical plant. In FIG. 7, the product liquid sent from the reactor is used for spectroscopic analysis, but the configuration may be such that the light emitted from inside the reactor is separated and the emission spectrum is acquired.

[0030] Figure 8 is an explanatory diagram of a predictive reaction model. The predictive reaction model is a reactor model that simulates a reactor on a computer, and when explanatory variables are input, it outputs a response variable.

[0031] The objective variable of the predictive response model is a quality value or the like. The explanatory variables include raw materials, amounts of raw materials, operating conditions, reactor, theoretical formula, room temperature, and production volume. Among the explanatory variables, the raw materials and operating conditions are manipulable factors that can be manipulated during production. Among the explanatory variables, room temperature and the like are environmental variables. If a reactor to be used is specified, the reactor can also be included in the environmental variables. Among the explanatory variables, the amount of raw materials and production volume are load variables.

[0032] Predictive reaction models can be created from theory or by machine learning, or a combination of theory and machine learning for each parameter is possible. As an example of creating a predictive reaction model from theory, a basic predictive model formula can be prepared and analyzed based on chemical theory, taking into account the reaction system, and then the coefficients can be corrected by data analysis. Theoretical models created from theory have the advantage of not needing existing data and being able to predict unknown reactions from experimental data, but if everything is based on a theoretical model, there may be some deviations from the results of actual operation. These differences can be compensated for with machine learning models. Machine learning can automatically determine which model to use and how to combine them.

[0033] FIG. 9 is an explanatory diagram of creating a model from theory. In the first step (STEP 1), a physical model is created as follows: For example, the production rate of product [C] is evaluated from the concentration [Ai] of raw material i (i = 1 to n), catalyst concentration [Cat], and the Arrhenius equation as follows: d[C] / dt = -(1 / ai)·d[Ai] / dt = k·[Cat]**a· exp(-E / RT)·Πi [Ai]**ai Then, a physical model is created by incorporating it into the material balance equation for the raw materials.

[0034] In the next step (STEP 2), activation energy E, order of action α, and ai (i=1 to n) are determined by multiple regression analysis. In the next step (STEP 3), analytical values ​​are predicted from actual operating data based on the created model, and any discrepancies with the actual data are corrected through data analysis, improving accuracy. For example, a raw material evaporation term is created using the data from the equipment, and a raw material addition term is created from the raw material addition amount and raw material analysis values.

[0035] FIG. 10 is an explanatory diagram of the creation of a model using machine learning. In the first step (STEP 1), a theoretical model of the predicted reaction is created based on machine learning. In the next step (STEP 2), reaction-related terms are created as individual explanatory variables within the theoretical model. For example, terms for raw material evaporation and raw material reflux can be created. In the next step (STEP 3), the data created in STEP 2 is added to the prediction formula as an explanatory variable, thereby improving prediction accuracy.

[0036] 11 and 12 are explanatory diagrams of the use of a predicted response model. In FIG. 11, of the process including the first to third steps, data up to the second step are used as explanatory variables of the predictive reaction model to predict events after the second step. In addition, one or more interim analysis points are set during the second step, and a final analysis point is set after the third step, and the quality of the product is confirmed at the interim analysis points and the final analysis point. If the data up to the second step are given as explanatory variables to the predictive reaction model, the quality values ​​of the interim analysis points and the final analysis points can be obtained as the objective variable.The first to third steps are a combination of steps that sequentially carry out chemical changes such as promoting reactions by changing the temperature or maintaining a constant temperature, or changing or stabilizing the state.

[0037] In Figure 12, a model that predicts the future based on current information makes it possible to predict even processes that require significant control. For example, if the current time is between the first and second progress analysis points, by providing data including the quality at the first progress analysis point as an explanatory variable to the predictive response model, the quality values ​​at the second progress analysis point and one final analysis point can be obtained as the dependent variable.

[0038] 13 is a flowchart showing the processing procedure of the chemical plant management apparatus 10. First, the chemical plant management apparatus 10 accumulates data of existing products (step S101). The data of existing products includes information on the raw materials, reactors, and operating conditions for producing the products. The chemical plant management apparatus 10 creates a predicted reaction model from the data of existing products and stores it in the auxiliary storage device 16, which serves as a storage unit (step S102). Here, when creating a predictive reaction model from existing product data, quality data 33 of each product may be used. As an example, information on spectroscopic analysis of the product may be used as one of the feature quantities of the explanatory variables, or may be used as the dependent variable. In particular, in chemical plants that handle a wide variety of products, there is often a shortage of information that can be used as the dependent variable, and using this information as the dependent variable allows the number of samples of the dependent variable to be increased. The chemical plant management apparatus 10 calculates the similarity between the created predictive reaction models, and groups the predictive reaction models so that similar models belong to the same group (step S103). Grouping may be based on the type of product such as product number, chemical properties, physical properties of the product, processing of a specific step included in the reactor or operating conditions, etc. Furthermore, if grouping cannot be performed based on these criteria, grouping may be performed based on important explanatory factors of the predictive reaction model.

[0039] When the chemical plant management device 10 receives input of new product data (step S104), it identifies a group similar to the new product data (step S105). Here, the new product data includes information on the reactor that produces the product. Then, it creates a predictive reaction model for the new product from the predictive reaction models of the similar group and stores it in the auxiliary storage device 16, which serves as a storage unit (step S106).

[0040] The chemical plant management device 10 identifies the raw materials and operating conditions of the new product from the raw materials, operating conditions, etc. of the similar group and the predicted reaction model of the new product (step S107), and outputs the identified raw materials and operating conditions (step S108), thereby ending the process. One example of the output is to provide information to a manager. The output may also be used to control the operation of a chemical plant, thereby enabling automatic operation of the chemical plant. Furthermore, it is also possible to modify the predicted reaction model based on the output and the actual production results. Here, the grouping in step S103 has been described as creating a set of groups based on the multidimensional parameters of the product. However, groups may be created for each of multiple property types related to the product. In this case, when new product data is received, similar groups are identified for each type in step S105, and the raw materials and operating conditions of the new product are identified from the multiple predicted reaction models included in the identified similar groups in step S107. This enables identification that takes into account the influence of each of the multiple property types. Here, the chemical plant management system 10 may determine, from the similar groups identified for each property type, similar groups based on the property types that are more important in identifying the raw materials and operating conditions of the new product, and use this to identify the raw materials and operating conditions of the new product. This enables identification that takes into account the influence of each property type, which is expected to improve accuracy. Here, when a new product is produced in a certain reactor, the product in the existing predictive reaction model included in the similar group used to create the predictive reaction model of the new product may be configured to be the same as the new product. That is, when specifying the materials and operating conditions for producing a new product in a certain reactor, the materials and operating conditions for a product that is the same as a new product produced in a different reactor in the past may be used. Furthermore, the reactor producing the new product and the reactor in the existing predictive reaction model included in the similar group used to create the predictive reaction model for the new product may be configured to be the same reactor. That is, when specifying the materials and operating conditions for producing a new product in a certain reactor, the materials and operating conditions for a product different from the new product produced in the same reactor in the past may be used.

[0041] The output of raw materials and operating conditions using a predictive reaction model may be performed before or during production. For example, it may be used to detect anomalies in the production of a new product. For example, judgment information, which is information about predetermined conditions that indicate the occurrence of an anomaly in the production of a product, may be stored in the auxiliary storage device 16 or the like. Predictions may be made in real time using a predictive reaction model during the production, and the prediction results may be compared with the judgment information to determine whether an anomaly will occur. Here, the predetermined conditions that indicate the occurrence of an anomaly may be various conditions determined for each plant facility or product. Examples include conditions such as the product characteristics not satisfying desired conditions or the pressure exceeding a predetermined threshold. If it is determined that an anomaly will occur, how the operating conditions should be changed may be determined based on the information about the anomaly, the operating conditions up to that point, and past operating performance data, and the determined operating conditions may be output as third operating conditions. Furthermore, instead of outputting the identified operating conditions as the third operating conditions, a configuration may be adopted in which predictions are made in real time using a predictive reaction model during generation, and instructions are given to a controller that controls the plant equipment to automatically change the operating conditions, etc.

[0042] Fig. 14 is a specific example of the display output by the display unit. The display output in Fig. 14 displays the quality prediction curve, product information, operating parameters, and quality prediction values. The quality prediction curve is a graphical representation of the results of predictions made using a predictive reaction model, and can show quality predictions under various operating conditions. The graph also shows quality control values ​​as the target range of quality values, allowing users to check whether the quality values ​​will reach the quality control values.

[0043] Product information includes the date of manufacture, product, part number, lot, reactor, and quality control range. The operating parameters include the current temperature and pressure, and the recommended operating temperature and pressure. The quality prediction value indicates the current reaction completion time and quality prediction value, and the reaction completion time and quality prediction value of the recommended operation. The recommended operation in terms of operating parameters and quality prediction values ​​corresponds to the guidance instructions of the quality prediction curve.

[0044] In addition, any information related to the generation can be used for the display output. For example, for each explanatory variable in a predictive reaction model of a new product, the correlation with the objective variable of the predictive reaction model may be evaluated using machine learning, and information on the explanatory variables and information on an index showing the correlation of the explanatory variables with the objective variable may be displayed in order of the highest correlation with the objective variable. Such a display makes it possible to clearly show what factors are affecting the quality of the product. Furthermore, the chemical plant management apparatus 10 may store a reactor model that simulates a reactor on a computer in the auxiliary storage device 16. In this case, the chemical plant management apparatus 10 can input information specified as raw materials and operating conditions for producing a new product into the reactor model and perform calculations to identify the substance that the reactor model will produce. In this way, by simulating the second reactor, it is possible to predict the substance that will be obtained as the second product. Furthermore, by inputting information on any raw materials and operating conditions into a reactor model and performing calculations to identify the substance produced by the reactor model, a new predictive reaction model can be obtained from the input information on the raw materials and operating conditions and information on the substance produced by the virtually obtained reactor model. This makes it possible to increase the number of samples of existing predictive reaction models to be used when producing new products.

[0045] As described above, the disclosed chemical plant management apparatus 10 includes a storage unit in a chemical plant that stores information on a first raw material to be fed into a first reactor, information on first operating conditions, and information on a first product produced by feeding the first raw material into the first reactor under the first operating conditions; an input unit that inputs a second product to be produced in a second reactor; and a control unit that generates a plurality of property groups based on the information on the first product, each of which includes a plurality of first products having a predetermined range of properties. The control unit identifies a predetermined property group among the plurality of property groups that includes the second product based on the properties of the second product and the property ranges of the plurality of property groups, and identifies a second raw material and second operating conditions for producing the second product in the second reactor based on the input information on the plurality of first products included in the identified predetermined property group.

[0046] In this configuration, by utilizing the track record of a known product (first product), the first product can be classified into one of multiple property groups, and the raw materials and operating conditions of the target product (second product) can be identified.By identifying the property group to which the target product (second product) should be classified based on the properties of the target product (second product), the raw materials and operating conditions of the second product can be identified, and the raw materials and operating conditions of the target product can be identified from information on products with similar properties.

[0047] The control unit may also be configured to generate, for each type of characteristic, a plurality of characteristic groups each including a plurality of first products having a predetermined range of characteristics based on information about the first product, assign information about the characteristic group in which the second product is included to information about the second product, and identify the second raw material and the second operating conditions for producing the second product based on the first raw material and the first operating conditions included in the assigned characteristic group. In this configuration, the first product is grouped by its characteristics, the groups into which the second product should be classified by its characteristics are identified, the first product is extracted based on the group of each characteristic of the second product, and the extracted first product is used as a similar product to identify the raw materials and operating conditions of the second product.

[0048] Furthermore, the range of the characteristics of the first product is determined based on information about the product number of the product representing the first product. Product numbers are assigned according to the composition of the product, so similar products can be easily identified by using the product number as a characteristic.

[0049] Further, the storage unit stores a second reactor model that simulates the second reactor on a computer, and the control unit inputs information on the second raw material and information on the second operating conditions into the second reactor model and performs calculations to identify a substance produced by the second reactor model. In this way, by simulating the second reactor, it is possible to predict the substance obtained as the second product.

[0050] The storage unit may be configured to store a first reactor model that simulates the first reactor on a computer, and the control unit may be configured to input information on the first raw material and information on the first operating conditions into the first reactor model and perform a calculation to identify a substance to be produced by the first reactor model, and to identify a second raw material and second operating conditions that produce the second product based on the information on the first raw material, the information on the first operating conditions, and information on the substance to be produced by the first reactor model. In this configuration, the first reactor is simulated, and the raw materials and operating conditions for the second product can be identified based on information on the virtually obtained first product.

[0051] The first product and the second product may be of the same product type. The first reactor and the second reactor may be the same reactor. In this way, by using information about the first product, which has the same product and reactor, it is possible to obtain information about the second product with high accuracy.

[0052] The first reactor may further include a first spectral input unit that disperses light emitted from within the first reactor to acquire a first emission spectrum, and the control unit may store information about the acquired first emission spectrum in a storage unit, and when identifying the second raw material and the second operating conditions, identify them based on the first emission spectrum in addition to the first operating conditions and the first operating conditions. The second reactor may further include a second spectral input unit that disperses light emitted from inside the second reactor and acquires a second emission spectrum, and the control unit may store information on the acquired second emission spectrum in a storage unit, and display information on the second operating conditions or a change in the second operating conditions on a display unit based on the second emission spectrum before or during operation of the second reactor. In this way, by evaluating the quality of the first product and the second product using spectroscopic analysis, it is possible to predict the quality of the second product and identify raw materials and operating conditions that will satisfy quality requirements.

[0053] The reactor may further include a display unit that displays information to an administrator, and the storage unit further stores judgment information, which is information on an index related to the production of a product and is information on a predetermined condition that indicates the occurrence of an abnormality in the production of the product. The control unit determines whether an abnormality will occur when operating under the second operating conditions based on the judgment information and a result of calculation performed by inputting information on the second raw material and information on the second operating conditions into the second reactor model, and if it is determined that an abnormality will occur when operating under the second operating conditions, the control unit specifies a third operating condition based on the information on the abnormality and the second operating conditions, and displays the third operating condition on the display unit. According to this configuration, it is possible to predict the production results of the second product in real time and display the operating conditions that will avoid abnormalities.

[0054] The control unit may further include an acquisition unit that acquires information about the second product produced in the second reactor and control history data of equipment related to control of the second reactor from a controller of the equipment, wherein the control unit generates information about control instructions for equipment related to control of the second reactor based on the specified second raw material and the second operating conditions, transmits the information to the controller, stores the acquired information about the second product produced in the second reactor and the control history data in a storage unit, and changes the second operating conditions based on the information about the second product produced in the second reactor and the control history data during or before operation of the second reactor. According to this configuration, the production result of the second product can be predicted in real time, and the chemical plant can be operated automatically.

[0055] Furthermore, the information on the first product may include information on a predictive reaction model related to a reaction that produces the first product in the first reactor, and when performing the process of dividing the information on the first product into one of a plurality of groups based on a range of characteristics of the first product, the control unit may determine a similarity of the predictive reaction models based on at least the range of characteristics of the first product, classify each piece of information on the first product into one of the plurality of groups based on the similarity, identify a predetermined group in which the characteristics of the second product fall within the range of characteristics of the group, and when specifying the second raw material and the second operating conditions based on the first raw material and the first operating conditions included in the specified predetermined group, create a predictive reaction model related to a reaction that produces the second product in the second reactor based on the predictive reaction model related to the first product included in the specified group, and specify the second raw material and the second operating conditions based on the predictive reaction model. In this configuration, a predictive reaction model of the second product can be generated using a predictive reaction model of the first product that is similar to the second product, and the raw materials and operating conditions for the second product can be determined.

[0056] Furthermore, the control unit may be configured to evaluate the correlation between each explanatory variable in a predictive reaction model for a reaction to produce the second product in the second reactor and the objective variable of the predictive reaction model, and to display information on the explanatory variables and information on an index indicating the correlation of the explanatory variables with the objective variable on the display unit in order of highest correlation with the objective variable. In this configuration, it is possible to clearly show what factors are affecting the quality of the second product.

[0057] The present invention is not limited to the above-described embodiment, but includes various modifications. For example, the above-described embodiment has been described in detail to clearly explain the present invention, and is not necessarily limited to those including all of the described configurations. Furthermore, not only can the configurations be deleted, but also replacements and additions of configurations are possible.

[0058] For example, in the above embodiment, for the sake of simplicity, a new product that has not been produced up to that point has been described as an example of the "second product." However, the product produced in the first reactor may also be used as the "second product" when produced in the second reactor. Furthermore, the example of small-lot production of a wide variety of products is given in order to demonstrate that prediction is possible regardless of whether or not there is a track record for the target product, and does not preclude the application of the present invention to products that are planned to be produced in large quantities. Furthermore, the timing of grouping the predicted response models may be such that the grouping is performed in advance and stored in the storage unit, or may be performed as needed when the data of the second product is received. [Explanation of symbols]

[0059] 10: Chemical plant management device, 11: Arithmetic unit, 12: Main memory device, 13: Input unit, 14: Display unit, 15: Communication interface, 16: Auxiliary memory device, 20: Data integration platform, 31: Raw material data, 32: Control performance data, 33: Quality data, 41: Product table, 42: Raw material table, 43: Operation condition table, 44: Prediction reaction model

Claims

1. a storage unit in a chemical plant that stores information about a first raw material to be fed into a first reactor, information about first operating conditions, and information about a first product produced by feeding the first raw material into the first reactor and using the first operating conditions; an input section for inputting a second product produced in the second reactor; a control unit that generates a plurality of characteristic groups based on information about the first products, each of which includes a plurality of the first products having characteristics within a predetermined range; The control unit identifying a predetermined property group including the second product among the plurality of property groups based on the property of the second product and the property ranges of the plurality of property groups; Identifying a second raw material and second operating conditions for producing the second product in the input second reactor based on information about the plurality of first products included in the identified predetermined property group. A chemical plant management device comprising:

2. 2. The plant management system according to claim 1, The control unit generating a plurality of property groups for each type of property, each group including a plurality of first products having a property within a predetermined range, based on information about the first product; adding information about the property group in which the second product is included to information about the second product; Identifying the second raw material and the second operating conditions for producing the second product based on information of the plurality of first products included in the assigned property group. A chemical plant management device comprising:

3. 2. The chemical plant management system according to claim 1, The range of properties of the first product is determined based on product part number information representing the first product. A chemical plant management device comprising:

4. 2. The chemical plant management system according to claim 1, the storage unit stores a second reactor model that simulates the second reactor on a computer, The control unit inputs information about the second raw material and information about the second operating conditions into the second reactor model and performs a calculation to identify a substance produced by the second reactor model. A chemical plant management device comprising:

5. 5. The chemical plant management system according to claim 4, the storage unit stores a first reactor model that simulates the first reactor on a computer, The control unit inputting information on the first raw material and information on the first operating conditions into the first reactor model and performing a calculation, thereby identifying a substance to be produced by the first reactor model; Identifying a second raw material and second operating conditions for producing the second product based on information on the first raw material, information on the first operating conditions, and information on substances produced by the first reactor model. A chemical plant management device comprising:

6. 2. The chemical plant management system according to claim 1, The first product and the second product are of the same product type. A chemical plant management device comprising:

7. 2. The chemical plant management system according to claim 1, The first reactor and the second reactor are the same reactor. A chemical plant management device comprising:

8. 2. The chemical plant management system according to claim 1, a first spectral input unit that disperses the light emitted from inside the first reactor and acquires a first emission spectrum; The control unit storing the acquired information of the first emission spectrum in a storage unit; When identifying the second raw material and the second operating conditions, the identification is based on the first emission spectrum in addition to information about the plurality of first products. A chemical plant management device comprising:

9. 2. The chemical plant management system according to claim 1, a second spectral input unit that disperses the light emitted from the second reactor and acquires a second emission spectrum; The control unit storing the acquired information of the second emission spectrum in a storage unit; displaying information on the second operating conditions or changes to the second operating conditions on a display unit based on the second emission spectrum before or during operation of the second reactor. A chemical plant management device comprising:

10. 5. The chemical plant management system according to claim 4, It also has a display unit that displays information to the administrator. the storage unit further stores judgment information that is information about an index related to the generation of a product, the judgment information being information about a predetermined condition that indicates the occurrence of an abnormality in the generation of the product; the control unit determines whether an abnormality will occur when the reactor is operated under the second operating conditions, based on a calculation result obtained by inputting information about the second raw material and information about the second operating conditions into the second reactor model and the determination information; If it is determined that an abnormality will occur when operating under the second operating conditions, a third operating condition is identified based on information about the abnormality and the second operating conditions, and the third operating condition is displayed on the display unit. A chemical plant control device comprising:

11. 2. The chemical plant management system according to claim 1, an acquisition unit that acquires information about the second product generated in the second reactor and control performance data of a device related to control of the second reactor from a controller of the device, The control unit generating information regarding control instructions for devices related to control of the second reactor based on the identified second raw material and the second operating conditions, and transmitting the information to the controller; storing the acquired information about the second product produced in the second reactor and the control performance data in a storage unit; changing the second operating conditions based on information on the second product produced in the second reactor and the control performance data during or before operation of the second reactor; A chemical plant management device comprising:

12. 2. The chemical plant management system according to claim 1, the information about the first product includes information about a predictive reaction model about a reaction that produces the first product in a first reactor; The control unit When performing a process of classifying information about the first product into one of a plurality of groups based on a range of characteristics of the first product, a similarity of the predicted reaction model is determined based on at least a range of characteristics of the first product, and the information about the first product is classified into one of the plurality of groups based on the similarity; identifying a predetermined group in which the properties of the second product fall within a range of the properties of the group, and identifying the second raw material and the second operating conditions based on the first raw material and the first operating conditions included in the identified predetermined group, creating a predictive reaction model for a reaction to produce the second product in the second reactor based on a predictive reaction model for the first product included in the identified predetermined group, and identifying the second raw material and the second operating conditions based on the predictive reaction model; A chemical plant management device comprising:

13. The chemical plant management system according to claim 12, The control unit evaluating a correlation between each explanatory variable in a predictive reaction model related to a reaction for producing the second product in the second reactor and a response variable of the predictive reaction model; Displaying information on the explanatory variables and information on indices indicating the correlation between the explanatory variables and the objective variable on a display unit in descending order of correlation with the objective variable. A chemical plant management device comprising:

14. a storage unit in a chemical plant that stores information about a first raw material to be fed into a first reactor, information about first operating conditions, and information about a first product produced by feeding the first raw material into the first reactor and using the first operating conditions; an input section for inputting a second product produced in the second reactor; a control unit that generates a plurality of characteristic groups based on information about the first products, each of which includes a plurality of the first products having characteristics within a predetermined range; The control unit identifying a predetermined property group including the second product among the plurality of property groups based on the property of the second product and the property ranges of the plurality of property groups; Identifying a second raw material and second operating conditions for producing the second product in the input second reactor based on information about the plurality of first products included in the identified predetermined property group. A chemical plant management device comprising:

15. In a chemical plant, a step of storing information on a first raw material to be fed into a first reactor, information on first operating conditions, and information on a first product produced by feeding the first raw material into the first reactor and using the first operating conditions in a storage unit; accepting an input of a second product to be produced in a second reactor; generating a plurality of property groups based on information about the first products, each of the property groups including a plurality of the first products having a predetermined range of properties; identifying a predetermined property group among the plurality of property groups, the predetermined property group including the second product, based on the property of the second product and the property ranges of the plurality of property groups; and identifying a second raw material and second operating conditions for producing the second product in the input second reactor based on information about the plurality of first products included in the identified predetermined property group. A chemical plant management method comprising:

Images