Information processing device, property estimation method, model production method, and program

Abstract

To rapidly and highly accurately estimate properties of produced oil generated in a plant.SOLUTION: An information processing device 10 includes: a first estimation unit 22 into which actual measured values of a plurality of operation parameters representing an operation condition of a plant are input and which outputs an estimated value of a first characteristic representing properties of produced oil generated in the plant; and a second estimation unit 24 into which the estimated value of the first characteristic is input and which outputs an estimated value of a second characteristic representing properties of the produced oil.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present disclosure relates to an information processing device, a property estimation method, a model generation method, and a program. [Background technology]

[0002] Soft sensors are sometimes used to analyze in real time the properties of products produced in plants such as oil refineries. For example, a technique is known in which process variables from the plant are given to a model, the plant state is simulated by a simulator, and the product properties are estimated using virtual outputs obtained from the simulator (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2010-79465 Summary of the Invention [Problem to be solved by the invention]

[0004] In the above-mentioned prior art, the state of the plant is simulated by a simulator, so unless the computer used for the simulation is high performance, it takes a long time to estimate the properties of the product.

[0005] An exemplary object of an embodiment of the present disclosure is to provide a technique for quickly and accurately estimating the properties of product oil produced in a plant. [Means for solving the problem]

[0006] An information processing device according to one embodiment of the present disclosure includes a first estimation unit that receives as input actual measured values of a plurality of operating parameters that indicate the operating conditions of the plant and outputs an estimated value of a first characteristic that indicates the properties of the product oil produced in the plant, and a second estimation unit that receives as input the estimated value of the first characteristic and outputs an estimated value of a second characteristic that indicates the properties of the product oil.

[0007] Another aspect of the present disclosure is a property estimation method, which includes the steps of: receiving as input actual measured values of a plurality of operating parameters indicating the operating conditions of a plant, and outputting an estimated value of a first property indicating the properties of a product oil produced by the plant; and receiving as input the estimated value of the first property, and outputting an estimated value of a second property indicating the properties of the product oil.

[0008] Yet another aspect of the present disclosure is a program that causes a computer to implement a function of receiving actual measured values of a plurality of operating parameters indicating the operating conditions of a plant and outputting an estimated value of a first characteristic indicating the properties of a product oil produced by the plant, and a function of receiving the estimated value of the first characteristic and outputting an estimated value of a second characteristic indicating the properties of the product oil.

[0009] Yet another aspect of the present invention is a model generation method, which includes the steps of: generating a first model using simulation results from a plant simulator that simulates a plant, inputting actual measured values of multiple operating parameters that indicate the operating conditions of the plant, and outputting an estimated value of a first characteristic that indicates the properties of a product oil produced in the plant; acquiring an estimated value of the first characteristic that is output from the first model when a set of actual measured values of the multiple operating parameters that indicate the actual operating conditions of the plant is input to the first model; acquiring actual measured values of a second characteristic that indicates the properties of the product oil and that correspond to the set; and generating a second model that indicates the correlation between the estimated value of the first characteristic and the actual measured value of the second characteristic. [Effects of the Invention]

[0010] According to the present disclosure, the properties of the product oil produced in a plant can be estimated quickly and with high accuracy. [Brief explanation of the drawings]

[0011] [Figure 1] 1 is a diagram illustrating an example of a schematic configuration of a plant according to an embodiment. [Figure 2]1 is a diagram schematically illustrating a configuration of an information processing device according to an embodiment. [Figure 3] 3 is a flowchart illustrating an example of a property estimation method according to an embodiment. [Figure 4] 1 is a flowchart illustrating an example of a model generation method according to an embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0012] The present disclosure provides an overview. The present disclosure relates to a technology for estimating values of characteristics that indicate the properties of a product oil produced in a plant from actual measured values of multiple operating parameters that indicate the operating conditions of the plant. Some of the characteristics that indicate the properties of the product oil are difficult to measure in real time, and highly accurate measurement requires the use of test equipment installed in a laboratory or the like. The present disclosure provides a so-called soft sensor for estimating such characteristics.

[0013] Because it is difficult to measure a characteristic that requires a long time to measure frequently, the amount of actual measurement data accumulated as past operating results is small, making it difficult to generate a highly accurate model or prediction formula for estimating the value of the characteristic. In the present disclosure, instead of directly estimating the value of a desired characteristic from the actual measurement values of multiple operating parameters, the value of a first characteristic, which serves as an intermediate variable, is estimated from the actual measurement values of multiple operating parameters, and the value of a second characteristic, which is a desired characteristic, is estimated from the value of the first characteristic. For example, as the first characteristic, which serves as an intermediate variable, a characteristic similar to the second characteristic and from which a larger amount of data can be obtained than the second characteristic, can be selected. According to the present disclosure, by estimating the value of the second characteristic via the value of the first characteristic, the value of the second characteristic, which is a desired characteristic, can be accurately estimated.

[0014] The subject of the information processing device or method of the present disclosure includes a computer. The computer executes a computer program to realize the functions of the subject of the information processing device or method of the present disclosure. The computer includes, as its main hardware configuration, a processor that operates according to the computer program. The type of processor is not important as long as it can realize the functions by executing the computer program. The processor is composed of one or more electronic circuits including semiconductor integrated circuits (IC, LSI, etc.). The computer program is recorded on a non-transitory recording medium such as a computer-readable ROM, optical disk, or hard disk drive. The computer program may be pre-stored on the recording medium or may be supplied to the recording medium via a wide area communication network including the Internet.

[0015] The technology of the present disclosure will be described below with reference to the drawings based on preferred embodiments. The embodiments are illustrative and do not limit the invention, and all features and combinations thereof described in the embodiments are not necessarily essential to the invention. Identical or equivalent components, parts, and processes shown in each drawing are designated by the same reference numerals, and redundant descriptions will be omitted where appropriate. Furthermore, the scale and shape of each part shown in each drawing are set for convenience to facilitate explanation and should not be interpreted as limiting unless otherwise specified. Furthermore, when terms such as "first" and "second" are used in this specification or claims, unless otherwise specified, they do not represent any order or importance, but are used to distinguish one configuration from another.

[0016] FIG. 1 is a diagram showing an example of a schematic configuration of a plant 50 according to an embodiment. The plant 50 is an apparatus that produces a product oil from a feedstock oil 52. The plant 50 is a distillation refinery that fractionates the feedstock oil 52 to produce a first product oil 54, a second product oil 56, and a third product oil 58. The feedstock oil 52 is, for example, a naphtha fraction. The first product oil 54 is, for example, a heavy naphtha fraction. The second product oil 56 is, for example, a light naphtha fraction. The third product oil 58 is, for example, a liquefied petroleum gas.

[0017] The oil types of the feed oil 52, the first product oil 54, the second product oil 56, and the third product oil 58 are not limited to the oil types exemplified above, and can be changed as appropriate depending on the configuration of the plant 50 used by the user, the type of feed oil 52 used by the user, or the type of product oil desired by the user. Furthermore, depending on the configuration of the plant 50 and the production plan of the plant 50, some of the first product oil 54, the second product oil 56, and the third product oil 58 may not be produced, or a fourth product oil (not shown) may also be produced.

[0018] Plant 50 includes a first fractionator 60 , a first condenser 62 , a second fractionator 64 , and a second condenser 66 .

[0019] The first fractionator 60 separates the heavy fraction contained in the feedstock 52 to produce a first product oil 54. The first fractionator 60 is also called a splitter. The feedstock, which is the residue obtained by separating the first product oil 54 from the feedstock 52, is sent from the first fractionator 60 to a first condenser 62 as a first overhead component 68. The first condenser 62 cools the first overhead component 68 and separates it into a first off-gas 70, which is a gas component, and a light fraction 72, which is a liquid component. The light fraction 72 is sent from the first condenser 62 to a second fractionator 64. A portion 72a of the light fraction 72 output from the first condenser 62 is refluxed to the first fractionator 60 and used to adjust the temperature of the first fractionator 60.

[0020] The second fractionator 64 separates the light naphtha fraction contained in the light fraction 72 sent from the first condenser 62 to produce a second product oil 56. The second fractionator 64 is also called a stabilizer. The light fraction, which is the residue obtained by separating the second product oil 56 from the light fraction 72, is sent from the second fractionator 64 to the second condenser 66 as a second overhead component 74. The second condenser 66 cools the second overhead component 74 and separates it into a second off-gas 76, which is a gas component, and a third product oil 58, which is a liquid component. A portion 58a of the third product oil 58 output from the second condenser 66 is refluxed to the second fractionator 64 and used to adjust the temperature of the second fractionator 64.

[0021] The operating parameters indicating the operating conditions of the plant 50 include, for example, at least one of a temperature parameter, a pressure parameter, a flow rate parameter, and a reflux ratio parameter. Temperature parameters include, for example, the feed temperature of the feedstock 52 supplied to the first fractionator 60, the temperature of each tray of the first fractionator 60 and the second fractionator 64, and the temperatures of the first condenser 62 and the second condenser 66. Pressure parameters may include the internal pressures of the first fractionator 60 and the second fractionator 64. Flow rate parameters may include the flow rate of the feedstock 52, the flow rate of the first product oil 54, the flow rate of the second product oil 56, the flow rate of the third product oil 58, the flow rate of the first off-gas 70, and the flow rate of the second off-gas 76. Flow rate parameters may include a flow rate ratio, or, for example, a value normalized based on the flow rate of the feedstock 52. Parameters related to reflux ratios may include the reflux ratio from first condenser 62 to first fractionator 60, the reflux ratio from second condenser 66 to second fractionator 64, and the like.

[0022] The second product oil 56 discharged from the second fractionator 64 may have a boiling point below a predetermined standard value (e.g., 40°C) and may be classified as a special flammable material as defined by the Fire Service Act. If the second product oil 56 is classified as a special flammable material, storage and handling become difficult. Therefore, an additive 80 with a high boiling point is added to the second product oil 56 to adjust the boiling point of the resulting product oil 82 after the addition to the second product oil 56, so that the resulting product oil 82 has a boiling point above a predetermined standard value (e.g., 40°C). The additive 80 is, for example, a raffinate, and has a boiling point of, for example, about 60°C. The additive 80 has high economic value because it can be used as a gasoline base stock. Therefore, from an economic standpoint, it is preferable that the amount of additive 80 added to the second product oil 56 be as small as possible, and it is preferable that the amount added be adjusted so that the boiling point of the resulting product oil 82 after the addition is equal to or higher than a control value slightly higher than the predetermined standard value (e.g., 41°C or 42°C). Note that the additive 80 is not limited to a raffinate, as long as it has a higher boiling point than the second product oil 56.

[0023] However, in order to accurately and precisely measure the boiling point of the second product oil 56, a test based on the equilibrium reflux boiling point test method (e.g., JIS K2233:2017) is required, and it takes time to obtain the actual measured value. Therefore, in the past, it was difficult to accurately measure the boiling point of the second product oil 56 in real time, so it was necessary to set a control value that took into account a margin depending on the frequency and accuracy of boiling point measurements. As a result, compared to when the control value was set close to the predetermined reference value, the amount of additive 80 added had to be increased by the amount of the margin set, which left room for improvement from an economic perspective. Therefore, in this embodiment, a so-called software sensor is used to enable the boiling point of the second product oil 56 to be estimated accurately and quickly. In this embodiment, estimation of the boiling point of the second product oil 56 is described as an example, but the present invention can also be applied to estimation of the boiling points of other product oils.

[0024] 2 is a diagram schematically illustrating the functional configuration of an information processing device 10 according to an embodiment. The information processing device 10, a data management device 42, a plant management device 44, and a user terminal 46 are connected to each other via a network 40 in a state in which they can communicate with each other. The plant management device 44 is connected to a plant device 48 in a state in which they can communicate with each other.

[0025] The information processing device 10, the data management device 42, the plant management device 44, and the user terminal 46 are each a general-purpose computer such as a server, a workstation, or a personal computer, or a mobile terminal such as a smartphone or a tablet computer. As an example, the information processing device 10, the data management device 42, and the plant management device 44 are servers. As an example, the user terminal 46 is a personal computer.

[0026] Each block shown in the block diagrams of this disclosure can be realized in terms of hardware by elements or mechanical devices, such as a processor such as a computer's CPU (Central Processing Unit) and memories such as ROM (Read Only Memory) and RAM (Random Access Memory), and in terms of software by a computer program or the like. Here, functional blocks realized by cooperation between hardware and software are depicted. Those skilled in the art will understand that these functional blocks can be realized in various ways by combining hardware and software.

[0027] A computer program that implements the functions of at least some of the functional blocks shown in Fig. 2 may be installed in the storage of one or more computers. The CPUs of the one or more computers may load the installed computer program into their main memory and execute it to perform the functions of the functional blocks shown in Fig. 2.

[0028] 2 may be executed by a single computer or may be distributed among multiple computers. When the functions of the functional blocks shown in Fig. 2 are distributed among multiple computers, the multiple computers may transmit and receive data via a communication network including a LAN (Local Area Network), a WAN (Wide Area Network), and the Internet.

[0029] Network 40 is configured by a communication network including at least one of a local area network (LAN), a wide area network (WAN), the Internet, and various mobile communication systems constructed by wireless base stations. Examples of the mobile communication system include mobile communication systems such as 3G, 4G, and 5G, LTE (Long Term Evolution), and wireless networks (e.g., Wi-Fi (registered trademark)) that can be connected to the Internet via a predetermined access point.

[0030] The data management device 42 acquires and manages data used by the information processing device 10. The data management device 42 is, for example, a data server that collects and stores various data related to plant equipment 48. The data management device 42 acquires necessary data from, for example, the plant management device 44 or a user terminal 46. The data management device 42 transmits data required by the information processing device 10 to the information processing device 10 in response to a request from the information processing device 10, for example.

[0031] The plant management device 44 acquires and manages data from plant equipment 48. The plant equipment 48 is equipment installed in the plant 50 or a group of equipment consisting of multiple equipment installed in the plant 50. The plant equipment 48 includes, for example, at least one of a first fractionator 60, a first condenser 62, a second fractionator 64, and a second condenser 66. The plant equipment 48 is equipped with sensors for acquiring actual measured values of operating parameters that indicate the operating conditions of the plant equipment 48. The types of sensors equipped in the plant equipment 48 are not particularly limited, and may include, for example, temperature sensors, pressure sensors, and flow rate sensors. The plant management device 44 acquires the measured values measured by the sensors equipped in the plant equipment 48.

[0032] The user terminal 46 is used by a user or operator who manages the plant. The user terminal 46 displays an input screen for inputting data to be used in the information processing device 10. The data input to the user terminal 46 is, for example, transmitted to the data management device 42 and managed by the data management device 42. For example, an actual measured value of a property indicating the properties of the product oil produced in the plant 50 is input to the user terminal 46. For example, an actual measured value of the equilibrium reflux boiling point measured on a sample of the second product oil 56 produced in the plant 50 is input to the user terminal 46. The user terminal 46 displays information output from the information processing device 10.

[0033] The information processing device 10 includes an acquisition unit 12, an estimation unit 14, a calculation unit 16, a notification unit 18, and a storage unit 20. The estimation unit 14 may include a first estimation unit 22 and a second estimation unit 24. The storage unit 20 may include driving data 26, actual measurement value data 28, and model data 30. If the storage unit 20 does not store these data, the information processing device 10 may generate these data or acquire them from a data management device 42 as necessary. The information processing device 10 may further include a model generation unit 32.

[0034] The acquisition unit 12 acquires actual measured values of a plurality of operating parameters that indicate the operating conditions of the plant 50. The actual measured values of the plurality of operating parameters are measured by sensors provided in the plant equipment 48. The types of the plurality of operating parameters are not particularly limited, but may be, for example, temperature, pressure, flow rate, reflux ratio, etc.

[0035] The actual measured values of the operating parameters acquired by the acquisition unit 12 may be the sensor measurements themselves or may be calculated values calculated from measurements by one or more sensors. The acquisition unit 12 may acquire measurements by one or more sensors and calculate the actual measured value of at least one operating parameter using the acquired measurements. For example, the acquisition unit 12 may acquire a first measured value by a first sensor of the plant equipment 48 and a second measured value by a second sensor of the plant equipment 48, and calculate the ratio of the first measured value to the second measured value as the actual measured value of the operating parameter. The acquisition unit 12 acquires the actual measured values of the operating parameters in association with the acquisition date and time. The acquisition date and time may be the date and time when the operating parameters are measured by a sensor included in the plant equipment 48, or the date and time when the data management device 42 or the plant management device 44 acquires the actual measured values of the operating parameters. The operating data 26 may be generated by the data management device 42. In this case, the acquisition unit 12 may acquire the operating data 26 from the data management device 42.

[0036] The acquisition unit 12 stores the acquired actual measurement values of the operating parameters in the storage unit 20 as operating data 26. The operating data 26 may include time-series values of the operating parameters. The time-series values of the operating parameters include the actual measurement values of the operating parameters and the dates and times when the actual measurement values of the operating parameters were acquired. The acquisition unit 12 accumulates the acquired time-series values of the operating parameters in the storage unit 20, thereby generating operating data 26 consisting of time-series values of the operating parameters from the past to the present. The time-series values of the operating parameters recorded as the operating data 26 are acquired at relatively short time intervals (for example, every 1 minute, every 4 minutes, every 10 minutes, or every 30 minutes).

[0037] The acquisition unit 12 further acquires actual measured values of properties indicating the properties of the product oil produced in the plant 50. The acquisition unit 12 acquires, for example, actual measured values of properties indicating the properties of the product oil managed by the data management device 42. The acquisition unit 12 acquires the actual measured values of properties indicating the properties of the product oil in association with the acquisition date and time. The acquisition date and time may be the date and time when the product oil is sampled from the plant 50, the date and time when a test on the product oil is conducted, or the date and time when the actual measured values of the properties indicating the properties of the product oil are input into the user terminal 46. An example of the actual measured value of the property indicating the properties of the product oil is the boiling point of the second product oil 56 produced in the plant 50, for example, the equilibrium reflux boiling point of the second product oil 56.

[0038] The acquisition unit 12 stores the acquired actual measurement values of the properties indicating the properties of the produced oil in the memory unit 20 as actual measurement data 28. The actual measurement data 28 may include time series values of the properties indicating the properties of the produced oil. The time series values of the properties indicating the properties of the produced oil include the actual measurement values of the properties indicating the properties of the produced oil and the date and time of acquisition of the actual measurement values of the properties indicating the properties of the produced oil. In this case, the acquisition unit 12 accumulates the acquired time series values of the properties indicating the properties of the produced oil in the memory unit 20, thereby generating actual measurement data 28 consisting of time series values of the properties indicating the properties of the produced oil from the past to the present. The time series values of the properties indicating the properties of the produced oil recorded as actual measurement data 28 may be acquired periodically or irregularly. The actual measurement data 28 may be generated by the data management device 42. In this case, the acquisition unit 12 may acquire the actual measurement data 28 from the data management device 42.

[0039] The estimation unit 14 uses the operating data 26 to estimate a value of a property indicating the properties of the product oil produced in the plant 50. The property estimated by the estimation unit 14 is the boiling point of the second product oil 56, for example, the equilibrium reflux boiling point of the second product oil 56. The estimation unit 14 may estimate the value of the property indicating the properties of the product oil from the operating data 26 using model data 30 stored in the memory unit 20. The model data 30 includes, for example, a prediction model that receives as input a set of actual measured values of multiple operating parameters included in the operating data 26 and outputs an estimated value of the property indicating the properties of the product oil. The number of multiple operating parameters used by the estimation unit 14 for estimation may be 5 or more, 10 or more, or 15 or more, or may be 100 or less, 50 or less, or 30 or less.

[0040] The first estimation unit 22 uses the operating data 26 to estimate the value of a first characteristic indicating the properties of the product oil produced in the plant 50. The first characteristic indicates, for example, the distillation properties of the product oil, and indicates the distillation temperature or distillation ratio of the product oil. The first characteristic is a characteristic different from the boiling point, but can be said to be a characteristic similar to the boiling point in that it is a characteristic related to temperature. As the distillation temperature, for example, the initial boiling point temperature, the X% distillation temperature (X = 5, 10, 20, 30, 40, 50, 60, 70, 90, 95, 97, etc.), the end temperature, etc. can be used. As the distillation ratio, for example, the distillation ratio of Y°C or higher (Y = 20, 25, 30, 35, 40, 45, 50, etc.) can be used. The first estimation unit 22 may use the operating data 26 to estimate multiple values of the first characteristic of the product oil, or may estimate multiple distillation temperatures with different ratios X, or may estimate multiple distillation ratios with different temperatures Y. The first estimation unit 22 may estimate at least one distillation temperature of the product oil and at least one distillation ratio of the product oil using the operating data 26. The number of the multiple first characteristics estimated by the first estimation unit 22 may be 2 or more, 5 or more, or 8 or more, or may be 20 or less, 15 or less, or 10 or less.

[0041] The first estimation unit 22 may output an estimated value of the first characteristic using a first model. The first model indicates a correlation between values of a plurality of parameters and values of the first characteristic. The first model is generated by, for example, a model generation unit 32 and stored in the storage unit 20 as model data 30. The first model receives, for example, a set of actual measured values of a plurality of driving parameters included in the driving data 26 as input, and outputs an estimated value of the first characteristic.

[0042] The second estimation unit 24 estimates the value of a second characteristic indicating the properties of the product oil produced in the plant 50 using the estimated value of the first characteristic output from the first estimation unit 22. The second characteristic indicates the boiling point of the product oil, for example, the equilibrium reflux boiling point. The second estimation unit 24 may output the estimated value of the second characteristic using a second model. The second model indicates the correlation between the estimated value of the first characteristic output from the first model and the actual measured value of the second characteristic. The second model is generated, for example, by the model generation unit 32 and stored in the memory unit 20 as model data 30.

[0043] The estimation unit 14 may output an estimated value of the second characteristic using a prediction model that combines the first model and the second model. By using the prediction model, the estimation unit 14 can directly output an estimated value of the second characteristic of the produced oil using as input a set of actual measured values of multiple parameters included in the operating data 26. The prediction model is generated, for example, by the model generation unit 32 and stored in the storage unit 20 as model data 30.

[0044] According to one example of a predictive model constructed for the above-mentioned plant 50, by inputting a set of approximately 20 measured values of operating parameters and using approximately 3 to 10 distillation properties (first characteristics of the second product oil 56) as intermediate variables, the root mean square error (RMSE) of the estimated value of the equilibrium reflux boiling point of the second product oil 56 can be reduced to 1°C or less (e.g., 0.6°C).

[0045] The calculation unit 16 uses the estimated value of the second characteristic output from the estimation unit 14 to calculate the amount of additive to be added to the product oil produced in the plant 50. The calculation unit 16 calculates the amount of additive to be added using the estimated value of the second characteristic and a predetermined control value of the second characteristic. For example, the calculation unit 16 uses the estimated value of the boiling point of the second product oil 56 and a predetermined control value of the boiling point to calculate the amount of additive 80 to be added so that the boiling point of the product oil 82 after the addition of the additive 80 is equal to or higher than the predetermined control value. The predetermined control value can be predetermined based on the standard value for special flammable materials specified in the Fire Service Act (e.g., 40°C), and can be set to a value slightly larger than the standard value, for example, 41°C or 42°C.

[0046] The notification unit 18 notifies the user of the estimated value output from the estimation unit 14 and the amount of additive added calculated by the calculation unit 16. The notification unit 18 may output data for displaying the estimated value and the amount of additive added on the user terminal 46. The notification unit 18 may notify the plant 50 of the amount of additive added, and control the amount of additive 80 added in the plant 50.

[0047] The notification unit 18 may notify the user when the estimated value of the second characteristic output from the estimation unit 14 is equal to or smaller than a predetermined threshold. The notification unit 18 may, for example, send an alert to the user terminal 46. Here, the predetermined threshold may be set to the same value as the above-mentioned predetermined reference value (e.g., 40°C), or may be set to a value slightly smaller than the predetermined reference value (e.g., 39°C or 38°C).

[0048] The model generation unit 32 generates a first model and a second model to be used in the estimation unit 14 .

[0049] The model generation unit 32 can generate the first model by machine learning using training data in which a set of values of a plurality of operating parameters is input and a value of the first characteristic of the produced oil is output. The model generation unit 32 can use at least one of linear regression, support vector machine, random forest, gradient boosting, and neural network as a machine learning algorithm for generating the first model.

[0050] To generate training data to be used in generating the first model, the model generation unit 32 may execute a simulation using a plant simulator that simulates the state of the plant 50. In this case, by inputting multiple sets of values of multiple operating parameters into the plant simulator, simulated values of the first characteristic corresponding to each of the multiple sets can be generated.

[0051] The model generation unit 32 can generate the first model by using the simulation results of the plant simulator as training data. The model generation unit 32 can generate the first model by using the training data to derive a correlation between a simulation value of the first characteristic and a set of values of a plurality of operating parameters corresponding to the simulation value of the first characteristic. Note that if training data to be used to generate the first model is prepared in advance, the model generation unit 32 may omit generating the training data. In this case, the training data to be used to generate the first model is acquired by the acquisition unit 12.

[0052] The model generation unit 32 can generate the second model by machine learning using training data in which the value of the first characteristic of the produced oil is input and the value of the second characteristic of the produced oil is output. The model generation unit 32 can use at least one of linear regression, support vector machine, random forest, gradient boosting, and neural network as a machine learning algorithm for generating the second model.

[0053] The model generation unit 32 may use the estimated value of the first characteristic output from the first model to generate training data to be used in generating the second model. In this case, the model generation unit 32 generates an estimated value of the first characteristic corresponding to the actual measured value of the second characteristic by inputting a set of actual measured values of a plurality of operating parameters corresponding to the actual measured value of the second characteristic to the first model, for example, based on the operating data 26 and the actual measured value data 28. The model generation unit 32 may generate in advance a combination of the actual measured value of a specific second characteristic and a set of actual measured values of a plurality of operating parameters corresponding to the actual measured value of the specific second characteristic, for example, based on the time-series values of the plurality of operating parameters included in the operating data 26 and the time-series values of the second characteristic included in the actual measured value data 28.

[0054] The model generation unit 32 can generate the second model by using the training data to derive a correlation between the actual measured value of the second characteristic and the estimated value of the first characteristic corresponding to the actual measured value of the second characteristic. Note that if the training data to be used for generating the second model is prepared in advance, the model generation unit 32 may omit generating the training data. In this case, the training data to be used for generating the second model is acquired by the acquisition unit 12.

[0055] The model generation unit 32 may generate a prediction model that combines the first model and the second model. The model generation unit 32 can generate a prediction model by combining the first model and the second model. By combining the first model and the second model, the prediction model is a model that uses the measured values of multiple operating parameters as input and the value of the second property of the produced oil as output.

[0056] The information processing device 10 does not need to include the model generation unit 32. In this case, the information processing device 10 can acquire the first model, the second model, or the prediction model generated by a device different from the information processing device 10 as model data 30, and output an estimated value of the second characteristic using the acquired model data 30.

[0057] 3 is a flowchart showing an example of a property estimation method according to an embodiment. The acquisition unit 12 acquires actual measured values of a plurality of operating parameters that indicate the operating conditions of the plant 50 (S10). The first estimation unit 22 receives the acquired actual measured values of the plurality of operating parameters as input and outputs an estimated value of a first property that indicates the properties of the produced oil (S12). The second estimation unit 24 receives the estimated value of the first property output from the first estimation unit 22 as input and outputs an estimated value of a second property that indicates the properties of the produced oil (S14).

[0058] FIG. 4 is a flowchart showing an example of a model generation method according to an embodiment. The model generation unit 32 inputs values of a plurality of operating parameters indicating operating conditions into a plant simulator and calculates a simulated value of a first characteristic indicating the properties of the produced oil (S20). The model generation unit 32 uses the simulation results to generate a first model that receives the values of the plurality of operating parameters as input and outputs a value of the first characteristic (S22). The acquisition unit 12 acquires a set of actual measured values of a plurality of operating parameters indicating the actual operating conditions of the plant 50 (S24). The model generation unit 32 inputs the acquired set of actual measured values of the plurality of operating parameters into the first model and acquires an estimated value of the first characteristic (S26). The acquisition unit 12 acquires an actual measured value of a second characteristic indicating the properties of the produced oil corresponding to the acquired set of actual measured values of the plurality of operating parameters (S28). The model generation unit 32 generates a second model that indicates the correlation between the estimated value of the first characteristic and the actual measured value of the second characteristic (S30).

[0059] According to this embodiment, the properties of the product oil produced in the plant can be estimated in real time with high accuracy. Instead of directly estimating the desired properties from the operating parameters indicating the operating conditions of the plant, by using the first characteristic indicating the properties of the product oil as an intermediate variable, the prediction accuracy can be improved even when the number of data points of the actual measured values of the desired properties is small.

[0060] According to this embodiment, the desired properties of the produced oil can be estimated in real time with high accuracy, and therefore the amount of additives to be added to make the desired properties of the produced oil equal to or greater than a predetermined control value can be more appropriately calculated. As a result, the margin for the amount of additives to be added can be reduced, and the amount of additives to be added can be kept to the minimum necessary, thereby improving economy.

[0061] The present disclosure has been described above based on the embodiments. These embodiments are merely examples, and it will be understood by those skilled in the art that various modifications are possible in the combination of each component or each treatment process, and that such modifications are also within the scope of the present disclosure.

[0062] In the above-described embodiment, the plant is a distillation refinery apparatus, and the desired property is the boiling point. The property estimation method according to the present disclosure may be applied to estimating the properties of product oil produced in any type of plant. In this case, a property that is difficult to measure in real time can be adopted as the second property, and a property similar to the second property can be used as the first property. For example, the first property can be a property that shares units (temperature, pressure, flow rate, etc.) with the second property, or a property related to the units (temperature, pressure, flow rate, etc.) of the second property.

[0063] In the present disclosure, parameters or values may be expressed as absolute values, relative values from a predetermined value, or other corresponding information.

[0064] The embodiment may be a program for causing a computer to implement the functions for implementing the above-described method, or a recording medium for storing the program. The recording medium for storing such a program may be a non-transitory, tangible, computer-readable storage medium, such as a non-volatile memory, a magnetic recording medium such as a magnetic tape or a magnetic disk, or an optical recording medium such as an optical disk.

[0065] Several aspects of the present disclosure are described below.

[0066] A first aspect is an information processing device including: a first estimation unit that receives as input actual measured values of a plurality of operating parameters that indicate the operating conditions of a plant and outputs an estimated value of a first characteristic that indicates the properties of a product oil produced in the plant; and a second estimation unit that receives as input the estimated value of the first characteristic and outputs an estimated value of a second characteristic that indicates the properties of the product oil. According to the first aspect, even if the number of data points for the actual measured values of the second characteristic is small, the value of the second characteristic can be estimated with high accuracy by estimating the value of the second characteristic via the first characteristic.

[0067] A second aspect is the information processing device according to the first aspect, wherein the first estimation unit outputs an estimate of the first characteristic using a first model generated from a simulation result of a plant simulator that simulates the plant. According to the second aspect, by using the simulation result, the first model can be generated using a larger number of operating conditions than actual operating conditions, thereby improving the estimation accuracy of the first model.

[0068] A third aspect is the information processing device according to the second aspect, wherein the second estimation unit outputs the estimated value of the second characteristic using a second model that indicates a correlation between the estimated value of the first characteristic output from the first model when a set of actual measured values of the plurality of operating parameters indicating actual operating conditions of the plant is input to the first model, and the actual measured value of the second characteristic of the product oil corresponding to the set. According to the third aspect, by generating the second model via the first model with high estimation accuracy, the estimation accuracy of the second model can be improved.

[0069] In a fourth aspect, the plant is a distillation refinery that fractionates a feedstock oil to produce the product oil, and the plurality of operating parameters include at least one of a parameter related to temperature, a parameter related to pressure, a parameter related to flow rate, and a parameter related to a reflux ratio. According to the fourth aspect, the information processing device of any one of the first to third aspects can estimate the second property of the product oil produced in the distillation refinery with high accuracy.

[0070] In a fifth aspect, the first characteristic represents a distillation property of the product oil, and the second characteristic represents a boiling point of the product oil. According to the fifth aspect, when attempting to estimate the boiling point of the product oil, the accuracy of estimating the boiling point can be improved by using the distillation property related to temperature as an intermediary.

[0071] In a sixth aspect, in the information processing device according to the fifth aspect, the second characteristic indicates the equilibrium reflux boiling point of the produced oil. According to the sixth aspect, the equilibrium reflux boiling point, which indicates a more accurate boiling point while requiring more time for measurement, can be estimated with high accuracy and quickly.

[0072] A seventh aspect is the information processing device according to the fifth or sixth aspect, further comprising a notification unit that notifies a user when the estimated value of the second characteristic output from the second estimation unit is equal to or less than a predetermined threshold. According to the seventh aspect, the user can be alerted when the boiling point of the product oil is equal to or less than the predetermined threshold, thereby improving user convenience.

[0073] An eighth aspect is the information processing device according to any one of the fifth to seventh aspects, further comprising a calculation unit that calculates an amount of additive to be added to the product oil using the estimated value of the second characteristic output from the second estimation unit, wherein the value of the second characteristic of the additive is greater than the value of the second characteristic of the product oil. According to the eighth aspect, the amount of additive to be added for adjusting the boiling point of the product oil can be appropriately calculated, thereby improving user convenience.

[0074] A ninth aspect is the information processing device according to the eighth aspect, wherein the calculation unit calculates the amount of additive to be added by using the estimated value of the second characteristic output from the second estimation unit and a predetermined control value of the second characteristic. According to the ninth aspect, it is possible to appropriately calculate the amount of additive to be added for adjusting the boiling point of the refined oil so that it satisfies the control value, thereby improving user convenience.

[0075] A tenth aspect is a property estimation method comprising the steps of: inputting actual measurement values of a plurality of operating parameters indicating the operating conditions of a plant and outputting an estimated value of a first characteristic indicating the properties of a product oil produced in the plant; and inputting the estimated value of the first characteristic and outputting an estimated value of a second characteristic indicating the properties of the product oil. According to the tenth aspect, even if the number of data points for the actual measurement values of the second characteristic is small, the value of the second characteristic can be estimated with high accuracy by estimating the value of the second characteristic via the first characteristic.

[0076] An eleventh aspect is a program that causes a computer to implement the following functions: inputting actual measured values of a plurality of operating parameters that indicate the operating conditions of a plant and outputting an estimated value of a first characteristic that indicates the properties of a product oil produced in the plant; and inputting the estimated value of the first characteristic and outputting an estimated value of a second characteristic that indicates the properties of the product oil. According to the eleventh aspect, even if the number of data points for the actual measured values of the second characteristic is small, the value of the second characteristic can be estimated with high accuracy by estimating the value of the second characteristic via the first characteristic.

[0077] A twelfth aspect of the present invention is a model generation method comprising the steps of: generating a first model using simulation results from a plant simulator that simulates a plant; inputting actual measured values of a plurality of operating parameters that indicate the operating conditions of the plant; and outputting an estimated value of a first characteristic that indicates the properties of a product oil produced by the plant; acquiring an estimated value of the first characteristic that is output from the first model when a set of actual measured values of the plurality of operating parameters that indicate the actual operating conditions of the plant is input to the first model; acquiring actual measured values of a second characteristic that indicates the properties of the product oil, the actual measured values corresponding to the set; and generating a second model that indicates a correlation between the estimated value of the first characteristic and the actual measured value of the second characteristic. According to the twelfth aspect, by using the simulation results, the first model can be generated using a greater number of operating conditions than the actual operating conditions, thereby improving the estimation accuracy of the first model. Even when the number of data points for the actual measured values of the second characteristic is small, the estimation accuracy of the second model can be improved by estimating the value of the second characteristic via the first characteristic.

[0078] A thirteenth aspect is the model generation method according to the twelfth aspect, further comprising the step of generating a prediction model using the first model and the second model, the prediction model receiving actual measured values of the plurality of operating parameters as input and outputting an estimated value of the second characteristic. According to the thirteenth aspect, by integrating the first model and the second model, it is possible to provide a prediction model that directly outputs an estimated value of the second characteristic from the values of the operating parameters.

[0079] Any combination of the configurations according to the above-described embodiments or aspects is also useful as an embodiment of the present disclosure. A new embodiment resulting from the combination will have the combined effects of the combined examples and modifications. It will also be understood by those skilled in the art that the functions to be performed by each component in the claims can be realized by each component shown in the examples and modifications, either individually or in combination. [Explanation of symbols]

[0080] 10...information processing device, 12...acquisition unit, 14...estimation unit, 16...calculation unit, 18...notification unit, 22...first estimation unit, 24...second estimation unit, 50...plant, 52...raw material oil, 56...second product oil, 80...additive, 82...product oil.

Claims

1. a first estimation unit that receives as input actual measured values of a plurality of operating parameters that indicate operating conditions of the plant and outputs an estimated value of a first characteristic that indicates the properties of a product oil produced in the plant; a second estimation unit that receives the estimated value of the first characteristic as an input and outputs an estimated value of a second characteristic that indicates the properties of the produced oil; Information processing device.

2. the first estimation unit outputs an estimated value of the first characteristic using a first model generated from a simulation result of a plant simulator that simulates the plant. The information processing device according to claim 1 .

3. the second estimation unit outputs the estimated value of the second characteristic using a second model that indicates a correlation between the estimated value of the first characteristic output from the first model when a set of actual measurement values of the plurality of operating parameters that indicate actual operating conditions of the plant is input to the first model, and the actual measurement value of the second characteristic of the product oil corresponding to the set. The information processing device according to claim 2 .

4. The plant is a distillation and refining apparatus that fractionates a feedstock oil to produce the product oil, the plurality of operating parameters include at least one of a parameter related to temperature, a parameter related to pressure, a parameter related to flow rate, and a parameter related to reflux ratio; The information processing device according to claim 1 .

5. The first property indicates the distillation property of the product oil, and the second property indicates the boiling point of the product oil. The information processing device according to claim 4 .

6. The second property indicates the equilibrium reflux boiling point of the product oil. The information processing device according to claim 5 .

7. a notification unit that notifies a user when the estimated value of the second characteristic output from the second estimation unit is equal to or smaller than a predetermined threshold value; The information processing device according to claim 5 .

8. Further, a calculation unit is provided that calculates the amount of additive to be added to the refined oil using the estimated value of the second characteristic output from the second estimation unit, the value of the second property of the additive is greater than the value of the second property of the product oil; The information processing device according to claim 5 .

9. the calculation unit calculates the amount of addition using the estimated value of the second characteristic output from the second estimation unit and a predetermined control value of the second characteristic. The information processing device according to claim 8 .

10. a step of receiving actual measured values of a plurality of operating parameters indicating operating conditions of a plant and outputting an estimated value of a first characteristic indicating properties of a product oil produced in the plant; and a step of inputting the estimated value of the first characteristic and outputting an estimated value of a second characteristic indicating the properties of the produced oil. Property estimation method.

11. a function of receiving as input actual measured values of a plurality of operating parameters indicating operating conditions of a plant and outputting an estimated value of a first characteristic indicating properties of a product oil produced in the plant; a function of inputting the estimated value of the first characteristic and outputting an estimated value of a second characteristic indicating the properties of the produced oil; A program that makes the computer realize the above.

12. generating a first model that uses simulation results of a plant simulator that simulates a plant, inputs actual measured values of a plurality of operating parameters that indicate operating conditions of the plant, and outputs an estimated value of a first characteristic that indicates the properties of a product oil produced in the plant; acquiring an estimated value of the first characteristic output from the first model when a set of actual measured values of the plurality of operating parameters indicating actual operating conditions of the plant is input to the first model; acquiring actual measured values of a second characteristic indicating the properties of the product oil, the actual measured values corresponding to the set; generating a second model that indicates a correlation between the estimated value of the first characteristic and the actual measured value of the second characteristic; Model generation method.

13. a step of generating a prediction model using the first model and the second model, the prediction model receiving actual measured values of the plurality of operating parameters as input and outputting an estimated value of the second characteristic; The model generation method of claim 12.

Images