Program, information processing device, and method

Abstract

To provide an information processing device, a method, and a program that can accurately predict a reaction state of heavy oil decomposition and control reaction.SOLUTION: An information processing device equipped with a processor for predicting a reaction state of a heavy oil executes steps of: accepting, as training data, oil properties of heavy oil to be subjected to heavy oil decomposition, operational parameters related to a heavy oil decomposition apparatus for performing the heavy oil decomposition, and values indicating a reaction state when the heavy oil is decomposed in the heavy oil decomposition apparatus according to the operational parameters (S101); training a first model using the training data to output the reaction state of the heavy oil decomposition in response to input of the oil properties of the heavy oil and the operational parameters (S102); and storing the trained first model in a storage unit (S103).SELECTED DRAWING: Figure 5

Description

[Technical Field]

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

[0002] To provide a method for predicting the operating conditions of petroleum refineries, which allows for a simple prediction of the operating conditions of the equipment that will give the optimal value of the objective function in linear programming, there is a technique for simulating the relationship between the operating conditions and the product yield. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2002-329187 Summary of the Invention [Problem to be solved by the invention]

[0004] In conventional technology, the reaction state of heavy oil, such as index values ​​for reaction rate parameters, is determined by simulation, which has the problem that it is not possible to predict the reaction state more accurately.

[0005] An object of the present disclosure is to provide a technology that can accurately predict the reaction state of heavy oil cracking and control the reaction. [Means for solving the problem]

[0006] The information processing device according to the present disclosure is an information processing device equipped with a processor, and executes the steps of accepting input, as learning data, the oil properties of heavy oil to be cracked, operating parameters for a heavy oil cracking device that performs the heavy oil cracking, and values ​​that indicate the reaction state when the heavy oil is cracked in the heavy oil cracking device using the operating parameters; learning a first model using the learning data that outputs the reaction state of the heavy oil cracking in response to input of the oil properties of the heavy oil and the operating parameters; and storing the learned first model in a memory unit. [Effects of the Invention]

[0007] According to the program of the present disclosure, the reaction state of heavy oil cracking can be predicted with high accuracy. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a block diagram showing a configuration of an information processing system 1. FIG. [Figure 2] 1 is a block diagram showing a configuration of an information processing device 10. FIG. [Figure 3] FIG. 2 is a diagram showing an example of the configuration of a heavy oil cracking unit 20. [Figure 4] 1 is a block diagram showing a functional configuration of an information processing device 10. FIG. [Figure 5] 10 is a flowchart showing an example of the flow of learning processing by the information processing device 10. [Figure 6] 10 is a flowchart showing an example of a flow of prediction processing performed by the information processing device 10. [Figure 7] 10 is a flowchart showing an example of the flow of a re-learning process performed by the information processing device 10. [Figure 8] 10 is a flowchart showing an example of the flow of a product yield calculation process performed by the information processing device 10. [Figure 9] 10 is a flowchart showing an example of a flow of optimization processing performed by the information processing device 10. [Figure 10]FIG. 2 is a diagram showing an example of the configuration of a heavy oil cracking unit 20. [Figure 11] FIG. 2 is a diagram showing an example of the configuration of a heavy oil cracking unit 20. DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. In the following description, the same components are denoted by the same reference numerals. The names and functions of the components are also the same. Therefore, detailed descriptions thereof will not be repeated.

[0010] <Summary of this disclosure> This disclosure describes a technology for predicting values ​​indicating the reaction state when heavy oil is cracked based on the operating parameters of a heavy oil cracking unit that cracks heavy oil and the oil properties of the heavy oil.

[0011] In this disclosure, heavy oil refers to oil extracted from the bottom of an atmospheric distillation tower or a vacuum distillation tower when crude oil is distilled, or a crude oil equivalent thereto. In this disclosure, heavy oil is described using an example in which the heavy oil is extracted from the bottom of an atmospheric distillation tower. Because the cracking reaction of heavy oil is a complex reaction affected by various operating parameters, it has been difficult to quantify, control, and optimize the reaction state in real time during actual operation. Because the reaction state changes from moment to moment in an actual heavy oil cracking unit during operation, it has been difficult to accurately predict it simply by performing a simulation.

[0012] In the present disclosure, a model is prepared that learns the relationship between values ​​indicating the reaction state of heavy oil and operating parameters and the oil properties of heavy oil. The reaction state of heavy oil is, for example, a parameter related to the reaction rate of oily components in heavy oil, an index value indicating equilibrium catalytic activity, etc. The index value indicating equilibrium catalytic activity is, for example, the equilibrium catalytic activity value obtained by analysis, the amount of metal components deposited on the catalyst, etc.

[0013] Using this model, the information processing system 1 predicts and quantifies values ​​indicating the reaction state of heavy oil in real time. The values ​​indicating the reaction state and operational parameters are input into a pre-prepared physical model that predicts product yields, making it possible to control the reaction. In this way, optimizing the cracking reaction of heavy oil contributes to improving product yields and reducing OPEX, such as by reducing catalyst input, thereby contributing to improving the profitability of the refinery.

[0014] <1. Configuration of Information Processing System 1> An information processing system 1 according to the present disclosure will be described with reference to Fig. 1. The information processing system 1 according to the present disclosure includes an information processing device 10, a heavy oil cracking unit 20, a user terminal 30, and a network 40.

[0015] 2 is a diagram showing the configuration of the information processing device 10. The information processing device 10 is, for example, a laptop computer, a rack-mounted or tower-type computer, a smartphone, or the like. The information processing device 10 may also be configured as a single system by a plurality of information processing devices, or may have a redundant configuration. The manner in which the multiple functions required to realize the information processing device 10 are allocated can be determined appropriately in consideration of the processing capabilities of each piece of hardware, the specifications required for the information processing device 10, and the like.

[0016] The information processing device 10 includes a processor 11, a memory 12, a storage 13, a communication IF 14, and an input / output IF 15.

[0017] The processor 11 is hardware for executing an instruction set written in a program, and is composed of an arithmetic unit, a register, a peripheral circuit, and the like.

[0018] The memory 12 is for temporarily storing programs and data to be processed by the programs, and is a volatile memory such as a DRAM (Dynamic Random Access Memory).

[0019] The storage 13 is a storage device for saving data, such as a flash memory, a hard disc drive (HDD), or a solid state drive (SSD).

[0020] The communication IF 14 is an interface for inputting and outputting signals so that the information processing device 10 can communicate with external devices. The communication IF 14 is connected to a network 40 such as a LAN, the Internet, or a wide area Ethernet by wire or wirelessly.

[0021] The input / output IF 15 functions as an interface with an input device (for example, a pointing device such as a mouse, a keyboard) for receiving an input operation, and an output device (for example, a display, a speaker, etc.) for presenting information.

[0022] The heavy oil cracking unit 20 is a unit that cracks heavy oil into lighter cracked oil through a predetermined cracking reaction. Cracked oil is, for example, light gas (including LPG), gasoline, middle distillate, etc. The heavy oil cracking unit 20 cracks heavy oil by, for example, fluid catalytic cracking (FCC), thermal cracking, hydrocracking, etc. In the following, FCC will be described as the heavy oil cracking unit 20.

[0023] The heavy oil cracking unit 20 has a function of controlling the reactor 21 and the regenerator 22, and a function of transmitting and receiving predetermined information to and from the information processing device 10 via communication.

[0024] Fig. 3 is a diagram showing a configuration example of a heavy oil cracking unit 20. The example shown in Fig. 3 is a case where the heavy oil cracking unit 20 cracks heavy oil by FCC. As shown in Fig. 3, the heavy oil cracking unit 20 includes a reactor 21 and a regenerator 22.

[0025] The reactor 21 is a device that brings the raw material heavy oil into contact with a catalyst to cause a cracking reaction and obtain cracked products. Specifically, the reactor 21 brings heavy oil, steam, and a catalyst into contact with the catalyst after the heavy oil, steam, and catalyst are input. The reactor 21 then produces cracked oil by cracking the heavy oil through a cracking reaction caused by the contact of the heavy oil with the catalyst. The reactor 21 also introduces steam into the resulting cracked oil to remove any cracked oil adhering to the catalyst. The reactor 21 then outputs the cracked oil. The reactor 21 also passes the used catalyst to the regenerator 22.

[0026] The regenerator 22 regenerates the catalyst used in the reactor 21. When a catalyst is used in the cracking reaction of heavy oil, coke (carbon) adheres to the catalyst surface, causing the catalyst to deteriorate. The regenerator 22 regenerates the catalyst by burning the coke adhered to the catalyst surface at high temperatures, and supplies the regenerated catalyst to the reactor 21 so as to maintain constant activity within the reactor 21. The regenerator 22 also emits exhaust gases produced by combustion.

[0027] The user terminal 30 is a terminal operated by a user. Here, the user is, for example, a person who operates and manages the heavy oil cracking unit 20. The user terminal 30 is, for example, a smartphone, a personal computer, or the like.

[0028] The information processing device 10, the heavy oil cracking unit 20, and the user terminal 30 are configured to be able to communicate with each other via a network 40.

[0029] 2. Functional configuration of information processing device 10

[0030] 4 is a block diagram showing the functional configuration of the information processing device 10. As shown in FIG.

[0031] The communication unit 110 performs processing for the information processing device 10 to communicate with external devices.

[0032] The storage unit 120 stores data and programs used by the information processing device 10. The storage unit 120 stores a training data DB 121, a model DB 122, and the like.

[0033] The learning data DB 121 is a database that holds learning data used when performing learning processing.

[0034] The learning data is a set of actual data values ​​indicating the oil properties of the heavy oil to be cracked, the operating parameters for the heavy oil cracking unit 20 that performs the heavy oil cracking, and the reaction state when the heavy oil is cracked in the heavy oil cracking unit 20 using the operating parameters.

[0035] The oil properties of the heavy oil are information about the oil properties of the heavy oil, such as the density, metal concentration, nitrogen concentration, etc. The operating parameters are, for example, parameters such as the heavy oil flow rate to the heavy oil cracking unit 20, the amount of catalyst, the amount of steam inflow, the pressure in the reactor 21, the internal temperature, the temperature of the catalyst, the ratio of catalyst to heavy oil, and parameters such as the pressure in the regenerator 22, the internal temperature, and the temperature of the catalyst.

[0036] For example, when the value indicating the reaction state is a parameter related to the reaction rate of heavy oil in heavy oil cracking, the learned data employs a first parameter related to the cracking reaction of heavy oil among the above operating parameters as the operating parameter.

[0037] For example, parameters related to the reaction rate include the reaction rate constant, frequency factor, activation energy, etc. The reaction rate constant of the Arrhenius equation is expressed, for example, by the following equation: In the following equation, k is the reaction rate constant, R is the gas constant, T is the absolute temperature, E is the activation energy, and A is the frequency factor.

number

[0038] Furthermore, when the value indicating the reaction state is an index value indicating equilibrium catalytic activity, the learned data employs, as the operating parameter, a second parameter relating to equilibrium catalytic activity among the operating parameters.

[0039] The model DB 122 is a database that holds various models and model parameters.

[0040] The model DB 122 holds a heavy oil reaction state prediction model (hereinafter referred to as the first model). The first model is a model that outputs the reaction state of heavy oil cracking in response to input of oil properties of heavy oil and operating parameters.

[0041] Specifically, the first model is a reaction rate constant prediction model (hereinafter referred to as the second model), which outputs a reaction rate constant in response to inputting the oil properties of heavy oil and the first parameter.

[0042] The first model may also be an equilibrium catalytic activity prediction model (hereinafter referred to as the third model), which outputs an index value indicating equilibrium catalytic activity in response to input of the oil properties of heavy oil and the second parameter.

[0043] Each of the first to third models may employ any model, such as a machine learning model or a neural network. The first to third models may represent the relationship between the oil properties of heavy oil, the operating parameters, and the reaction state of heavy oil cracking using, for example, a linear regression model. The model DB 122 may also hold both the second model and the third model. In the present disclosure, an example will be described in which the model DB 122 holds both the second model and the third model.

[0044] The model DB 122 also stores models other than those described above. For example, the model DB 122 stores a physical model that outputs a product yield or an index value that contributes to the product yield by inputting the oil properties of the heavy oil, the first parameter, the reaction rate constant of heavy oil cracking calculated by the second model, the second parameter, and an index value indicating the equilibrium catalytic activity calculated by the third model.

[0045] The control unit 130 performs functions such as a reception control unit 131, a transmission control unit 132, an input unit 133, a learning unit 134, an acquisition unit 135, a prediction unit 136, a determination unit 137, a calculation unit 138, an optimization unit 139, and an output unit 140 by the processor 11 of the information processing device 10 performing processing according to a program.

[0046] The reception control unit 131 controls the process in which the information processing device 10 receives a signal from an external device in accordance with a communication protocol.

[0047] The transmission control unit 132 controls the process in which the information processing device 10 transmits a signal to an external device in accordance with a communication protocol.

[0048] The input unit 133 receives input, as learning data, of the oil properties of the heavy oil to be cracked, operating parameters for the heavy oil cracking unit 20 that performs the heavy oil cracking, and values ​​that indicate the reaction state when the heavy oil is cracked in the heavy oil cracking unit 20 using the operating parameters. Specifically, the input unit 133 acquires the learning data from the learning data DB 121.

[0049] The learning unit 134 learns the second model and the third model using the acquired learning data.

[0050] Specifically, the learning unit 134 learns the second model using the oil properties of the heavy oil, the first parameter, and the reaction rate constant from the learning data. For example, if the second model is a linear regression model, the learning unit 134 learns the second model using the reaction rate constant as the objective variable and the oil properties of the heavy oil and the first parameter as explanatory variables.

[0051] Cracking reactions of heavy oil, which has numerous continuous boiling point components, are difficult to control due to the interaction of multiple operating parameters, not just the number of feedstock components. Because feedstock properties change frequently during actual operation, cracking reactions could not be optimized. This led to lower product yields and worsening profits. In particular, catalyst-based cracking reactions, such as FCC and hydrocracking, require consideration of catalyst degradation, which makes it difficult to accurately control the reaction, leading to increased OPEX, such as the excessive addition of new catalyst.

[0052] By training the second model, it is possible to learn the correlation between the reaction rate constants that change with changes in feedstock properties and the operating parameters. Therefore, by using the trained second model, it is possible to accurately predict the reaction rate constants in actual operation.

[0053] Here, the second model learns the correlation between the first parameters and the reaction rate constants of the cracking reaction when heavy oil cracking is performed with the first parameters for each oily component of heavy oil. For example, the learning unit 134 learns the correlation between the reaction rate constants and the operating parameters so that the second model outputs the reaction rate constants for each of the oily components of heavy oil, such as oil equivalent to vacuum residue (VR) and oil equivalent to vacuum gas oil (VGO), which react with other oily components through heavy oil cracking to turn into gasoline, coke, exhaust gas, etc.

[0054] By training the second model in this way, the learning unit 134 can learn not only the number of feedstock components in the cracking reaction of heavy oil, which has countless continuous boiling point components, but also the correlations between the interactions of multiple operating parameters. The second model trained in this way can accurately predict the reaction rate constant.

[0055] Then, the learning unit 134 stores the learned second model in the model DB 122.

[0056] Furthermore, the learning unit 134 learns a third model using the oil properties of the heavy oil, the second parameter, and the index value indicating equilibrium catalytic activity from the learning data. For example, if the third model is a linear regression model, the learning unit 134 learns the third model using the index value indicating equilibrium catalytic activity as the objective variable and the oil properties of the heavy oil and the second parameter as explanatory variables.

[0057] As in the above example, catalyst circulates within the reactor 21 and regenerator 22 of the heavy oil cracking unit 20. While catalyst degradation that occurs in the reactor 21 can be restored to some extent in the regenerator 22, even the regenerator 22 cannot completely restore the catalyst. Therefore, in order to maintain constant catalyst activity, it is necessary to supply fresh catalyst in addition to restoring it in the regenerator 22. However, the amount of fresh catalyst input has been determined based on past experience. In actual operation, the catalyst degradation rate varies depending on factors such as changes in the heavy oil being processed. This has made it difficult to adjust the catalyst activity to the desired level.

[0058] By learning the above-mentioned third model, it is possible to learn the correlation between the oil properties of heavy oil in actual operation, the second parameter, and the index value indicating equilibrium catalytic activity. Therefore, by using the learned third model, it is possible to accurately predict the index value indicating equilibrium catalytic activity in actual operation. This has the effects of stabilizing equilibrium catalytic activity, optimizing the amount of new catalyst added, and reducing the cost and effort of equilibrium catalyst analysis. In particular, by using the learned third model, it is not necessary to consider the time required for analysis, so equilibrium catalytic activity can be obtained in real time if the oil properties of heavy oil in actual operation and the second parameter are known.

[0059] Then, the learning unit 134 stores the learned third model in the model DB 122.

[0060] The learning unit 134 can also re-learn the first model using the learning data, the third parameter which is an operating parameter for the heavy oil cracking unit 20 currently performing heavy oil cracking, and a value indicating the predicted reaction state. The learning unit 134 performs re-learning, for example, when the analysis results are obtained.

[0061] Specifically, the learning unit 134 acquires the learning data, third parameters, which are operating parameters for the heavy oil cracking unit 20 that is actually in operation and are described below, and values ​​indicating the reaction state obtained by analysis, described below, and re-learns each model. For example, the learning unit 134 re-learns the second model from the learning data, a first parameter of the third parameters, and the analyzed reaction rate constant. Also, for example, the learning unit 134 re-learns the third model from the learning data, a second parameter of the third parameters, and the analyzed index value indicating the equilibrium catalytic activity.

[0062] In this way, the learning unit 134 can re-learn the second and third models when, for example, analysis results are acquired, thereby obtaining index values ​​indicating the reaction rate constant and equilibrium catalytic activity with greater accuracy in real time. For example, to obtain a predetermined equilibrium catalytic activity, it is necessary to adjust the amount of new catalyst added. To accurately determine the amount of new catalyst added, analysis results of the equilibrium catalytic activity are required. The analysis results of the equilibrium catalytic activity require time after the catalyst is removed for analysis. Therefore, it takes time to reflect the analysis results in the simulation used for prediction. However, because the learning unit 134 can re-learn the second and third models when analysis results are acquired, it is possible to obtain index values ​​indicating the reaction rate and equilibrium catalytic activity with greater accuracy in real time, and the prediction accuracy can quickly follow the effects of, for example, catalyst changes.

[0063] The acquisition unit 135 acquires third parameters, which are operating parameters for the heavy oil cracking unit currently performing heavy oil cracking, the oil properties of the heavy oil, and the trained second and third models.

[0064] Specifically, the acquisition unit 135 acquires the third parameter and the oil properties of the heavy oil to be input into the heavy oil cracking unit 20 from the heavy oil cracking unit 20. The acquisition unit 135 may acquire the third parameter and the oil properties of the heavy oil by receiving them from the user terminal 30. The acquisition unit 135 also acquires the trained second model and third model from the model DB 122.

[0065] The acquisition unit 135 also acquires the analysis results of the reaction state. Specifically, the acquisition unit 135 acquires the analysis results of the reaction state based on the actual operating parameters, decomposition products, catalysts, etc.

[0066] The prediction unit 136 predicts a value indicating the reaction state by inputting the oil properties of the heavy oil and the third parameter into the first model.

[0067] Specifically, the prediction unit 136 predicts the reaction rate constant by inputting the oil properties of the heavy oil and the first parameter of the third parameters into the second model.

[0068] Furthermore, the prediction unit 136 predicts an index value indicating equilibrium catalytic activity by inputting the oil properties of the heavy oil and the second parameter of the third parameters into the third model.

[0069] The determination unit 137 determines the optimum amount of new catalyst to be introduced into the heavy oil cracking unit 20 based on the index value indicating the equilibrium catalyst activity.

[0070] Specifically, the determination unit 137 calculates the amount of catalyst to be added to bring the equilibrium catalyst activity to a predetermined value based on the index value indicating the equilibrium catalyst activity, and then determines the calculated amount as the optimal amount of new catalyst.

[0071] The calculation unit 138 uses the oil properties of the heavy oil, the third parameter, and the value indicating the reaction state to determine the product yield or an index value that contributes to the product yield.

[0072] Specifically, the calculation unit 138 first acquires a reaction model from the model DB 122. Next, the calculation unit 138 calculates the product yield or an index value contributing to the product yield using the oil properties of the heavy oil, the first parameter of the third parameter, the calculated reaction rate constant for heavy oil cracking, the second parameter of the third parameter, the calculated index value indicating the equilibrium catalytic activity, and the reaction model. The product yield is, for example, the yield of cracked oil obtained by the heavy oil cracking unit 20. The index value contributing to the product yield is, for example, an operating parameter required to calculate the product yield. The index value is, for example, the reaction temperature, reaction time, catalyst concentration, etc.

[0073] The optimization unit 139 determines the operating parameters that will achieve the optimal product yield using the determined product yield or an index value that contributes to the product yield, the oil properties of the heavy oil, the operating parameters, and values ​​indicating the reaction state.

[0074] Specifically, the optimization unit 139 first finds an optimal product yield using the product yield or an index value contributing to the product yield, the oil properties of the heavy oil, the first parameter and reaction rate constant of the third parameter, and the second parameter and index value indicating the equilibrium catalytic activity. For example, the optimization unit 139 searches for the optimal product yield by inputting the oil properties of the heavy oil, the third parameter, the found reaction rate, and the found equilibrium catalytic activity into a model for searching for the optimal product yield.

[0075] Next, the optimization unit 139 determines optimal operating parameters using the oil properties of the heavy oil, the first parameter of the third parameter, the reaction rate, the second parameter of the third parameter, an index value indicating equilibrium catalytic activity, and the determined optimal product type ratio. Note that the optimization unit 139 may also be configured to determine optimal operating parameters using the trained second and third models.

[0076] The output unit 140 outputs the predicted reaction rate constant and the index value indicating the equilibrium catalytic activity to an output device, etc. The predicted reaction rate constant and the index value indicating the equilibrium catalytic activity may be configured to be output by the output unit 140 to an external device via communication and displayed.

[0077] The output unit 140 outputs the determined amount of new catalyst. In response to this, the user issues a catalyst injection command to a catalyst injection device (not shown) from the user terminal 30 or the like, whereby new catalyst is injected into the heavy oil cracking unit 20.

[0078] Furthermore, the output unit 140 outputs the determined product yield or an index value that contributes to the product yield.

[0079] Furthermore, the output unit 140 outputs the optimum operating parameters that have been found.

[0080] <3.Operation> The processing in the information processing device 10 will be described below with reference to the drawings.

[0081] <3.1. Learning process> 5 is a flowchart showing an example of the flow of learning processing performed by the information processing device 10. The information processing device 10 executes this processing at any timing (for example, upon receiving a learning processing start signal, etc.).

[0082] In step S101, the input unit 133 receives input, as learning data, of the oil properties of the heavy oil to be subjected to heavy oil cracking, operating parameters for the heavy oil cracking unit 20 that performs the heavy oil cracking, and values ​​that indicate the reaction state when the heavy oil is cracked in the heavy oil cracking unit 20 using the operating parameters.

[0083] In step S102, the learning unit 134 learns the second model and the third model using the acquired learning data.

[0084] In step S103, the learning unit 134 stores the learned second and third models in the model DB 122, and then ends the process.

[0085] <3.2. Prediction processing> 6 is a flowchart showing an example of the flow of prediction processing performed by the information processing device 10. The information processing device 10 executes this processing by inputting the oil properties of heavy oil and operating parameters, etc.

[0086] In step S201, the acquisition unit 135 acquires third parameters, which are operating parameters for the heavy oil cracking unit 20 currently performing heavy oil cracking, the oil properties of the heavy oil, and the learned second and third models.

[0087] In step S202, the prediction unit 136 predicts the reaction rate constant by inputting the oil properties of the heavy oil and the first parameter of the third parameters into the second model.

[0088] In step S203, the prediction unit 136 predicts an index value indicating equilibrium catalytic activity by inputting the oil properties of the heavy oil and the second parameter of the third parameters into the third model.

[0089] In step S204, the output unit 140 outputs the predicted reaction rate constant and the index value indicating the equilibrium catalytic activity to an output device or the like.

[0090] The information processing device 10 can execute the prediction process at any timing, and can therefore predict values ​​indicating the reaction state of the heavy oil cracking unit 20 in operation in real time.

[0091] <3.3. Re-learning process> 7 is a flowchart showing an example of the flow of the relearning process performed by the information processing device 10. The information processing device 10 executes this process at any timing (for example, upon obtaining the analysis result of the reaction state, receiving a signal to start the relearning process, etc.).

[0092] In step S211, the acquisition unit 135 acquires a third parameter, which is an operating parameter for the heavy oil cracking unit 20 currently performing heavy oil cracking, the oil properties of the heavy oil, the reaction state obtained by analysis, and the learned second and third models.

[0093] In step S212, the learning unit 134 re-learns the second model using the learning data, the third parameters which are operating parameters for the heavy oil cracking unit 20 currently performing heavy oil cracking, and the analyzed reaction rate constants.

[0094] In step S213, the learning unit 134 re-learns the third model using the learning data, the third parameters which are operating parameters for the heavy oil cracking unit 20 currently performing heavy oil cracking, and an index value indicating the analyzed equilibrium catalytic activity.

[0095] In step S214, the learning unit 134 stores the re-learned second and third models in the model DB 122, and then ends the process.

[0096] The information processing device 10 can execute the prediction process at any timing, and can therefore predict values ​​indicating the reaction state of the heavy oil cracking unit 20 in operation in real time.

[0097] <3.4. Product Yield Calculation Processing> 8 is a flowchart showing an example of the flow of a product yield calculation process performed by the information processing device 10. The information processing device 10 executes this process by inputting the oil properties of the heavy oil and the operating parameters, etc. Note that processes common to the prediction process are assigned the same reference numerals and will not be described again.

[0098] In S304, the calculation unit 138 calculates the product yield or an index value that contributes to the product yield using the oil properties of the heavy oil, the third parameter, and the value indicating the reaction state.

[0099] In S305, the output unit 140 outputs the calculated product yield or the index value that contributes to the product yield, and the process ends.

[0100] <3.5. Optimization Processing> 9 is a flowchart showing an example of the flow of optimization calculation processing performed by the information processing device 10. The information processing device 10 executes this processing at any timing. Note that processing common to the prediction processing and the product yield calculation processing is assigned the same reference numerals and description thereof will be omitted.

[0101] In S401, the optimization unit 139 determines the optimal product yield using the product yield or an index value contributing to the product yield, the oil properties of the heavy oil, the first parameter of the third parameter and the reaction rate, and the second parameter of the third parameter and an index value indicating the equilibrium catalytic activity.

[0102] In S405, the optimization unit 139 uses the determined product yield or an index value contributing to the product yield, the oil properties of the heavy oil, the operating parameters, and values ​​indicating the reaction state to determine the operating parameters that will achieve the determined optimal product yield.

[0103] In S406, the output unit 140 outputs the optimum operating parameters thus found, and the process ends.

[0104] As described above, according to the present disclosure, the oil properties of the heavy oil to be cracked, the operating parameters for the heavy oil cracking unit that performs the heavy oil cracking, and values ​​that indicate the reaction state when the heavy oil is cracked in the heavy oil cracking unit using the operating parameters are input as learning data, and a first model that outputs the reaction state of heavy oil cracking in response to the input of the oil properties of the heavy oil and the operating parameters is learned using the learning data, and the learned first model is stored in a memory unit, thereby making it possible to obtain a first model for accurately predicting the reaction state of heavy oil cracking.

[0105] Furthermore, according to the present disclosure, a third parameter, which is an operating parameter for a heavy oil cracking unit currently performing heavy oil cracking, the oil properties of the heavy oil, and a trained first model are obtained, and the oil properties of the heavy oil and the third parameter are input into the first model to predict a value indicating the reaction state, and the value indicating the reaction state is output, thereby making it possible to accurately predict the reaction state of heavy oil cracking.

[0106] Furthermore, according to the present disclosure, a first model is obtained that outputs a value indicating the reaction state when heavy oil is cracked in the heavy oil cracking unit using the operating parameters in response to input of the oil properties of the heavy oil to be cracked and the operating parameters related to the heavy oil cracking unit that performs the heavy oil cracking; the oil properties of the heavy oil and the operating parameters are input; a value indicating the reaction state is determined using the oil properties of the heavy oil, the operating parameters, and the first model; and the product yield or an index value that contributes to the product yield is determined using the oil properties of the heavy oil, the operating parameters, and the value indicating the reaction state, thereby improving the product yield and contributing to improved profits of the refinery.

[0107] Furthermore, according to the present disclosure, the product yield can be improved by determining the operating parameters that realize the optimal product yield using the determined product yield or an index value that contributes to the product yield, the oil properties of the heavy oil, the operating parameters, and a value indicating the reaction state.

[0108] <4.Other> Although the embodiments of the disclosure have been described above, they can be implemented in various other forms and can be implemented with various omissions, substitutions, and modifications. These embodiments and modifications, as well as those with omissions, substitutions, and modifications, are included in the technical scope of the claims and their equivalents.

[0109] For example, each function of the information processing device 10 may be configured in another device. For example, each DB of the storage unit 120 may be constructed as an external database. Furthermore, each function of the information processing device 10 may be configured in another device.

[0110] In the above disclosure, the heavy oil cracking unit 20 employs FCC as an example, but the present disclosure is not limited to this. The configuration of the present disclosure can also be applied to the heavy oil cracking unit 20 employing thermal cracking and hydrocracking. Figures 10 and 11 show configuration examples for the heavy oil cracking unit 20 employing thermal cracking and hydrocracking. Note that the following configuration example is an example for illustrative purposes and does not limit the heavy oil cracking unit 20 to thermal cracking and hydrocracking.

[0111] Fig. 10 is a diagram showing an example of the configuration of a heavy oil cracking unit 20. The example in Fig. 10 shows a case where the heavy oil cracking unit 20 cracks heavy oil by thermal cracking. As shown in Fig. 10, the heavy oil cracking unit 20 includes a distillation column 23, a heating furnace 24, and a reactor 25. The heavy oil cracking unit 20 has a function of controlling the distillation column 23, the heating furnace 24, and the reactor 25 during thermal cracking, and a function of transmitting and receiving predetermined information to and from the information processing device 10 via communication.

[0112] The distillation tower 23 receives heavy oil as a raw material and separates and outputs each component of the heavy oil (light gas, naphtha fraction, middle distillate, heavy components, and coke) according to boiling point by distillation. Among the components output from the distillation tower 23, the heavy oil output from the bottom of the distillation tower 23 is sent to the heating furnace 24.

[0113] The heating furnace 24 heats the heavy oil separated by the distillation column 23 to a temperature required for thermal cracking, and outputs high-temperature heavy oil. The heating furnace 24 passes the output high-temperature heavy oil to the reactor 25.

[0114] The reactor 25 cracks the high-temperature heavy oil heated by the heating furnace 24 and outputs cracked gas and coke. The reactor 25 operates in a batch cycle or semi-batch cycle. After the reaction, the reactor 25 is purged of the heavy oil remaining inside the reactor 25, and decoked to remove the coke adhering to the surface of the reactor 25, and then warmed up in preparation for the next reaction.

[0115] Fig. 11 is a diagram showing an example of the configuration of a heavy oil cracking unit 20. The example in Fig. 11 shows a case where the heavy oil cracking unit 20 cracks heavy oil by hydrocracking. As shown in Fig. 11, the heavy oil cracking unit 20 includes a reactor 26 and a hydrogen separation tank 27. The heavy oil cracking unit 20 has a function of controlling the reactor 26 and the hydrogen separation tank 27 during hydrocracking, and a function of transmitting and receiving predetermined information to and from the information processing device 10 via communication.

[0116] The reactor 26 is a device that brings the feedstock heavy oil into contact with a catalyst and hydrogen to cause a cracking reaction and obtain cracked products (light gas, naphtha fraction, middle distillate, and heavy components, as well as coke and sludge components). Specifically, the reactor 26 receives heavy oil, hydrogen, and a catalyst, and then brings the heavy oil into contact with the hydrogen and catalyst. Next, the reactor 26 produces cracked products by cracking the heavy oil through a cracking reaction caused by contacting the heavy oil with the hydrogen and catalyst. The reactor 26 then outputs cracked oil.

[0117] The hydrogen separation tank 27 separates the cracked products and hydrogen (and the catalyst in the case of slurry bed hydrocracking) produced by the reactor 26. Specifically, the hydrogen separation tank 27 has a high-pressure hydrogen separation tank, a medium-pressure hydrogen separation tank, and a low-pressure hydrogen separation tank (and a catalyst separation tank in the case of slurry bed hydrocracking), and separates the cracked products and hydrogen (and the catalyst in the case of slurry bed hydrocracking) in each tank. The hydrogen separation tank 27 outputs the separated cracked products and hydrogen (and the catalyst in the case of slurry bed hydrocracking).

[0118] <Additional Notes> The matters described in the above embodiments will be supplemented below. (Supplementary Note 1) An information processing device (10) for predicting a heavy oil reaction state, comprising a processor (11), An information processing device that executes the steps of: receiving input (S101) of oil properties of heavy oil to be cracked, operating parameters for a heavy oil cracking unit that performs heavy oil cracking, and values ​​indicating the reaction state when the heavy oil is cracked in the heavy oil cracking unit using the operating parameters as learning data; learning (102) of a first model that outputs the reaction state of the heavy oil cracking in response to input of the oil properties of the heavy oil and the operating parameters; and storing the learned first model in a memory unit (S103).

[0119] (Appendix 2) The information processing device described in (Appendix 1), wherein the operating parameter is a first parameter among the operating parameters relating to the cracking reaction of the heavy oil, the value indicating the reaction state is a parameter relating to the reaction rate of the heavy oil in the heavy oil cracking, and the first model is a model that outputs the parameter relating to the reaction rate in response to inputting the oil properties of the heavy oil and the first parameter.

[0120] (Appendix 3) The operating parameter is a second parameter relating to equilibrium catalytic activity among the operating parameters, the value indicating the reaction state is an index value indicating the equilibrium catalytic activity of the heavy oil in the heavy oil cracking, and the first model is a model that outputs an index value indicating the equilibrium catalytic activity in response to inputting the oil properties of the heavy oil and the second parameter. (Appendix 1)

[0121] (Appendix 4) An information processing device described in any of (Appendix 1) to (Appendix 3), which executes the steps of: acquiring third parameters, which are operating parameters for a heavy oil cracking unit currently performing heavy oil cracking, the oil properties of the heavy oil, and a trained first model (S201); predicting a value indicating the reaction state by inputting the oil properties of the heavy oil and the third parameters into the first model (S202, S203); and outputting the value indicating the reaction state (S204).

[0122] (Appendix 5) An information processing device according to (Appendix 4), which executes the steps of (S212, S213) re-learning the first model using the learning data, the third parameter, and a value indicating the reaction state predicted by the predicting step, and the step of storing the re-learned first model in a memory unit (S214).

[0123] (Supplementary Note 6) An information processing device (10) having a processor (11), the information processing device executing the steps of: acquiring a first model (S201) that outputs a value indicating a reaction state when the heavy oil is cracked in the heavy oil cracking device according to the operating parameters in response to input of oil properties of heavy oil to be cracked and operating parameters related to a heavy oil cracking unit that performs the heavy oil cracking; accepting input of the oil properties of the heavy oil and the operating parameters (S202, S203); determining a value indicating the reaction state using the oil properties of the heavy oil, the operating parameters, and the first model (S203); determining the product yield or an index value contributing to the product yield using the oil properties of the heavy oil, the operating parameters, and the value indicating the reaction state (S304); and outputting the determined product yield or the index value contributing to the product yield (S305).

[0124] (Appendix 7) The operating parameter is a first parameter among the operating parameters relating to the cracking reaction of the heavy oil, the value indicating the reaction state is a parameter relating to the reaction rate of the heavy oil in the heavy oil cracking, the first model is a model that outputs a parameter relating to the reaction rate in response to inputting the oil properties of the heavy oil and the first parameter, and in the step of determining the product yield or an index value contributing to the product yield, the oil properties of the heavy oil, the first parameter, and the determined parameter relating to the reaction rate of heavy oil cracking are used to determine the product yield or an index value contributing to the product yield. (Appendix 6) An information processing device as described in

[0125] (Appendix 8) The operating parameter is a second parameter related to equilibrium catalytic activity among the operating parameters, the value indicating the reaction state is an index value indicating the equilibrium catalytic activity of the heavy oil in the heavy oil cracking, the first model is a model that outputs an index value indicating the equilibrium catalytic activity in response to inputting the oil properties of the heavy oil and the second parameter, and in the step of determining the index value contributing to the product yield, the index value contributing to the product yield is determined using the oil properties of the heavy oil, the second parameter, and the determined index value indicating the equilibrium catalytic activity. (Appendix 6)

[0126] (Appendix 9) The information processing device according to (Appendix 8) executes the steps of: receiving input of a first parameter related to the cracking reaction of heavy oil among the operating parameters (S201); and, in response to input of the oil properties of the heavy oil and the first parameter, outputting a parameter related to the reaction rate of the heavy oil cracking; and, using the oil properties of the heavy oil and the first parameter, determining a parameter related to the reaction rate (S202). In the step of obtaining, the oil properties of the heavy oil, the first model, and the second model are obtained; and in the step of determining the product yield or an index value contributing to the product yield, the product yield or an index value contributing to the product yield is determined using the oil properties of the heavy oil, the first parameter, the determined parameter related to the reaction rate of heavy oil cracking, the second parameter, and the determined index value indicating the equilibrium catalytic activity.

[0127] (Appendix 10) In the acquiring step, a reaction model is provided that outputs the product yield or an index value contributing to the product yield by inputting the oil properties of the heavy oil, the first parameter, the obtained parameter related to the reaction rate of the heavy oil cracking, the second parameter, and the obtained index value indicating the equilibrium catalytic activity; and in the step of acquiring the oil properties of the heavy oil, the first model, and the second model and determining the product yield or the index value contributing to the product yield, the information processing device described in (Appendix 9) determines the product yield or the index value contributing to the product yield using the oil properties of the heavy oil, the first parameter, the obtained parameter related to the reaction rate of the heavy oil cracking, the second parameter, the obtained index value indicating the equilibrium catalytic activity, and the reaction model.

[0128] (Appendix 11) An information processing device according to any one of (Appendix 7) to (Appendix 10), which executes a step (S405) of determining operating parameters that realize an optimal product yield using the determined product yield or an index value that contributes to the product yield, the oil properties of the heavy oil, the operating parameters, and a value indicating the reaction state.

[0129] (Appendix 12) The information processing device according to any one of (Appendix 1) to (Appendix 11), wherein the first model is a linear regression model or a neural network model, and in the learning step, the first model is learned so as to represent a correlation between the oil properties of the heavy oil, the operating parameters, and the reaction state.

[0130] (Supplementary Note 13) A method in which a computer (e.g., an information processing device 10) having a processor (11) executes the steps of: receiving input (S101) of oil properties of heavy oil to be cracked, operating parameters for a heavy oil cracking unit that performs heavy oil cracking, and values ​​indicating a reaction state when the heavy oil is cracked in the heavy oil cracking unit using the operating parameters as learning data; learning a first model (S102) that outputs a reaction state of the heavy oil cracking in response to input of the oil properties of the heavy oil and the operating parameters; and storing the learned first model in a memory unit (S103).

[0131] (Appendix 14) A program to be executed by a computer (e.g., an information processing device 10) having a processor (11), the program causing the processor to execute the following steps: a step (S101) of accepting input, as learning data, of oil properties of heavy oil to be subjected to heavy oil cracking, operating parameters for a heavy oil cracking unit that performs heavy oil cracking, and values ​​indicating a reaction state when the heavy oil is cracked in the heavy oil cracking unit using the operating parameters; a step (S102) of learning, using the learning data, a first model that outputs a reaction state of the heavy oil cracking in response to input of the oil properties of the heavy oil and the operating parameters; and a step (S103) of storing the learned first model in a memory unit. [Explanation of symbols]

[0132] 1: Information processing system 10: Information processing device 11: Processor 12: Memory 13: Storage 14: Communication IF 15: Input / output interface 20: Heavy oil cracking equipment 21: Reactor 22: Regenerator 23: Distillation column 24:Heating furnace 25: Reactor 26: Reactor 27: Hydrogen separation tank 30: User terminal 40: Network 110: Communications Department 120: Storage section 121: Learning data DB 122: Model DB 130: Control unit 131: Reception control section 132: Transmission control section 133: Input section 134: Learning Department 135: Acquisition section 136: Prediction section 137: Decision section 138: Calculation section 139: Optimization section 140: Output section

Claims

[Claim 1] An information processing device including a processor, A step of receiving an input of an oil property of a target heavy oil for which a value indicating a reaction state is predicted by a first model, and an operating parameter for which a value indicating a reaction state is predicted by the first model; determining a value indicating the reaction state in real time using the oil properties of the heavy oil, the operating parameters, and the first model; A step of calculating a product yield or an index value contributing to the product yield using the oil properties of the heavy oil, the operating parameters, and the value indicating the reaction state; outputting the obtained product yield or an index value contributing to the product yield; Run The first model is learned using the oil properties of the heavy oil to be cracked, the operating parameters for the heavy oil cracking unit that performs the heavy oil cracking, and the values ​​indicating the reaction state when the heavy oil is cracked in the heavy oil cracking unit by the operating parameters as learning data, and is a learning model that can output in real time a value indicating the reaction state when the heavy oil is cracked in the heavy oil cracking unit by the operating parameters in response to inputting the oil properties of the heavy oil to be cracked and the operating parameters for the heavy oil cracking unit that performs the heavy oil cracking. Information processing device.

Images