System for managing optimal operating point of oil refining process and predicting quality of oil product using artificial intelligence model and method thereof

Abstract

This method for predicting the quality of an oil product produced by a hydrodesulfurization process comprises the steps of: collecting measurement information measured by a plurality of sensors provided in hydrodesulfurization process equipment; and predicting the quality of the oil product from the collected information by using an artificial intelligence model including a first prediction model that has learned first training data and a second prediction model that has learned second training data.

Description

A system and method for managing the optimal operating point of a refining process and predicting the quality of refined products using an artificial intelligence model

[0001] The present disclosure relates to a system and method for managing the optimal operating point of a refining process and predicting the quality of refined oil products using an artificial intelligence model. Specifically, the present disclosure relates to a system that utilizes an artificial intelligence model to optimize operating variables in a process to achieve optimal quality or economic efficiency and provides guidance on the operating points of the operating variables. Furthermore, the present disclosure predicts quantitative indicators related to the quality of refined oil products produced during the process under conditions consistent with the manipulated guidance.

[0002] Unlike chemical products, essential oil products are mixtures, and their quality cannot be controlled by their individual components. Therefore, quality control for these essential oil products requires periodic analysis and management of their components through distillation assessments.

[0003] However, the product's test cycle is excessively long. Furthermore, the existing operating process relies heavily on the operator's experience and expertise, making it difficult to comprehensively consider various variables and determine optimal operating points for operating variables such as temperature, flow rate, and pressure. Frequent changes to the optimal operating point can occur due to internal and external environmental factors, increasing the need for optimal operating point selection that considers the relationship between energy yields.

[0004] Therefore, in this disclosure, a system for predicting the quality of essential oil products using an artificial intelligence model is proposed.

[0005] The technical challenge that the present disclosure seeks to solve is to provide a system that manages the optimal operating point of a refining process and predicts the quality of refined products using an artificial intelligence model.

[0006] The technical problems of the present disclosure are not limited to the technical problems described above, and technical problems not mentioned can be clearly understood by a person having ordinary skill in the technical field to which the present disclosure pertains from this specification and the attached drawings.

[0007] In order to solve the above-described technical problem, the method according to the embodiments provides a method for managing the optimal operating point of a refining process and predicting the quality of a refined product using an artificial intelligence model.

[0008] According to embodiments, a method for predicting the quality of a refined petroleum product produced by a residual hydrodesulfurization process includes the steps of collecting measurement information measured by a plurality of sensors provided in a hydrodesulfurization process facility; and the step of predicting the quality of the refined petroleum product from the collected information using an artificial intelligence model including a first prediction model that has learned first learning data and a second prediction model that has learned second learning data. The first learning data includes measurement information measured by the plurality of sensors for each quality test performed within a first period and label information indicating the result of each quality test, and the second learning data includes measurement information measured by the plurality of sensors for each quality test performed within a second period and label information indicating the result of each quality test, and a start time of the first period is earlier than a start time of the second period, and an end time of the first period is earlier than or equal to an end time of the second period.

[0009] Furthermore, the step of predicting the quality includes a step of predicting the quality of the refined oil product as an average or weighted average of the results predicted by each of the first prediction model and the second prediction model.

[0010] Furthermore, the second period is characterized by being a period from the present time to 6 months prior.

[0011] Furthermore, each of the measurement information included in the first learning data and the measurement information included in the second learning data further includes first derivative information indicating a ratio of the flow rate of the refined oil product finally produced by the hydrodesulfurization process to the intermediate crude oil introduced into the fractionator in the hydrodesulfurization process, and second derivative information indicating a temperature of a catalyst added to a reactor into which the initial crude oil (feed) in the hydrodesulfurization process is introduced.

[0012] Furthermore, each of the measurement information included in the first learning data and the measurement information included in the second learning data further includes third derivative information indicating the temperature of the layer from which the refined product is extracted in the distillation tower.

[0013] Furthermore, the quality of the essential oil product is derived based on the temperature information when the essential oil product evaporates to a preset range during a standard test or the lowest temperature information at which the essential oil product ignites.

[0014] Furthermore, the quality prediction method further includes a step of receiving a quality test result of a refined oil product produced by the hydrodesulfurization process; and a step of feeding back the artificial intelligence model based on the received quality test result and measurement information measured by the plurality of sensors.

[0015] Furthermore, the quality prediction method further includes a step of confirming optimal operation guidance provided by a system that provides guidance for process operation; and a step of controlling facilities linked to the plurality of sensors based on the optimal operation guidance.

[0016] According to embodiments, a server for predicting the quality of a refined product produced by a residual hydrodesulfurization process comprises at least one processor; and a memory storing instructions for directing the at least one processor to perform at least one step, wherein at least one step comprises: collecting information measured by a plurality of sensors provided in a hydrodesulfurization process facility; and predicting the quality of the refined product from the collected information using an artificial intelligence model including a first prediction model that has learned first learning data and a second prediction model that has learned second learning data. Here, the first learning data includes measurement information measured by the plurality of sensors for each quality test performed within a first period and label information indicating the result of each quality test, and the second learning data includes measurement information measured by the plurality of sensors for each quality test performed within a second period and label information indicating the result of each quality test, and the start time of the first period is earlier than the start time of the second period, and the end time of the first period is earlier than or equal to the end time of the second period.

[0017] According to embodiments, a non-transitory computer-readable recording medium records a computer program that is executed by a device for predicting the quality of a refined petroleum product produced by a residual hydrodesulfurization process. The computer program includes collecting measurement information measured by a plurality of sensors provided in a hydrodesulfurization process facility; and predicting the quality of the refined petroleum product from the collected information using an artificial intelligence model including a first prediction model that has learned first learning data and a second prediction model that has learned second learning data. The first learning data includes measurement information measured by the plurality of sensors for each quality test performed within a first period and label information indicating the result of each quality test, and the second learning data includes measurement information measured by the plurality of sensors for each quality test performed within a second period and label information indicating the result of each quality test, wherein a start time of the first period is earlier than a start time of the second period, and an end time of the first period is earlier than or equal to an end time of the second period.

[0018] The technical solutions of the present disclosure are not limited to the above-described technical solutions, and solutions not mentioned can be clearly understood by a person having ordinary skill in the art to which the present disclosure pertains from this specification and the attached drawings.

[0019] The system according to the embodiments enables the operation of a refining process adaptively to the ever-changing market prices of refined petroleum products.

[0020] The system according to the embodiments provides the effect of increasing accuracy by enabling more efficient learning of recent period patterns by referencing or fusing both a model that learns long-term patterns and a model that learns recent period patterns.

[0021] The system according to the embodiments enables rapid prediction of the quality of refined petroleum products produced in various manipulable variables during the hydrodesulfurization process.

[0022] The system according to the embodiments can help managers easily find the cause visually when the predicted results of an artificial intelligence model change rapidly, and can facilitate the discovery of rapid changes in experimental values ​​or errors in instruments such as sensors.

[0023] The effects according to the present disclosure are not limited to the effects described above, and effects not mentioned can be clearly understood by a person having ordinary skill in the art to which the present disclosure pertains from this specification and the attached drawings.

[0024] Figure 1 is a flow chart showing the overall flow of the process of producing refined oil products.

[0025] Figure 2 is a flow chart showing the overall flow of the residual hydro desulfurization process according to embodiments.

[0026] FIG. 3 is a diagram showing a system for managing the optimal operating point of a refining process and predicting the quality of a refined product using an artificial intelligence model according to embodiments.

[0027] FIG. 4 is a flowchart illustrating a method by which a system according to embodiments predicts the quality of a refined product using artificial intelligence.

[0028] Figure 5 shows quality indices of refined products predicted by the system according to embodiments and quality indices of refined products at actual corresponding points in time.

[0029] Figures 6 and 7 illustrate operating variables that can be manipulated by the system according to embodiments to optimize the hydrodesulfurization process.

[0030] Figure 8 illustrates the operating variables that can be manipulated for optimization of the hydrodesulfurization process by the system according to embodiments to achieve optimal quality or economy.

[0031] FIG. 9 is a flowchart illustrating a method by which a system according to embodiments analyzes and manages an optimal operating point of a refining process.

[0032] Figure 10 shows the prediction results of the operating variables that can be manipulated by the system according to the embodiments to optimize the hydrodesulfurization process.

[0033] Figures 11 to 14 illustrate examples of UIs provided by systems according to embodiments.

[0034] Figure 15 is an example of a configuration diagram of a system or server according to embodiments.

[0035] The present disclosure may be modified in various ways and encompasses numerous embodiments. Specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present disclosure to specific embodiments, but rather to encompass all modifications, equivalents, and alternatives falling within the spirit and technical scope of the present disclosure. Throughout the description of each drawing, similar reference numerals have been used to designate similar components.

[0036] Terms such as "first," "second," "A," and "B" may be used to describe various components, but the components should not be limited by the terms. The terms are used solely to distinguish one component from another. For example, without departing from the scope of the present disclosure, a first component could be referred to as a second component, and similarly, a second component could also be referred to as a first component. The term "and / or" includes any combination of multiple related listed items or any one of multiple related listed items.

[0037] When a component is referred to as being "connected" or "connected" to another component, it should be understood that it may be directly connected or connected to that other component, but that there may be other components in between. Conversely, when a component is referred to as being "directly connected" or "connected" to another component, it should be understood that there are no other components in between.

[0038] The terminology used in this application is only used to describe specific embodiments and is not intended to limit the present disclosure. The singular expression includes the plural expression unless the context clearly indicates otherwise. In this application, it should be understood that the terms "comprise" or "have" indicate the presence of a feature, number, step, operation, component, part, or combination thereof described in the specification, but do not exclude in advance the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts, or combinations thereof.

[0039] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by a person of ordinary skill in the art to which this disclosure pertains. Terms defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and shall not be interpreted in an idealized or overly formal sense unless explicitly defined herein.

[0040] The present disclosure relates to a system and method for monitoring the optimal operating point of a refining process and predicting the quality of refined products using an artificial intelligence model. Specifically, the present disclosure provides a method for optimizing multiple manipulable operating factors (or variables) during a refining process, and relates to a method for predicting the expected quality indices of refined products during the optimization process. Furthermore, the present disclosure can determine a target quality based on the predicted quality indices of refined products and provide optimal operating guidance for achieving the target quality.

[0041] Below, the process of monitoring the optimal operating point of a refining process and predicting the quality of a refined product using an artificial intelligence model according to embodiments is described in detail.

[0042] Figure 1 is a flow chart showing the overall flow of the process of producing refined oil products.

[0043] As illustrated in Figure 1, the refining process involves removing impurities from crude oil stored in large tanks through a desalter process, preheating the crude oil through a heat exchanger process, and then heating it to a high temperature through a furnace process. Subsequently, fractions are separated based on differences in boiling points through an atmospheric distillation column process.

[0044] The residue desulfurization process described in this specification is one of the oil refining processes and can be referred to by various synonyms such as the residual hydro desulfurization process and the heavy oil hydro desulfurization process.

[0045] In Fig. 2, specific examples of the desulfurization process and / or the hydrodesulfurization process according to the embodiments are examined.

[0046] Figure 2 is a flow chart showing the overall flow of the residual hydro desulfurization process according to embodiments.

[0047] The system according to the embodiments of the present disclosure shown in FIG. 2 can optimize the operation guidance of the desulfurization process and predict the quality indicators of the refined oil products to be finally produced.

[0048] Referring to FIG. 2, the hydrodesulfurization process according to the embodiments may include a feed / reactor process (20), a separator process (21), and a fractionation process (22).

[0049] The feed / reactor process (20) heats crude oil (feed) with one or more first heaters (heaters, 200) and introduces the heated crude oil into one or more reactors (reactors, 201) to perform a reaction process. The reactor (201) performs a hydrodesulfurization process to reduce the sulfur content of the crude oil by adding hydrogen to the crude oil and reacting it with a catalyst. When the hydrodesulfurization process is performed by the reactor (201), a high-temperature and high-pressure separation process is performed by applying a preset high temperature and pressure included in the separator process (21).

[0050] The hydrodesulfurization process according to the embodiments may be performed in multiple stages in one or more reactors by varying the temperature ranges. For example, the hydrodesulfurization process may be performed in two to six reactors by varying the temperature ranges.

[0051] The separator process (21) may include a high-temperature and high-pressure separation process through a high-temperature and high-pressure separator (separator, 210), a low-pressure steam generation process (not shown), an additional heating process through a second heater (212), and other separation processes. The separator process (21) performs a pretreatment process to separate refined oil products contained in the crude oil hydrodesulfurized by the feed / reactor process (20).

[0052] The high-temperature and high-pressure separation process is a process for separating crude oil or gas-liquid under high-temperature / high-pressure conditions, and separates hydrodesulfurized crude oil into a gaseous component (Gas), i.e., low-pressure steam (LP steam) discharged from a high-temperature and high-pressure separator (210), and a liquid component (Liquid).

[0053] The gas component, i.e., the low-pressure steam discharged from the high-temperature and high-pressure separator (210), contains not only excess hydrogen but also a large amount of carbon monoxide, carbon dioxide, propane, hydrogen sulfide, etc., and also contains other light components such as methane and ethane. In particular, the hydrogen components among these can be fed into a hydrogen purification facility (not shown) through a low-temperature / high-pressure separator, etc., and can be used in the reactor (201) of the feed / reactor process (20) after generating high-purity hydrogen through a compressor (not shown).

[0054] At this time, the low-pressure steam discharged from the high-temperature, high-pressure separator (210) can be utilized in various ways for heat exchange at the process facility site, and this low-pressure steam can be collected or generated in a low-pressure steam generator (not shown).

[0055] The liquid component can be introduced into a high-temperature / low-pressure separator (not shown), and the raw material discharged as a liquid component again from the high-temperature / low-pressure separator can be heated by a second heater (212) and introduced into a distillation column (fractionator, 220).

[0056] The fractionation process (22) refers to a process of separating the incoming raw materials into various refined oil products using a distillation tower (220). The distillation tower (220) can separate various refined oil products contained in the raw materials heated by the second heater (212). In addition, the distillation tower (220) can separate crude oils extracted as gaseous components in a high-temperature, high-pressure separator (not shown) through a high-temperature, high-pressure separator (210), and can also separate crude oils released as gaseous components from the high-temperature, high-pressure separator (210) and crude oils extracted by various separators. The distillation tower (220) can separate such crude oils to produce petroleum semi-finished products such as LPG, naphtha, kerosene, diesel, gasoline, and residual oil.

[0057] Because refined crude oil is composed of various mixtures, quality control of refined oil products, unlike other chemical products, is difficult. Quality control of refined oil products is achieved through regular distillation assessments, which analyze and manage the product's components. For example, diesel, a major refined oil semi-finished product, is extracted once a day, taking four hours. Residue is extracted every three to four days, and naphtha is extracted every seven days. These extractions are then used to derive experimental values ​​and maintain quality.

[0058] However, the testing cycle for semi-finished products is problematically long. Furthermore, the existing operating process relies heavily on the operator's experience and expertise, making it difficult to comprehensively consider various variables and determine the optimal temperature operating point for the separation process. Internal and external environmental factors can lead to frequent changes in the optimal operating point, increasing the need for selecting an optimal operating point that considers the relationship between energy yields.

[0059] Accordingly, starting from FIG. 3 below, a system according to the present disclosure is proposed that uses artificial intelligence to predict the quality of a semi-finished product and to provide optimal operation guidance for a hydrodesulfurization process.

[0060] FIG. 3 is a diagram showing a system for managing the optimal operating point of a refining process and predicting the quality of a refined product using an artificial intelligence model according to embodiments.

[0061] Referring to FIG. 3, the system (30) according to the embodiments can operate simultaneously or separately with a first system (31) for predicting the quality of a refined product and a second system (32) for managing the optimal operating point of a refined product process, and information generated by the first system (31) and the second system (32) is transmitted to a manager through a user interface (33) or provides information to a terminal device of the manager.

[0062] The first system (31) predicts the quality of the final refined products to be produced based on the measured values ​​of the manipulable variables and sensors included in the hydrodesulfurization process. Furthermore, the first system (31) predicts the quality of the refined products when the variables and sensors are manipulated using the optimal operating guidance generated by the second system (32). Here, the first system (31) predicts the quality of the refined products by predicting the values ​​of quality indicators.

[0063] That is, the refined petroleum product whose quality is predicted here may include at least one of diesel, residual oil, and naphtha, and the quality of the refined petroleum product may be temperature information when the refined petroleum product evaporates at an arbitrary rate during a related standard test, may be information on the lowest temperature at which each refined petroleum product ignites, or may be an indicator derived secondarily based on the temperature information or the lowest temperature information.

[0064] A device or software module in which a first system (31) according to embodiments is constructed may include a collection unit (311) that collects measurement values ​​of a plurality of sensors included in a hydrodesulfurization process, and a quality prediction unit (313) that predicts in real time the quality of refined petroleum product(s) to be produced based on the collected measurement values, and may further include an artificial intelligence unit (312) that includes one or more artificial intelligence models used by the quality prediction unit (313) to predict quality. The artificial intelligence unit (312) may include one or more artificial intelligence models composed of a neural network model or a machine learning model. Specifically, the artificial intelligence unit (312) may use one or more artificial intelligence models capable of learning time-series learning data that takes specific conditions of process equipment as input information and the quality of refined petroleum products produced as a result of the process as label information. One or more artificial intelligence models can predict time-series data, and thus include at least one of the following models: recurrent neural networks (RNNs), long short-term memory neural networks (LSTMs), gated recurrent units (GRUs), one-dimensional convolutional neural networks (1D CNNs), transformers, and deep learning-based Prophet.

[0065] Meanwhile, the system (30) or the first system (31) according to the embodiments may further include an artificial intelligence learning unit (not shown) that trains the artificial intelligence unit (312) based on learning data or provides feedback to the artificial intelligence unit (312). The artificial intelligence learning unit collects and preprocesses learning data to train the artificial intelligence unit (312). In addition, the sensor measurement information of oil refineries confirmed in real time and the quality information of the final produced oil products can be tracked in real time to generate learning data, and the generated learning data can be input to the artificial intelligence unit (312) to train it.

[0066] The quality prediction unit (313) may directly transmit the quality information predicted by the artificial intelligence unit (312) to the user interface (33), but may also statistically process (e.g., average or weighted average) the values ​​predicted by multiple detailed artificial intelligence models included in the artificial intelligence unit (312) to derive a final quality index. As another example, the artificial intelligence unit (312) according to the embodiments may be an ensemble model in which multiple detailed artificial intelligence models are integrated. Hereinafter, detailed models, i.e., prediction models of various embodiments, are introduced.

[0067] The artificial intelligence unit (312) according to the embodiments may include a first prediction model that has learned first learning data. Here, the first learning data may include measurement information measured by the plurality of sensors for each quality test performed within a first period (e.g., a period of up to 7 years from the present time) and label information indicating the results of each quality test. In other words, the first prediction model is a model that has learned measurement information measured by the plurality of sensors included in hydrodesulfurization process facilities for 7 years and the quality test results of refined oil products produced based on the measurement information.

[0068] In addition, the artificial intelligence unit (312) according to the embodiments may further include a second prediction model that has learned the second learning data, and the second learning data may include measurement information measured by the plurality of sensors for each quality test performed within a second period (e.g., a period from the present time to 6 months ago) and label information indicating the result of each quality test. Here, the first period may be longer than the second period. That is, the start time of the first period is earlier than the start time of the second period, and the end time of the first period is earlier than or the same as the end time of the second period (e.g., the first period is past data for 7 years, and the second period is past data for 6 months).

[0069] The artificial intelligence unit (312) according to the embodiments may further include prediction models that learn by dividing various periods in addition to the first prediction model and the second prediction model.

[0070] The artificial intelligence unit (312) according to the embodiments provides the effect of increasing accuracy by enabling more efficient learning of recent period patterns by referencing or fusing both the model that learns long-term patterns and the model that learns recent period patterns, as described above.

[0071] In addition, the first system (31) according to the embodiments enables the quality of refined oil products produced to be quickly predicted in various variables that can be manipulated during the hydrodesulfurization process using the above configuration.

[0072] Meanwhile, the system (30) according to the embodiments can further operate a second system (32) that controls the hydrodesulfurization process or provides control guidance in a direction that optimizes the cost invested in the hydrodesulfurization process and the profit that can be obtained from the products produced, i.e., the economic feasibility.

[0073] The second system (32) according to the embodiments may further include an optimization unit (322) that optimizes the manipulable operating variables in the hydrodesulfurization process, and an operation guidance generation unit (323) that generates operation guidance of the hydrodesulfurization process based on the values ​​of the variables that can be optimized by the optimization unit (322). Meanwhile, the second system (32) according to the embodiments may further include a target quality determination unit (321) that can receive minimum quality satisfaction information of a specific refined product targeted by a manager through a user interface (33) and determine the target quality of the specific refined product based on the information.

[0074] The system (30) according to the embodiments can optimize the operation guidance of the hydrodesulfurization process in real time considering economic feasibility by linking the first system (31) and the second system (32). For example, the system (30) can verify the operating variables of the current hydrodesulfurization process through the optimization unit (322) based on the second system (32) and identify the operating variables that require manipulation. Thereafter, based on the operation guidance generation unit (323), the process operation guidance is provided to the manager through the user interface (33) and the manager can be confirmed as to whether to change the values ​​of the operating variables. At this time, the first system (31) can predict the expected quality information of the refined petroleum products when the operating variables change through the artificial intelligence unit (312) and the quality prediction unit (313), and can also predict the quality of the refined petroleum products being produced based on the values ​​of the current operating variables. Here, if the first system (31) predicts the quality of a specific refined product, the derived predicted quality can be confirmed by the target quality determination unit (321) of the second system, and then the target quality of the refined product can be determined. The second system (32) can optimize the operating variables manipulated in the hydrodesulfurization process for the quality of the targeted refined product through the optimization unit (322), and the optimization direction is generated by the operation guidance generation unit (323) and reported to the user through the user interface (33).

[0075] The system according to the embodiments provides the effect of increasing accuracy by enabling more efficient learning of recent period patterns by referencing or fusing both a model that learns long-term patterns and a model that learns recent period patterns.

[0076] Additionally, the system according to the embodiments enables rapid prediction of the quality of refined oil products produced in various manipulable variables during the hydrodesulfurization process.

[0077] Some or all of the operations described in the description of the invention related to the above-described FIG. 3 may be performed by a control unit within the server according to FIG. 15. FIG. 4 below shows examples of operations performed by a first system (31) according to embodiments, and FIG. 9 shows examples of operations performed by a second system (32) according to embodiments.

[0078] FIG. 4 is a flowchart illustrating a method by which a system according to embodiments predicts the quality of a refined product using artificial intelligence.

[0079] Referring to FIG. 4, some or all of the operations according to the embodiments may be performed by the system (30) or the first system (31) of FIG. 3.

[0080] Referring to FIG. 4, the system according to the embodiments can collect (400) measurement information measured by a plurality of sensors installed in the hydrodesulfurization process equipment. Step 400 can be performed by the collection unit (311) of FIG. 3, and can be performed by the input unit (1510) or control unit (1530) of the server (1500) shown in FIG. 15.

[0081] Thereafter, the system according to the embodiments can predict (401) the quality of the refined product from the collected information using an artificial intelligence model. Step 401 can be predicted by the artificial intelligence unit (312) or the quality prediction unit (313) of FIG. 3, and can be performed by the control unit (1530) of the server (1500) shown in FIG. 15. The quality of the refined product may mean, for example, an output result directly output from the artificial intelligence unit (312), but may also mean a result derived by statistically processing (for example, an average or a weighted average) the results output by detailed artificial intelligence models included in the artificial intelligence unit (312) in the quality prediction unit (313).

[0082] The artificial intelligence model used in step 401 may refer to the artificial intelligence unit (312) described in FIG. 3.

[0083] Thereafter, the system according to the embodiments can receive (402) the quality test results of the refined oil product produced by the hydrodesulfurization process. That is, the system according to the embodiments can check the values ​​of multiple sensors within the equipment within the hydrodesulfurization process in real time.

[0084] And, when the test results of the refined oil product produced according to the values ​​are available, the system according to the embodiments can receive the test result information. Thereafter, the system according to the embodiments can provide feedback (403) to the artificial intelligence model based on the received quality test results and the measurement information measured by the plurality of sensors. The system according to the embodiments can provide feedback to the artificial intelligence unit (312) by structuring the values ​​of the plurality of sensors and the test results (actual quality information) of the refined oil product together into a data structure. The artificial intelligence unit (312) can perform learning based on the data received as feedback.

[0085] Meanwhile, the learning data learned by the artificial intelligence unit (312) according to the embodiments may include the following information. The artificial intelligence unit (312) mentioned in Fig. 3 may include 1) measurement information measured by multiple sensors for each quality test performed within a specific period, and 2) label information indicating the results of each quality test.

[0086] At this time, the measurement information may further include first derivative information (Flow Ratio) indicating the ratio of the flow rate of the refined product finally produced by the hydrodesulfurization process to the intermediate crude oil fed into the fractionator in the hydrodesulfurization process. For example, the measurement information may include DSLFlowRatio information indicating the ratio of the flow rate of the diesel refined in a specific layer of the distillation column (Diesel Sidecut) to the flow rate of the crude oil (feed) fed into the distillation column. This is because if the flow rate of the introduced crude oil differs, it may affect the flow rate and quality of the final refined product, and thus, the artificial intelligence model needs to consider this when generating weights.

[0087] Additionally, the measurement information may further include secondary derivative information indicating the temperature of the catalyst added to the reactor into which the initial crude oil (feed) is introduced during hydrodesulfurization. The catalyst temperature may represent the temperature of the reactor, where the crude oil components are separated (fractional distillation) depending on the temperature. As the temperature continues to rise, the reactor temperature changes, which means that the components being separated may change, leading to changes in the flow rate of refined products extracted from each layer of the distillation column. Therefore, since the components changed in the reactor affect the temperature tag in the distillation column, this also needs to be considered in the AI ​​model to generate weights.

[0088] Thereafter, the system according to the embodiments can confirm the optimal operation guidance provided by the system providing guidance for process operation (performed at 323 in FIG. 3), and further perform an operation of controlling the facilities linked to the plurality of sensors based on the confirmed optimal operation guidance (performed at 322 in FIG. 3).

[0089] Figure 5 shows quality indices of refined products predicted by the system according to embodiments and quality indices of refined products at actual corresponding points in time.

[0090] Specifically, FIG. 5 illustrates graphs comparing the predicted results (Prediction) of the quality index of the refined oil product produced at a specific time by the artificial intelligence unit (312) according to embodiments for each refined oil product with the test result quality index of the actual refined oil product. Graph (500) is a graph comparing the predicted results of the quality index of DSL D95 produced during the period from May 12, 2023 to July 7, 2023 with the test result quality index of the actual DSL D95. Graph (501) is a graph comparing the predicted results of the quality index of DSL Flafh Pt produced during the period from May 12, 2023 to July 7, 2023 with the test result quality index of the actual DSL Flafh Pt. Graph (502) is a graph comparing the predicted quality index of Naph D95 generated during the period from May 14, 2023 to July 9, 2023 with the actual test result quality index of Naph D95. Graph (503) is a graph comparing the predicted quality index of TAR D5 generated during the period from May 12, 2023 to July 7, 2023 with the actual test result quality index of TAR D5.

[0091] Below, the process by which the system according to the embodiments manages the optimal operating point of the refining process and provides operating guidance will be described in detail. The operations illustrated in FIGS. 6 to 10 below can be performed by the second system (32) of FIG. 3.

[0092] Figures 6 and 7 illustrate operating variables that can be manipulated by the system according to embodiments to optimize the hydrodesulfurization process.

[0093] Referring to FIGS. 6 and 7, examples of manipulated operating variables within a hydrodesulfurization facility are shown before the system according to embodiments analyzes the optimal operating point of a refining process and generates optimal operating guidance. That is, FIGS. 6 and 7 show examples of manipulated operating variables or data for multiple sensors that can be collected within a hydrodesulfurization process by the system according to embodiments.

[0094] Referring to FIGS. 6 and 7, the system according to the embodiments can control a first variable (separator inlet temperature) (601) indicating the temperature of crude oil introduced into a high-temperature, high-pressure separator (210) via a reactor (201), or collect measurement information of the first variable (601). Specifically, as shown in FIGS. 6 and 7, the first variable (601) indicates the temperature of an intermediate product in which a hydrodesulfurization reaction is completed by one or more hydrodesulfurization reactors (201) after crude oil (feed) is heated by one or more first heaters (200, 700e1, 700e2) in a feed / reactor process, and then the intermediate product is introduced into the high-temperature, high-pressure separator (210).

[0095] The first variable (601) can be manipulated by the amount of fuel injected into one or more of the first heaters (200, 700e1, 700e2) described above as shown in FIGS. 6 and 7 or by controlling the one or more of the first heaters (200, 700e1, 700e2). Accordingly, the system according to the embodiments can control one or more of the first heaters (200, 700e1, 700e2) described above to optimize the first variable (601).

[0096] Furthermore, referring to FIGS. 6 and 7, the system according to the embodiments can control a second variable (602) indicating the temperature of the crude oil introduced into the distillation tower (220) via the second heater (212, 700g) used in the separation process, or collect measurement information of the second variable (602). Specifically, as shown in FIGS. 6 and 7, the second variable (602) means the temperature after the liquid substance discharged from the high-temperature, high-pressure separator (210) is introduced into the high-temperature, low-pressure separator (not shown), and the intermediate crude oil discharged as the liquid substance of the high-temperature, low-pressure separator (not shown) is heated by one or more second heaters (212, 700g). That is, the second variable (602) indicates the temperature of the intermediate crude oil introduced into the distillation tower (220).

[0097] The second variable (602) can be manipulated by the amount of fuel injected into one or more of the second heaters (212, 700g) described above as shown in FIGS. 6 and 7 or by controlling the one or more of the second heaters (212, 700g). Accordingly, the system according to the embodiments can control one or more of the second heaters (212, 700g) described above to optimize the second variable (602).

[0098] Also, referring to FIG. 6, the system according to the embodiments can control a third variable (603) indicating the flow rate of high-pressure / medium-pressure steam introduced into the distillation column (220), or collect measurement information of the third variable (603). Specifically, as shown in FIG. 6, the third variable (603) means the flow rate of high-pressure / medium-pressure steam introduced into the distillation column (220). The system according to the embodiments can manipulate the third variable (603) by controlling a device that discharges high-pressure / medium-pressure steam (700f) introduced into the distillation column (220) or some components of the distillation column.

[0099] Also, referring to FIG. 6, the system according to the embodiments can control the fourth variable (604) indicating the flow rate of the refined oil product distilled and separated in the distillation tower (220), or collect measurement information of the fourth variable (604) by related equipment.

[0100] The system according to the embodiments will be described below with reference to FIG. 7 to explain how it manipulates and measures the first to fourth variables described above and how it optimizes the hydrodesulfurization process (700).

[0101] In the process of the hydrodesulfurization process (700), the optimized process operation considering economic feasibility should maximize or maximize the result by limiting all costs incurred during the process operation (e.g., fuel supplied to the heater, amount of medium-pressure steam supplied to the distillation tower, etc.) from the total profit that can be obtained from the product.

[0102] Accordingly, the system according to the embodiments can manipulate control variables within the objective function to maximize the predefined objective function, or adjust the manipulateable detailed variables within the hydrodesulfurization process (700).

[0103] The objective function according to the embodiments is calculated by deducting all costs incurred in operating the process from the total sum of gains that can be obtained from the product. For example, the objective function is derived by adding all of 1) the product of the flow rate and price of the final refined products (diesel, residual oil, naphtha, etc.) (700a, 700b, 700c) produced (first profit), 2) the product of the production amount and price of steam (700d) discharged as a gaseous component of the high-temperature, high-pressure separator (210) and the product of the production amount and price (second profit), 3) the cost of fuel used in one or more first heaters (700e1, 700e2) for heating crude oil introduced into the reactor (201) (first cost), 4) the cost of fuel used in the second heater (700g) used in the separation process (second cost), and 5) the product of the amount of high-pressure / intermediate-pressure steam (700f) introduced into the distillation column and the price (third cost).

[0104] The system according to the embodiments can determine some or all of the first profit, the second profit, the first cost, the second cost, and the third cost that can maximize the objective function. Here, the first profit can be calculated by multiplying the production amount by the price (which can be set by the user) for each refined product. The second profit can be calculated by multiplying the production amount by the price of steam (which can be set by the user). In addition, the first cost and the second cost can be calculated by multiplying the amount of fuel used in the heater by the price (which can be set by the user).

[0105] Accordingly, the manager can input the prices (current prices or expected prices) of each refined product, low-pressure steam, and fuel used in the heater, and can recommend other variables of the objective function considering the optimal economic feasibility based on the prices (i.e., so that the objective function is maximized), and can control or adjust the first to fourth variables based on the values ​​of the recommended variables.

[0106] For example, the manager can set the production ratio or production policy of each refined product (e.g., a policy to maximize the economic efficiency of diesel), and when the price (current price or expected price) of the fuel used in the low-pressure steam and heater is input, some variables can be fixed according to the policy, other variables of the objective function can be recommended considering the optimal economic efficiency (i.e., so that the objective function is maximized), and the first to fourth variables can be controlled or adjusted based on the values ​​of the recommended variables.

[0107] With this configuration, the system according to the embodiments can operate the refining process adaptively to the ever-changing market prices of refined oil products.

[0108] In the following Figure 8, a specific method for optimizing the manipulable operating variables in the hydrodesulfurization process is presented.

[0109] Figure 8 illustrates the operating variables that can be manipulated for optimization of the hydrodesulfurization process by the system according to embodiments to achieve optimal quality or economy.

[0110] Some or all of the operations shown in FIG. 8 may be performed by a system according to embodiments, for example, the first system (31) or the second system (32) of FIG. 3. Some or all of the operations shown in FIG. 8 may be performed by a server (1500) or a control unit (1530) according to embodiments of FIG. 15.

[0111] Referring to FIG. 8, the system according to the embodiments can output (800) the values ​​of the second, third, and fourth variables that maximize the objective function by applying an optimization algorithm. The system according to the embodiments can confirm the target quality of the desired refined oil product and confirm the number of cases of the second to fourth variables that maximize the objective function while achieving the target quality of the refined oil product.

[0112] Thereafter, the system according to the embodiments can calculate (801) the values ​​of the optimal separator inlet temperature (HHPS Inlet temperature) (i.e., the first variable) from which the value of the second variable can be derived, after fixing the value of the second variable. Thereafter, the economic effect is examined (801) by estimating the value of the objective function based on the calculated first to fourth variables.

[0113] Thereafter, the system according to the embodiments monitors the change in heat quantity (802) of the heat exchanger heading to the first heater (Rx. Feed HTR) (200 in FIG. 6, 700e1 and 700e2 in FIG. 7). The system according to the embodiments selects the most appropriate heat exchange method based on the result of observing the change in heat quantity (802) of the heat exchanger.

[0114] For example, the system according to the embodiments may close the feed bypass (Bypass, heat exchanger shown in FIG. 7) to lower the separator inlet temperature (i.e., the first variable), so that the heat of the feed hydrodesulfurized by the reactor (201) can be taken by the initially supplied cold feed. In this case, since the initially supplied cold feed receives heat (804), the amount of fuel input to the first heater (Rx. Feed HTR) (700e1, 700e2) can be reduced, thereby lowering the first cost (804a) <reduction in first cost>. If the separator inlet temperature (i.e., the first variable) is lowered, the amount of low-pressure steam (700d) discharged as a gas component from the high-temperature, high-pressure separator (210) can be reduced (805), thereby lowering the second benefit (805a) <reduction in second benefit>. On the other hand, in this case, since the amount of raw materials discharged as liquid components from the high-temperature and high-pressure separator (210) increases (803), the amount of refined oil products increases, thereby increasing the first profit <increase in first profit>. However, in this case, the amount of raw materials that must be heated in the second heater (700g) increases (803a) <increase in second cost>, and since the amount of refined oil products produced increases, the amount of high-pressure / intermediate-pressure steam (700f) also increases <increase in third cost>. That is, the system according to the embodiments adjusts the first variable to observe the degree of change in the above-described profit or cost, and determines the case where the objective function can be maximized the most.

[0115] As another example, the system can open a feed bypass to maintain the heat of the hydrodesulfurized feed by the reactor (201) in order to increase the separator inlet temperature (i.e., the first variable) so that the heat is not taken away by the cold feed. In this case, since the initially supplied cold feed does not receive heat (804), a larger amount of fuel may be required for the first heater (700e1, 700e2), thereby increasing the first cost (804a) <increase in first cost>. If the separator inlet temperature (i.e., the first variable) is increased, the amount of low-pressure steam (700d) discharged as a gas component from the high-temperature, high-pressure separator (210) may increase (805), thereby increasing the second profit (805a) <increase in second profit>. On the other hand, in this case, the amount of raw materials discharged as liquid components from the high-temperature, high-pressure separator (210) decreases (803), so the amount of refined oil products decreases, lowering the first profit <reduction in first profit>. However, in this case, the amount of raw materials to be heated in the second heater (700g) decreases (803a) <reduction in second cost>, and since the amount of refined oil products produced decreases, the amount of high-pressure / medium-pressure steam (700f) also decreases <reduction in third cost>.

[0116] That is, the system according to the embodiments fixes the second variable (801) as described above, and then adjusts the first variable (802) through the operation of the heat exchanger, etc., observes the degree of change in the above-described profit or cost (803a, 804a, 805a), and determines the case in which the objective function can be maximized.

[0117] The system according to the embodiments can generate optimal process operation guidance based on the number of cases of the determined first to fourth variables and provide it to the terminal device of the manager.

[0118] The system according to the embodiments enables the operation of a refining process adaptively to the ever-changing market prices of refined petroleum products.

[0119] FIG. 9 is a flowchart illustrating a method by which a system according to embodiments analyzes and manages an optimal operating point of a refining process.

[0120] Some or all of the operations shown in FIG. 9 may be performed by a system according to embodiments, for example, the first system (31) or the second system (32) of FIG. 3. Some or all of the operations shown in FIG. 8 may be performed by a server (1500) or a control unit (1530) according to embodiments of FIG. 15.

[0121] Referring to FIG. 9, the system according to the embodiments can obtain target quality information (900) for refined petroleum products. Target quality information may, for example, refer to an economic feasibility policy established by an administrator, or a refined petroleum product production policy that reflects the importance of each refined petroleum product. Target quality information may, for example, refer to a production rate optimized for the constantly changing prices of refined petroleum products, or in some cases, a policy that maximizes the production of a specific refined petroleum product.

[0122] The system according to the following embodiments can determine (901) the values ​​of the manipulable operating variables in the hydrodesulfurization process to maximize the value of the objective function based on the target quality information.

[0123] At this time, step 901 can first determine the values ​​of the second variable, the third variable, and the fourth variable that maximize the value of the objective function as described in Fig. 8 (901a), and when the derived value of the second variable is fixed, the optimal value of the first variable can be determined (901b) by comparing the first cost and the second profit calculated according to the value of the first variable.

[0124] Next, the system according to the embodiments can provide operation guidance of the hydrodesulfurization process to the terminal device of the manager based on the values ​​of the determined operating variables (902). At this time, the system according to the embodiments can further determine detailed operating instructions that can be operated within the process operation based on the determined operating variables (for example, steps 802 to 805 of FIG. 8), wherein the detailed operating instructions can further include opening and closing of a heat exchanger that exchanges heat between a first flow path that delivers crude oil to one or more first heaters and a second flow path that delivers crude oil introduced into a high-temperature, high-pressure separator (804, 804a).

[0125] The system according to the embodiments enables the operation of a refining process adaptively to the ever-changing market prices of refined petroleum products.

[0126] Figure 10 shows the prediction results of the operating variables that can be manipulated by the system according to the embodiments to optimize the hydrodesulfurization process.

[0127] Specifically, FIG. 10 illustrates graphs showing the results of predictions based on the methods shown in FIGS. 6 to 8 for the operational guidance of the first to fourth variables that can be manipulated in the hydrodesulfurization process by the system according to embodiments for each refined oil product. Graph (1000) is a graph showing the results of predictions of the H22110 Inlet temperature. Graph (1001) is a graph showing the results of predictions of the H22111 Inlet temperature. Graph (1002) is a graph showing the results of predictions of the change in the HHPS Inlet - H2310 Inlet temperature. Graph (1003) is a graph showing the results of predictions of the LP STM production volume.

[0128] Below, FIGS. 11 to 14 illustrate examples of a UI (User Interface) that provides process operation guidance and quality information on refined products provided by a system according to embodiments.

[0129] Figures 11 to 14 illustrate examples of UIs provided by systems according to embodiments.

[0130] FIG. 11 is an example of a page that visually represents the results of quality information of refined products predicted by a system according to embodiments and the actual test results (actual values). The first component (1100) is a component that sets a period for viewing the trend of the quality of refined products. When the period is set in the first component (1100), the second component (1002) and the third component (1101) appear. The second component (1102) presents data measured by a plurality of sensors and identification information (tag names and descriptions) of each of the plurality of sensors. When one of the plurality of sensors displayed in the second component (1102) is selected, the system according to embodiments provides the trend of the measured values ​​of the corresponding sensor for the set period. The third component (1101) presents the quality index of the refined product for a specific period predicted by the system (or artificial intelligence model) according to embodiments and the quality index of the actual refined product for a specific period.

[0131] FIG. 12 is a separator guidance page, and is an example of a page that allows you to check the economic results when only the separator inlet temperature (first variable) is controlled, as shown in FIGS. 8 and 9. The first component (1200) represents the amount of change in the selected first variable. The second component (1201) is a component that represents the hourly, daily, monthly, and yearly economic results, that is, a component that represents the economic cost that can be saved daily, monthly, and yearly based on an economic index. The third component (1201) is a component that represents an economic index expressed as a profit that can be calculated based on a target flow rate of a refined product or a cost incurred in refining the target flow rate of the refined product. The fourth component (1203) is a component that represents the changes in the first to third costs based on changes in the operating variables. The fifth component (1204) is a component that represents the fuel usage and steam generation of a plurality of heaters, that is, a component that provides information about costs. The sixth component (1205) is a component that provides operational guidance including one or more suggestions for achieving economic savings. The seventh component (1206) is a component that can manually control the separator inlet temperature (the first variable). The eighth component (1207) is a component that can control the output temperatures of multiple heaters (including setting whether they are reference or fixed). Finally, the ninth component (1208) is a component that inputs information on the prices of fuels used in the heaters, the price of steam, and the prices of each of the refined products, including diesel, naphtha, and T-AR.

[0132] Fig. 13 is an example of a page that allows checking the economic results when the first to fourth variables are controlled, considering the entire hydrodesulfurization process that is not limited to the separator, unlike Fig. 12. The first component (1300) represents the amount of change in the selected first variable. The second component (1301) is a component that represents the hourly, daily, monthly, and yearly economic results, that is, a component that represents the economic cost that can be saved daily, monthly, and yearly based on an economic index. The third component (1302) is a component that represents an economic index expressed as a profit that can be calculated based on a target flow rate of a refined product or a cost incurred in refining the target flow rate of the refined product. The fourth component (1303) is a component that represents the changes in the first to third costs based on changes in the operating variables. The fifth component (1304) is a component that represents the fuel consumption and steam generation of multiple heaters, that is, a component that provides information about costs. The sixth component (1305) is a component that provides operational guidance including one or more suggestions for achieving economic savings. The seventh component (1306) is a component that can manually control the separator inlet temperature (the first variable). The eighth component (1307) is a component that can control the output temperatures of multiple heaters (including whether to set a reference or fixed temperature). Finally, the ninth component (1308) is a component that inputs information on the prices of fuels used in the heaters, the price of steam, and the prices of each of the refined products, including diesel, naphtha, and T-AR.

[0133] Figure 14 is a page visually representing the functions, operations, and influence of the AI ​​unit according to embodiments. In the first component (1401), when a desired period and target oil product are selected, a graph depicting the quality indicators for the selected oil product over time is presented. When the user clicks or drags within the graph, the second component (1402) visually displays how the influence of the AI ​​unit or a subordinate detailed model changes based on the start and end points.

[0134] In the second component (1402), the left component is a graph that quantitatively shows how sensor measurement information used as input values ​​of the artificial intelligence unit according to embodiments affects the artificial intelligence model, and presents correlations and absolute values. In the second component (1402), the right component includes a chart drawn based on values ​​converted into percentiles from the values ​​in the left component.

[0135] With this configuration, the system according to the embodiments can help the manager to easily visually find the cause when the predicted result of the artificial intelligence model changes rapidly, and can easily discover sudden changes in experimental values ​​or errors in instruments such as sensors.

[0136] Figure 15 is an example of a configuration diagram of a system or server according to embodiments.

[0137] Referring to FIG. 15, the server (1500) includes an input unit (1510), an output unit (1520), a control unit (1530), a storage unit (1540), and a communication unit (1550).

[0138] The input unit (1510) receives commands or information from an administrator. The input unit (1510) may include one or more of a microphone and a key input unit for receiving audio signals.

[0139] The output unit (1520) outputs command processing results or various information to the manager. For example, the output unit (1520) outputs information generated by a system that manages the optimal operating point of a refining process and predicts the quality of refined oil products using an artificial intelligence model. For this purpose, the output unit (1520) may include a display, a speaker, a haptic output unit, and an optical output unit, although not illustrated in the drawing. The display may be provided in the form of a flat panel display, a flexible display, an opaque display, a transparent display, electronic paper (E-paper), or any other form well known in the art to which the present disclosure pertains. A touch pad may be laminated on the display to form a touch screen, and a touch key may be implemented through such a touch screen. In addition to the display and the speaker, the output unit (1520) may also be configured to further include any form of output means well known in the art to which the present disclosure pertains.

[0140] The control unit (1530) connects and controls components within the server (1500). For example, it controls each component so that information generated from a system that manages the optimal operating point of a refining process and predicts the quality of a refined oil product using an artificial intelligence model can be output through the output unit (1520). As another example, when judgment information is input by an administrator, the control unit (1530) generates a response signal including judgment information. The control unit (1530) may be configured to include a central processing unit (CPU), a microprocessor unit (MPU), a microcontroller unit (MCU), a graphic processing unit (GPU), a neural processing unit (NPU), or any other form of processor well known in the art of the present disclosure.

[0141] The storage unit (1540) stores data, programs, applications, etc. required for the server (1500) to operate. The storage unit (1540) may include non-volatile memory, volatile memory, a hard disk, an optical disk, a magneto-optical disk, or any type of computer-readable recording medium well known in the art to which the present disclosure pertains.

[0142] The communication unit (1550) communicates with a system or other system that manages the optimal operating point of the refining process and predicts the quality of the refined product using an artificial intelligence model through a wired or wireless network.

[0143] The embodiments of the present disclosure disclosed in this specification and drawings are intended only to provide specific examples to facilitate easy explanation of the technical content of the present disclosure and to aid understanding of the present disclosure, and are not intended to limit the scope of the present disclosure. It will be apparent to those skilled in the art to which the present disclosure pertains that other modifications based on the technical concepts of the present disclosure are possible in addition to the embodiments disclosed herein.

[0144] Although the above has been described with reference to preferred embodiments of the present disclosure, it will be understood by those skilled in the art that various modifications and changes can be made to the present disclosure without departing from the spirit and scope of the present disclosure as set forth in the claims below.

[0145] The system and method for managing the optimal operating point of a refining process and predicting the quality of refined products using the artificial intelligence model described above can be applied to the refining industry.

Claims

1. A method for predicting the quality of refined oil products produced by the residual hydrodesulfurization process, A step of collecting measurement information measured by multiple sensors installed in a hydrodesulfurization process facility; and A step of predicting the quality of the refined oil product from the collected information using an artificial intelligence model including a first prediction model that has learned the first learning data and a second prediction model that has learned the second learning data, The above first learning data includes measurement information measured by the plurality of sensors for each quality test performed within the first period and label information indicating the result of each quality test. The above second learning data includes measurement information measured by the plurality of sensors for each quality test performed within the second period and label information indicating the result of each quality test. The start time of the first period is earlier than the start time of the second period, and the end time of the first period is earlier than or equal to the end time of the second period. Quality prediction methods.

2. In paragraph 1, The step of predicting the above quality is: A step of predicting the quality of the refined oil product by using an average or a weighted average of the results predicted by each of the first prediction model and the second prediction model, Quality prediction methods.

3. In paragraph 1, The above second period is the period from the present time to 6 months ago. Quality prediction methods.

4. In paragraph 1, Each of the measurement information included in the first learning data and the measurement information included in the second learning data, First derivative information indicating the ratio of the flow rate of the refined oil product finally produced by the hydrodesulfurization process to the intermediate crude oil introduced into the fractionator in the hydrodesulfurization process, and Further comprising second derivative information indicating the temperature of the catalyst added to the reactor into which the first crude oil (feed) is introduced in the above hydrodesulfurization process. Quality prediction methods.

5. In paragraph 4, Each of the measurement information included in the first learning data and the measurement information included in the second learning data, Further comprising third derivative information indicating the temperature of the layer from which the refined product is extracted in the distillation tower. Quality prediction methods.

6. In paragraph 1, The quality of the above essential oil product is derived based on the temperature information when the essential oil product evaporates to a preset range during a standard test or the minimum temperature information at which the essential oil product ignites. Quality prediction methods.

7. In paragraph 1, The above quality prediction method is, A step of receiving the quality test results of the refined oil product produced by the above hydrodesulfurization process; and Further comprising a step of feeding back the artificial intelligence model based on the received quality test results and measurement information measured by the plurality of sensors; Quality prediction methods.

8. In paragraph 1, The above quality prediction method is, A step of confirming the optimal operation guidance provided by the system providing guidance for process operation; and Further comprising a step of controlling the facilities linked to the plurality of sensors based on the above optimal driving guidance; Quality prediction methods.

9. As a server that predicts the quality of refined oil products produced by the residual hydrodesulfurization process, at least one processor; and A memory storing instructions that instruct at least one processor to perform at least one step, At least one of the above steps: A step of collecting information measured by multiple sensors installed in a hydrodesulfurization process facility; and A step of predicting the quality of the refined oil product from the collected information using an artificial intelligence model including a first prediction model that has learned the first learning data and a second prediction model that has learned the second learning data; The above first learning data includes measurement information measured by the plurality of sensors for each quality test performed within the first period and label information indicating the result of each quality test. The above second learning data includes measurement information measured by the plurality of sensors for each quality test performed within the second period and label information indicating the result of each quality test. The start time of the first period is earlier than the start time of the second period, and the end time of the first period is earlier than or equal to the end time of the second period. Server.

10. A non-transitory computer-readable recording medium having recorded thereon a computer program executed by a device for predicting the quality of a refined petroleum product produced by a residual hydrodesulfurization process, The above computer program: Collecting measurement information measured by multiple sensors installed in the hydrodesulfurization process equipment; and Predicting the quality of the refined oil product from the collected information using an artificial intelligence model including a first prediction model that has learned the first learning data and a second prediction model that has learned the second learning data, The above first learning data includes measurement information measured by the plurality of sensors for each quality test performed within the first period and label information indicating the result of each quality test. The above second learning data includes measurement information measured by the plurality of sensors for each quality test performed within the second period and label information indicating the result of each quality test. The start time of the first period is earlier than the start time of the second period, and the end time of the first period is earlier than or equal to the end time of the second period. Non-transitory computer-readable recording medium.

Images