Steady-state real-time optimization method, device, storage medium and electronic device

Abstract

The present application relates to the field of optimization control technology, and specifically to a steady-state real-time optimization method, device, storage medium and electronic device, including: obtaining current operating condition data; selecting a steady-state real-time optimization model that matches the current operating condition data from a preset model classification library; optimizing the current operating condition data according to the matched steady-state real-time optimization model. The present application pre-establishes a model classification library, and selects a converged model that is closest to the current operating condition data, so as to reduce the optimization calculation time and improve the convergence rate, while solving the problem that the existing model has high requirements for initial values.

Description

Technical Field

[0001] The present application relates to the field of optimization control technology, and in particular to a steady-state real-time optimization method, device, storage medium and electronic device. Background Art

[0002] Real-time optimization technology uses mathematical and mechanism models, combined with the market prices of raw materials, products and utilities, to simulate and optimize the operating parameters of the plant in real time based on the current operating status and plant constraints of the refining and chemical plants, and transmits the optimization results to the advanced control system to achieve the purpose of improving the efficiency of the plant or saving energy. According to the different operating states, real-time optimization systems can be divided into two types: steady-state and dynamic. Compared with dynamic real-time optimization, steady-state real-time optimization technology mainly uses mechanism models and is suitable for solving large-scale nonlinear optimization problems. In addition, steady-state detection is the most important difference between steady-state and dynamic real-time optimization. The steady-state real-time optimization system needs to perform steady-state detection on the device before and after operation to ensure that the process system is in a relatively stable operating state.

[0003] The steady-state real-time optimization method currently used in the market uses the simultaneous equation method (Equation Oriented, EO) to solve the model calculation. This algorithm has high requirements on the accuracy of the model's initial values. Unreasonable initial values ​​often cause the model to not converge. Summary of the invention

[0004] In response to the above problems, the present application provides a steady-state real-time optimization method, device, storage medium and electronic device, which solve the technical problem in the related art that the model does not converge due to unreasonable model initial values ​​in solving the simultaneous equation method.

[0005] In a first aspect, the present application provides a steady-state real-time optimization method, the method comprising:

[0006] Get current working condition data;

[0007] Selecting a steady-state real-time optimization model that matches the current operating condition data from a preset model classification library;

[0008] The current operating condition data is optimized and calculated according to the matched steady-state real-time optimization model.

[0009] In some embodiments, the process of establishing the preset model classification library includes:

[0010] Collect historical operating data;

[0011] According to the historical operating condition data, a steady-state real-time optimization model is trained to obtain multiple steady-state real-time optimization models that have completed convergence under different operating condition data;

[0012] The preset model classification library is established according to the multiple steady-state real-time optimization models.

[0013] In some embodiments, after collecting the historical operating condition data, the method further includes:

[0014] The historical operating condition data is preprocessed.

[0015] In some embodiments, establishing the preset model classification library according to the multiple steady-state real-time optimization models includes:

[0016] The preset model classification library is established according to the multiple steady-state real-time optimization models through a clustering algorithm.

[0017] In some embodiments, the steady-state real-time optimization model is trained according to the historical operating condition data to obtain multiple steady-state real-time optimization models that have completed convergence under different operating condition data, including:

[0018] Classifying the historical operating condition data to obtain historical operating condition data of different operating conditions;

[0019] The historical operating data of the different operating conditions are used as input to train the machine learning model, and multiple steady-state real-time optimization models that converge under different operating conditions are obtained.

[0020] In some embodiments, the step of selecting a steady-state real-time optimization model that matches the current operating condition data from a preset model classification library includes:

[0021] A steady-state real-time optimization model matching the current operating condition data is selected from a preset model classification library through a classification proximity algorithm.

[0022] In some embodiments, after obtaining the current operating condition data, the method further includes:

[0023] The current operating condition data is preprocessed.

[0024] In a second aspect, a steady-state real-time optimization device is provided, the device comprising:

[0025] An acquisition unit, used to acquire current working condition data;

[0026] A selection unit, used to select a steady-state real-time optimization model that matches the current operating condition data from a preset model classification library;

[0027] A calculation unit is used to optimize the current operating condition data according to the matched steady-state real-time optimization model.

[0028] In a third aspect, a storage medium stores a computer program that can be executed by one or more processors and can be used to implement the steady-state real-time optimization method as described in the first aspect.

[0029] In a fourth aspect, an electronic device comprises a memory and a processor, wherein a computer program is stored in the memory, and the memory and the processor are communicatively connected to each other, and when the computer program is executed by the processor, the steady-state real-time optimization method as described in the first aspect is executed.

[0030] The present application provides a steady-state real-time optimization method, device, storage medium and electronic device, including: obtaining current operating condition data; selecting a steady-state real-time optimization model that matches the current operating condition data from a preset model classification library; optimizing the current operating condition data according to the matched steady-state real-time optimization model. The present application pre-establishes a model classification library, from which a converged model that is closest to the current operating condition data is selected, so as to reduce the optimization calculation time and improve the convergence rate, while solving the problem that the existing model has high requirements for initial values. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0032] Figure 1 A flow chart of a steady-state real-time optimization method provided in an embodiment of the present application;

[0033] Figure 2 A schematic diagram of the model selection process provided in the embodiment of the present application;

[0034] Figure 3 A comparison diagram of the convergence before and after using the steady-state real-time optimization method of the present application provided in the embodiment of the present application;

[0035] Figure 4 A schematic diagram of the structure of a steady-state real-time optimization device provided in an embodiment of the present application;

[0036] Figure 5 A connection block diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0037] The following will describe the implementation methods of the present application in detail with reference to the accompanying drawings and embodiments, so that the implementation process of how the present application applies technical means to solve technical problems and achieve corresponding technical effects can be fully understood and implemented accordingly. The embodiments of the present application and the various features in the embodiments can be combined with each other without conflict, and the technical solutions formed are all within the scope of protection of the present application.

[0038] As can be seen from the background technology, the steady-state real-time optimization method currently used in the market uses the simultaneous equation method (Equation Oriented, EO) to solve the model calculation. This algorithm has high requirements on the accuracy of the initial value of the model. Unreasonable initial values ​​often cause the model to not converge.

[0039] In view of this, the present application provides a steady-state real-time optimization method, device, storage medium and electronic device, which solve the technical problem in the related art that the model does not converge due to unreasonable model initial values ​​in solving the simultaneous equation method.

[0040] Embodiment 1

[0041] Figure 1 A flow chart of a steady-state real-time optimization method provided in an embodiment of the present application is shown as follows: Figure 1 As shown, the method includes:

[0042] S101, obtaining current working condition data.

[0043] It should be noted that the current operating data include: cracking furnace operating data (including the start and stop status of the cracking furnace: on or off. Feed status: LPG, circulating ethane propane, light naphtha, naphtha, cracking tail oil, group cracking), raw material analysis data (laboratory analysis data and online analysis data, see Table 1 for specific information), cracking furnace operation data (cracking furnace feed amount, dilution steam flow, total average COT (Coil Outlet Temperature), cracking furnace outlet pressure COP (Coil Outlet Pressure)), cracking gas online analysis data (molar percentage of H2, CH4, C2H4, C2H6, C3H6 and C3H8), TMT measurement data (measurement date and measurement temperature).

[0044]

[0045] Table 1

[0046] In some embodiments, after obtaining the current operating condition data, the method further includes:

[0047] The current operating condition data is preprocessed.

[0048] It should be noted that the preprocessing includes sorting and cleaning the acquired current operating condition data, deleting zero-value data, duplicate data and abnormal data in the data, and then standardizing the data to ensure the accuracy of subsequent calculations.

[0049] S102: Selecting a steady-state real-time optimization model that matches the current operating condition data from a preset model classification library.

[0050] In some embodiments, the step of selecting a steady-state real-time optimization model that matches the current operating condition data from a preset model classification library includes:

[0051] A steady-state real-time optimization model matching the current operating condition data is selected from a preset model classification library through a classification proximity algorithm.

[0052] It should be noted that by collecting data under different working conditions and models that have converged under different working conditions, a model classification library is established using the clustering k-means clustering algorithm (K-means). When a new working condition appears, based on the current working condition data obtained, the classification k-nearest neighbor algorithm (KNN) is used to automatically select a steady-state real-time optimization model that is the same or similar to the current working condition in the preset model classification library. This ensures that the initial value of the model is close to the current working condition, and even if individual key instruments fail, the normal convergence of the model can be guaranteed.

[0053] Specifically, Figure 2 The figure shows a schematic diagram of the model selection process. First, according to the cracking furnace operating data, cracking furnace operation data, laboratory analysis data and online analysis data in the current operating data, the same or most similar operating conditions are found, and then the most similar steady-state real-time optimization model is found through the classification neighbor algorithm according to the data under the operating condition, and the model is used as the calculation model.

[0054] S103: Optimizing and calculating the current operating condition data according to the matched steady-state real-time optimization model.

[0055] It should be noted that after the steady-state real-time optimization model for calculation is obtained by selection, the current operating condition data is used as input to perform optimization calculation through the model.

[0056] In some embodiments, the process of establishing the preset model classification library includes:

[0057] Collect historical operating data;

[0058] According to the historical operating condition data, a steady-state real-time optimization model is trained to obtain multiple steady-state real-time optimization models that have completed convergence under different operating condition data;

[0059] The preset model classification library is established according to the multiple steady-state real-time optimization models.

[0060] It should be noted that the historical operating data include: cracking furnace operating data (including the start and stop status of the cracking furnace: on or off. Feed status: LPG, circulating ethane propane, light naphtha, naphtha, cracking tail oil, group cracking), raw material analysis data (laboratory analysis data and online analysis data, see Table 1 for specific information), cracking furnace operation data (cracking furnace feed amount, dilution steam flow, total average COT (Coil Outlet Temperature), cracking furnace outlet pressure COP (Coil Outlet Pressure)), cracking gas online analysis data (molar percentage of H2, CH4, C2H4, C2H6, C3H6 and C3H8), TMT measurement data (measurement date and measurement temperature).

[0061] In some embodiments, after collecting the historical operating condition data, the method further includes:

[0062] The historical operating condition data is preprocessed.

[0063] It should be noted that in order to establish a steady-state real-time optimization model, historical operating data is collected and stored in a database. Data preprocessing is to organize and clean the data and delete zero-value data, duplicate data, and abnormal data. 70% of the organized data is used as a training set and 30% of the data is used as a test set. The model is established through machine learning. During the model establishment process, the data needs to be standardized to eliminate the model deviation caused by the order of magnitude and ensure the accuracy of model establishment and prediction.

[0064] It should be further explained that during the production process of the ethylene unit, the production load will be adjusted according to the plan, the cracking furnace will be put into operation, the raw materials of the cracking furnace will be changed, and the cracking state of the cracking furnace (co-cracking and group cracking, etc.) will be changed. The steady-state real-time optimization model must be able to change with the adjustment of the device and achieve the convergence of the model under different working conditions. This has high requirements for the accuracy of the model, the perfection of the adaptive program, and the accuracy of the field measurement instruments. The 5089 hours of cracking furnace operating data collected were statistically analyzed. There are 43 working conditions in these data. The working condition changes of some cracking furnaces are shown in Table 3: In working condition 1, the 5# cracking furnace is turned from stopped to started, and naphtha raw materials are fed; in working condition 2, the 1# cracking furnace is turned from stopped to started, and the cracking tail oil is fed, and at the same time, the 4# cracking furnace is turned from started to stopped. The specific working condition changes of the cracking furnace are shown in Table 2, where 1 represents "on" and "feeding", and 0 represents "stop" and "no feeding".

[0065]

[0066] Table 2

[0067] In some embodiments, establishing the preset model classification library according to the multiple steady-state real-time optimization models includes:

[0068] The preset model classification library is established according to the multiple steady-state real-time optimization models through a clustering algorithm.

[0069] It should be noted that by collecting data under different working conditions and models that have converged under different working conditions, a model classification library is established using the clustering k-means clustering algorithm (K-means).

[0070] In some embodiments, the steady-state real-time optimization model is trained according to the historical operating condition data to obtain multiple steady-state real-time optimization models that have completed convergence under different operating condition data, including:

[0071] Classifying the historical operating condition data to obtain historical operating condition data of different operating conditions;

[0072] The historical operating data of the different operating conditions are used as input to train the machine learning model, and multiple steady-state real-time optimization models that converge under different operating conditions are obtained.

[0073] It should be noted that the models are trained and learned under different working conditions respectively to obtain the corresponding steady-state real-time optimization models under different working conditions, so as to ensure that the models can converge under different working conditions, improve the convergence rate and reduce the initial value requirements.

[0074] like Figure 3 As shown, it is a comparison diagram of the convergence before and after using the steady-state real-time optimization method of the present application. Figure 3 The first figure shows the convergence of the model 12 hours before and after the steady-state real-time optimization method of the present application is used. The model automatically runs once every 2 hours. Figure 3 It can be seen that the use of the steady-state real-time optimization method of the present application can not only reduce the model calculation time, but also improve the convergence rate of the model.

[0075] In summary, the embodiment of the present application provides a steady-state real-time optimization method, including: obtaining current operating condition data; selecting a steady-state real-time optimization model that matches the current operating condition data from a preset model classification library; and optimizing the current operating condition data according to the matched steady-state real-time optimization model. The present application pre-establishes a model classification library, from which a converged model that is closest to the current operating condition data is selected, so as to reduce the optimization calculation time and improve the convergence rate, while solving the problem that the existing model has high requirements for initial values.

[0076] Embodiment 2

[0077] Based on the steady-state real-time optimization method disclosed in the above embodiment of the present invention, Figure 4 Specifically disclosed is a steady-state real-time optimization device that applies the steady-state real-time optimization method.

[0078] like Figure 4 As shown, an embodiment of the present invention discloses a steady-state real-time optimization device, which includes:

[0079] An acquisition unit 401 is used to acquire current operating condition data;

[0080] It should be noted that the current operating data include: cracking furnace operating data (including the start and stop status of the cracking furnace: on or off. Feed status: LPG, circulating ethane propane, light naphtha, naphtha, cracking tail oil, group cracking), raw material analysis data (laboratory analysis data and online analysis data, see Table 1 for specific information), cracking furnace operation data (cracking furnace feed amount, dilution steam flow, total average COT (Coil Outlet Temperature), cracking furnace outlet pressure COP (Coil Outlet Pressure)), cracking gas online analysis data (molar percentage of H2, CH4, C2H4, C2H6, C3H6 and C3H8), TMT measurement data (measurement date and measurement temperature).

[0081] In some embodiments, after obtaining the current operating condition data, the device further includes:

[0082] A preprocessing unit is used to preprocess the current operating condition data.

[0083] It should be noted that the preprocessing includes sorting and cleaning the acquired current operating condition data, deleting zero-value data, duplicate data and abnormal data in the data, and then standardizing the data to ensure the accuracy of subsequent calculations.

[0084] A selection unit 402 is used to select a steady-state real-time optimization model that matches the current operating condition data from a preset model classification library;

[0085] In some embodiments, the selection unit 402 is used to select a steady-state real-time optimization model that matches the current operating condition data from a preset model classification library through a classification proximity algorithm.

[0086] It should be noted that by collecting data under different working conditions and models that have converged under different working conditions, a model classification library is established using the clustering k-means clustering algorithm (K-means). When a new working condition appears, based on the current working condition data obtained, the classification k-nearest neighbor algorithm (KNN) is used to automatically select a steady-state real-time optimization model that is the same or similar to the current working condition in the preset model classification library. This ensures that the initial value of the model is close to the current working condition, and even if individual key instruments fail, the normal convergence of the model can be guaranteed.

[0087] Specifically, Figure 2 The figure shows a schematic diagram of the model selection process. First, according to the cracking furnace operating data, cracking furnace operation data, laboratory analysis data and online analysis data in the current operating data, the same or most similar operating conditions are found, and then the most similar steady-state real-time optimization model is found through the classification neighbor algorithm according to the data under the operating condition, and the model is used as the calculation model.

[0088] The calculation unit 403 is used to perform optimization calculation on the current operating condition data according to the matched steady-state real-time optimization model.

[0089] It should be noted that after the steady-state real-time optimization model for calculation is obtained by selection, the current operating condition data is used as input to perform optimization calculation through the model.

[0090] In some embodiments, the process of establishing the preset model classification library includes:

[0091] A collection unit, used for collecting historical operating condition data;

[0092] A training unit, used for training a steady-state real-time optimization model according to the historical operating condition data, to obtain a plurality of steady-state real-time optimization models that have completed convergence under different operating condition data;

[0093] An establishing unit is used to establish the preset model classification library according to the multiple steady-state real-time optimization models.

[0094] It should be noted that the historical operating data include: cracking furnace operating data (including the start and stop status of the cracking furnace: on or off. Feed status: LPG, circulating ethane propane, light naphtha, naphtha, cracking tail oil, group cracking), raw material analysis data (laboratory analysis data and online analysis data, see Table 1 for specific information), cracking furnace operation data (cracking furnace feed amount, dilution steam flow, total average COT (Coil Outlet Temperature), cracking furnace outlet pressure COP (Coil Outlet Pressure)), cracking gas online analysis data (molar percentage of H2, CH4, C2H4, C2H6, C3H6 and C3H8), TMT measurement data (measurement date and measurement temperature).

[0095] In some embodiments, after collecting the historical operating condition data, the collection unit is further used to:

[0096] The historical operating condition data is preprocessed.

[0097] It should be noted that in order to establish a steady-state real-time optimization model, historical operating data is collected and stored in a database. Data preprocessing is to organize and clean the data and delete zero-value data, duplicate data, and abnormal data. 70% of the organized data is used as a training set and 30% of the data is used as a test set. The model is established through machine learning. During the model establishment process, the data needs to be standardized to eliminate the model deviation caused by the order of magnitude and ensure the accuracy of model establishment and prediction.

[0098] It should be further explained that during the production process of the ethylene unit, the production load will be adjusted according to the plan, the cracking furnace will be put into operation, the raw materials of the cracking furnace will be changed, and the cracking state of the cracking furnace (co-cracking and group cracking, etc.) will be changed. The steady-state real-time optimization model must be able to change with the adjustment of the device and achieve the convergence of the model under different working conditions. This has high requirements for the accuracy of the model, the perfection of the adaptive program, and the accuracy of the field measurement instruments. The 5089 hours of cracking furnace operating data collected were statistically analyzed. There are 43 working conditions in these data. The working condition changes of some cracking furnaces are shown in Table 3: In working condition 1, the 5# cracking furnace is turned from stopped to started, and naphtha raw materials are fed; in working condition 2, the 1# cracking furnace is turned from stopped to started, and the cracking tail oil is fed, and at the same time, the 4# cracking furnace is turned from started to stopped. The specific working condition changes of the cracking furnace are shown in Table 2, where 1 represents "on" and "feeding", and 0 represents "stop" and "no feeding".

[0099] In some embodiments, the establishing unit is used to establish the preset model classification library according to the multiple steady-state real-time optimization models through a clustering algorithm.

[0100] It should be noted that by collecting data under different working conditions and models that have converged under different working conditions, a model classification library is established using the clustering k-means clustering algorithm (K-means).

[0101] In some embodiments, the training unit comprises:

[0102] A classification subunit, used for classifying the historical operating condition data to obtain historical operating condition data of different operating conditions;

[0103] The training subunit is used to train the machine learning model using the historical operating data of the different operating conditions as input, so as to obtain multiple steady-state real-time optimization models that converge under different operating conditions.

[0104] It should be noted that the models are trained and learned under different working conditions respectively to obtain the corresponding steady-state real-time optimization models under different working conditions, so as to ensure that the models can converge under different working conditions, improve the convergence rate and reduce the initial value requirements.

[0105] like Figure 3 As shown, it is a comparison diagram of the convergence before and after using the steady-state real-time optimization method of the present application. Figure 3 The first figure shows the convergence of the model 12 hours before and after the steady-state real-time optimization method of the present application is used. The model automatically runs once every 2 hours. Figure 3 It can be seen that the use of the steady-state real-time optimization method of the present application can not only reduce the model calculation time, but also improve the convergence rate of the model.

[0106] In summary, the embodiment of the present application provides a steady-state real-time optimization device, including: obtaining current operating condition data; selecting a steady-state real-time optimization model that matches the current operating condition data from a preset model classification library; and optimizing the current operating condition data according to the matched steady-state real-time optimization model. The present application pre-establishes a model classification library, from which a converged model that is closest to the current operating condition data is selected, so as to reduce the optimization calculation time and improve the convergence rate, while solving the problem that the existing model has high requirements for initial values.

[0107] Embodiment 3

[0108] This embodiment also provides a computer-readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (for example, an SD or DX memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a disk, an optical disk, a server, an App application store, etc., on which a computer program is stored. When the computer program is executed by a processor, the method steps of the first embodiment can be implemented, and this embodiment will not be repeated here.

[0109] Embodiment 4

[0110] Figure 5 A connection block diagram of an electronic device 500 provided in an embodiment of the present application, such as Figure 5 As shown, the electronic device 500 may include: a processor 501 , a memory 502 , a multimedia component 503 , an input / output (I / O) interface 504 , and a communication component 505 .

[0111] The processor 501 is used to execute all or part of the steps in the steady-state real-time optimization method in Example 1. The memory 502 is used to store various types of data, which may include instructions of any application or method in the electronic device, and data related to the application.

[0112] The processor 501 can be an application specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field programmable gate array (FPGA), a controller, a microcontroller, a microprocessor or other electronic components, and is used to execute the steady-state real-time optimization method in the above-mentioned embodiment 1.

[0113] The memory 502 may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic storage device, a flash memory, a magnetic disk or an optical disk.

[0114] The multimedia component 503 may include a screen and an audio component, the screen may be a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone, and the microphone is used to receive external audio signals. The received audio signal may be further stored in a memory or sent via a communication component. The audio component also includes at least one speaker for outputting audio signals.

[0115] The I / O interface 504 provides an interface between the processor 501 and other interface modules, which may be a keyboard, a mouse, buttons, etc. These buttons may be virtual buttons or physical buttons.

[0116] The communication component 505 is used for wired or wireless communication between the electronic device 500 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G or 4G, or a combination of one or more of them, so the corresponding communication component 505 may include: Wi-Fi module, Bluetooth module, NFC module.

[0117] In summary, the present application provides a steady-state real-time optimization method, device, storage medium and electronic device, the method comprising: obtaining current operating condition data; selecting a steady-state real-time optimization model that matches the current operating condition data from a preset model classification library; optimizing the current operating condition data according to the matched steady-state real-time optimization model. The present application pre-establishes a model classification library, from which a converged model that is closest to the current operating condition data is selected, so as to reduce the optimization calculation time and improve the convergence rate, while solving the problem that the existing model has high requirements for initial values.

[0118] In the several embodiments provided in the present application, it should be understood that the disclosed method can also be implemented in other ways. The method embodiments described above are merely illustrative.

[0119] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0120] Although the implementation methods disclosed in this application are as above, the above contents are only the implementation methods adopted for facilitating the understanding of this application, and are not intended to limit this application. Any technician in the technical field to which this application belongs can make any modifications and changes in the form and details of the implementation without departing from the spirit and scope disclosed in this application, but the scope of patent protection of this application shall still be based on the scope defined in the attached claims.

Claims

1. A steady-state real-time optimization method, which is applied to an ethylene cracking unit for petrochemical industry, characterized in that: The method comprises: Collecting historical operating data, including: cracking furnace operating data, raw material analysis data, cracking furnace operation data, cracking furnace outlet pressure data, cracking gas online analysis data and TMT measurement data; Preprocessing the historical operating condition data includes organizing and cleaning the data, deleting zero-value data, duplicate data, and abnormal data in the data, and using 70% of the organized data as a training set and 30% of the data as a test set; According to the pre-processed historical operating condition data, a steady-state real-time optimization model is trained to obtain a plurality of steady-state real-time optimization models that converge under different operating condition data and are used to perform steady-state real-time optimization of parameters of the ethylene cracking unit under different operating conditions; Establishing a preset model classification library according to the multiple steady-state real-time optimization models; Acquire current operating condition data and preprocess the current operating condition data; the preprocessing of the current operating condition data includes sorting and cleaning the acquired historical operating condition data, deleting zero-value data, duplicate data and abnormal data in the data, and then standardizing the data; According to the preprocessed current operating condition data, the same or most similar operating condition data is found from the historical operating condition data, and a steady-state real-time optimization model that is the same or most similar to the operating condition data is matched in a preset model classification library through a classification proximity algorithm; The variable values ​​of the matched steady-state real-time optimization model are used as initial values ​​for large-scale nonlinear programming to optimize the current operating condition data; the initial value is close to the final optimal solution value of the optimization calculation of the current operating condition data to reduce the initial value requirements, shorten the optimization calculation time and improve the convergence rate.

2. The method according to claim 1, characterized in that The method of training a steady-state real-time optimization model according to the historical operating condition data to obtain multiple steady-state real-time optimization models that converge under different operating condition data includes: Classifying the historical operating condition data to obtain historical operating condition data of different operating conditions; The historical operating data of the different operating conditions are used as input to train the machine learning model, and multiple steady-state real-time optimization models that converge under different operating conditions are obtained.

3. A steady-state real-time optimization device, applied to an ethylene cracking device for petrochemical industry, characterized in that: The device comprises: A preset model classification library is used to collect historical operating data, the historical operating data including: cracking furnace operating data, raw material analysis data, cracking furnace operation data, cracking furnace outlet pressure data, cracking gas online analysis data and TMT measurement data; preprocessing the historical operating data, including sorting and cleaning the data, deleting zero-value data, duplicate data and abnormal data in the data, and using 70% of the sorted data as a training set and 30% of the data as a test set; training a steady-state real-time optimization model based on the preprocessed historical operating data to obtain multiple steady-state real-time optimization models that converge under different operating data and are used for steady-state real-time optimization of parameters under different operating conditions of the ethylene cracking unit; establishing the preset model classification library based on the multiple steady-state real-time optimization models; An acquisition unit is used to acquire current operating condition data and preprocess the current operating condition data; the preprocessing of the current operating condition data includes sorting and cleaning the acquired historical operating condition data, deleting zero-value data, duplicate data and abnormal data in the data, and then standardizing the data; A selection unit is used to find the same or most similar working condition data from the historical working condition data according to the preprocessed current working condition data, and match the same or most similar steady-state real-time optimization model to the working condition data in the preset model classification library through a classification proximity algorithm; A calculation unit is used to use the variable values ​​of the matched steady-state real-time optimization model as initial values ​​for large-scale nonlinear programming solution to optimize the current operating condition data; the initial value is close to the final optimal solution value of the optimization calculation of the current operating condition data to reduce the initial value requirements, reduce the optimization calculation time and improve the convergence rate.

4. A storage medium, characterized in that: The computer program stored in the storage medium can be executed by one or more processors and can be used to implement the steady-state real-time optimization method as described in any one of claims 1 to 2.

5. An electronic device, characterized in that: It comprises a memory and a processor, wherein a computer program is stored in the memory, and the memory and the processor are communicatively connected with each other, and when the computer program is executed by the processor, the steady-state real-time optimization method as described in any one of claims 1 to 2 is executed.

Images