Furnace tube coking sensing and predicting method and system for tubular cracking furnace

Abstract

The invention discloses a furnace tube coking sensing and predicting method and system for a tubular cracking furnace. The method comprises the following steps: constructing digital twin bodies of the cracking furnace; virtual-real spatial data between the cracking furnace and the cracking furnace digital twinborn body are interacted; based on interactive virtual-real spatial data, in combination with multi-source data deep learning, a time-space dynamic change process of a temperature field, a pressure field and a flow field in a cracking furnace tube is reconstructed, and furnace tube coke cleaning or burning optimal opportunity prediction of furnace tube coke statistical inference and trend prediction is realized. The method has important significance for guaranteeing the safety of petrochemical production.

Description

Technical Field

[0001] This invention belongs to the technical fields of safe and stable operation and process safety control of petrochemical production equipment, information engineering and data diagnosis, and specifically relates to a method and system for sensing and predicting coking in the tubes of a tubular pyrolysis furnace. Background Technology

[0002] Ethylene cracking furnaces are a crucial component of ethylene plants, playing a leading role in ethylene and even the entire petrochemical production process. The primary function of the cracking furnace is to process feedstock into cracked gas, which is then piped to subsequent production equipment in the process flow for further processing into ethylene, propylene, and various byproducts. The production capacity and technological level of the cracking furnace determine the production scale, capacity, and product quality of the ethylene plant.

[0003] As is well known, pyrolysis units are fraught with potential safety hazards, high risks, and difficulties in online control. Among these, the most typical are abnormal furnace operating conditions (e.g., burner backfire / extinguishing, flameout, etc.) and abnormal furnace tube operating conditions (e.g., coking, blockage, cracks, damage, deformation, etc.). Whether in the furnace or the furnace tubes, any abnormality or malfunction can, at best, affect production, and at worst, trigger catastrophic accidents. Therefore, establishing methods for detecting safety hazards such as abnormal furnace flames, localized coking in furnace tubes, and corrosion damage in pipelines is necessary and meaningful for the prevention and control of safety risks in ethylene units, especially pyrolysis furnaces.

[0004] The cracking furnace is the core equipment of an ethylene cracking unit and also a major area prone to catastrophic risks. Taking the tubular cracking furnace commonly used by domestic and foreign companies as an example, the ethylene production feedstock is preheated to about 6000℃ in the convection chamber tubes of the cracking furnace before rapidly flowing through the furnace tubes inside the furnace chamber. During this process, the material is rapidly heated to about 8000℃, undergoing chemical reactions to complete the cracking process of breaking down large molecules into smaller molecules. Due to secondary reactions such as polymerization and condensation during the cracking process, hydrocarbons inevitably coke or form carbon deposits that accumulate on the inner wall of the furnace tubes. Furthermore, with the extension of the operating cycle, the depth of cracking, and the increasing weight of the production material, the degree of coking becomes increasingly severe, even clogging the furnace tubes and affecting the safety of the unit. Currently, the typical practice is to burn off or remove coke based on changes in the furnace tube outlet temperature (COT), or to periodically burn off coke online and shut down for cleaning. The above practices have obvious limitations: First, coking of furnace tubes is complex and diverse, and COT is mainly related to coking near the tube opening, making it difficult to reflect the coking situation of the entire furnace tube; second, cleaning coke too early or too late, or burning coke for too long or too short a time, will not only affect production but also damage the furnace tubes or accelerate their aging.

[0005] Coking is a significant factor affecting the safety of ethylene plant production. Accurately detecting coking and rationally scheduling decoking methods and timing has long been a technical challenge hindering safe, stable, long-term, full-capacity, and high-quality production. Considering that coking areas in furnace tubes are poor heat conductors, leading to decreased pyrolysis efficiency, localized heat release, and even burn-through, and that as coking accumulates in the tubes, it can clog pipes, impacting production safety, and in severe cases, easily causing tube blockage or even tube rupture, current methods used by enterprises are mainly based on the pressure difference between the furnace tube inlet and outlet, tube temperature, and tube surface appearance, which have significant limitations. Therefore, this invention establishes a method for detecting coking in the entire furnace tube and assisting in decision-making regarding decoking or decoking timing, using key parameters such as pyrolysis temperature, reaction pressure, and residence time as input points. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention proposes a method and system for sensing and predicting coking in tubular pyrolysis furnace tubes, which is of great significance for ensuring the safety of petrochemical production.

[0007] To achieve the above objectives, the present invention provides the following solution:

[0008] A method for detecting and predicting coking in the tubes of a tubular pyrolysis furnace includes the following steps:

[0009] Constructing a digital twin of a pyrolysis furnace;

[0010] Interacting with each other in virtual and physical space between the pyrolysis furnace and its digital twin;

[0011] Based on interactive virtual-real space data and combined with deep learning of multi-source data, the spatiotemporal dynamic changes of temperature, pressure and flow fields inside the pyrolysis furnace tubes are reconstructed, enabling statistical inference and trend prediction of coking in the furnace tubes to predict the optimal time for coking or burning.

[0012] Preferably, the method for constructing a digital twin of a pyrolysis furnace includes:

[0013] With the pyrolysis device as the core, and focusing on the physical entity PE, virtual entity VE, digital twin data DD, security-oriented service Ss, and inter-entity connection CN, a digital twin conceptual model is constructed, integrating information data and physical data.

[0014] Based on integrated information and physical data, a bottom-up approach is adopted to establish a reference architecture for a digital twin system.

[0015] Based on the aforementioned digital twin system reference architecture, a digital model or information transmission model of the PE is established, bidirectional data communication between the VE and the entity is built, a digital shadow model of the ethylene unit process is formed, and then a complete digital twin model is formed, constructing a digital twin of the pyrolysis process.

[0016] Preferably, the method for interacting with the virtual-real spatial data between the pyrolysis furnace and its digital twin includes:

[0017] Based on 3D virtual simulation of the ethylene cracking system and combined with a digital twin model, the data generation, exchange and flow relationships of each link in the ethylene process are cleared through the digital main line, so as to realize the dynamic reproduction and simulation of the production process and safety risks in the virtual environment, as well as the seamless flow of data and process traceability of each production process.

[0018] Preferably, the method for reconstructing the spatiotemporal dynamic changes of the temperature field, pressure field, and flow field inside the pyrolysis furnace tube includes:

[0019] Construct an expandable set of models S for the temperature, pressure, and flow rate changes in the convection chamber of a pyrolysis furnace and the pyrolysis reaction of materials;

[0020] Based on the expandable model set S, numerical simulation is used to quantitatively analyze the coupling relationship between changes in furnace tube feed characteristics, changes in the internal and external environment of the furnace tube, and changes in material characteristics at the convection chamber outlet.

[0021] A 3D mesh was used to divide the convection chamber of the pyrolysis furnace into meshes D. Based on the expandable model set S, the temperature, pressure, flow velocity, and flow rate data of each node in mesh D were calculated.

[0022] By utilizing the coupling relationship between the changes in the feed characteristics of the furnace tube, the changes in the internal and external environment of the furnace tube, and the changes in the material characteristics at the outlet of the convection chamber, and combining the temperature, pressure, velocity, and flow rate data of each node in the grid, a multi-layer convolutional neural network is trained to obtain a deep learning model.

[0023] The temperature, pressure, flow velocity, and radiation characteristics of the tensor field at any point in the container and the transmission pipeline are described by the field, and the simulation data of the temperature, pressure, flow velocity, and radiation characteristics of the tensor field at the grid nodes are obtained.

[0024] Using spatiotemporal simulation data from digital twins, a tensor field function fitting model based on deep learning, as well as gradient, divergence, and curl function models, are obtained.

[0025] By combining temperature, pressure, flow rate, and radiation data collected on-site from the ethylene plant, the deep learning-based tensor field function fitting model, as well as the gradient, divergence, and curl function models, are updated. Using the updated model and spatiotemporal interpolation method, soft measurement data of the spatiotemporal variation of temperature, pressure, flow rate, and radiation characteristics at any point inside the ethylene cracking furnace tube are obtained.

[0026] Preferred methods for predicting the optimal timing for furnace tube descaling or descaling based on statistical inference and trend forecasting of furnace tube coking include:

[0027] The method of using online measured data to drive digital twins is used to visualize the location, degree, and change process of coking in furnace tubes.

[0028] By using historical data from the actual production processes of multiple groups of pyrolysis furnaces of the same model, the basic characteristics and mutual influence laws of flow, heat transfer, mass transfer and pyrolysis reaction processes inside the pyrolysis furnace tubes are revealed.

[0029] Using a combination of data mining and analysis and machine learning from pyrolysis furnaces, a causal relationship tree with coking and creep as the core was constructed, as well as a multi-factor regression analysis model for coking thickness in furnace tubes and a statistical inference model for coking location and thickness in furnace tubes based on local eddy currents in temperature and pressure fields.

[0030] Using measured operating data and deep learning models of temperature and pressure fields, the internal temperature and pressure fields of the furnace tubes are calibrated and soft-measured to achieve online prediction of the spatiotemporal changes of temperature and pressure fields.

[0031] A predictive data-driven digital twin system is adopted, which combines a multi-factor regression analysis model for coking thickness and a statistical inference model for coking location to predict the coking thickness and location of furnace tubes. Combined with domain expert knowledge, the optimal coking treatment strategy and the best time for treatment are determined, providing auxiliary decision-making for coking treatment.

[0032] The present invention also provides a furnace tube coking sensing and prediction system for a tubular pyrolysis furnace, comprising: a construction module, an interaction module, and a prediction module;

[0033] The building module is used to construct a digital twin of the pyrolysis furnace;

[0034] The interaction module is used to interact with the virtual-real space data between the pyrolysis furnace and its digital twin.

[0035] The prediction module is used to reconstruct the spatiotemporal dynamic changes of the temperature field, pressure field, and flow field inside the pyrolysis furnace tube based on interactive virtual-real spatial data and combined with deep learning of multi-source data, so as to predict the optimal time for furnace tube decoking or burning by statistical inference and trend prediction of coking in the furnace tube.

[0036] Preferably, the building module includes: a data integration unit, a reference architecture building unit, and a digital twin building unit;

[0037] The data integration unit is used to construct a digital twin conceptual model with the pyrolysis device as the core, around the physical entity PE, virtual entity VE, digital twin data DD, security-oriented service Ss, and inter-entity connection CN, and to integrate information data and physical data.

[0038] The reference architecture building unit is used to establish a digital twin system reference architecture based on integrated information data and physical data, using a bottom-up approach.

[0039] The digital twin construction unit is used to establish a digital model or information transmission model of PE based on the digital twin system reference architecture, build bidirectional data communication between VE and the entity, form a digital shadow model of the ethylene unit process, and then form a complete digital twin model to construct a digital twin of the pyrolysis process.

[0040] Preferably, the process of interacting with the virtual-physical spatial data between the pyrolysis furnace and its digital twin includes:

[0041] Based on 3D virtual simulation of the ethylene cracking system and combined with a digital twin model, the data generation, exchange and flow relationships of each link in the ethylene process are cleared through the digital main line, so as to realize the dynamic reproduction and simulation of the production process and safety risks in the virtual environment, as well as the seamless flow of data and process traceability of each production process.

[0042] Preferably, the prediction module includes: an expandable model set S construction unit, a coupling relationship analysis unit, a change calculation unit, a deep learning model construction unit, a grid node simulation data calculation unit, a spatiotemporal vector field plotting unit, and a soft measurement data calculation unit;

[0043] The expandable model set S building unit is used to construct an expandable model set S for the temperature, pressure, and flow rate changes in the convection chamber of the pyrolysis furnace and the material pyrolysis reaction;

[0044] The coupling relationship analysis unit is used to quantitatively analyze the coupling relationship between the changes in the characteristic quantity of furnace tube feed, the changes in the internal and external environment of the furnace tube, and the changes in the characteristic quantity of material at the convection chamber outlet, based on the expandable model set S and through numerical simulation.

[0045] The change calculation unit is used to divide the convection chamber of the pyrolysis furnace into a 3D grid D, and calculate the temperature, pressure, flow rate and flow rate data of each node of the grid D based on the expandable model set S.

[0046] The deep learning model building unit is used to utilize the coupling relationship between the changes in the characteristics of the furnace tube feed, the changes in the internal and external environment of the furnace tube, and the changes in the characteristics of the material at the convection chamber outlet. Combined with the temperature, pressure, velocity, and flow rate data of each node in the grid, a multi-layer convolutional neural network is trained to obtain a deep learning model.

[0047] The grid node simulation data calculation unit is used to obtain the simulation data of the temperature, pressure, flow velocity, and radiation characteristics of the tensor field at any point in the container and the transmission pipeline by using the temperature, pressure, flow velocity, and radiation characteristics of the field at the grid node.

[0048] The spatiotemporal vector field plotting unit is used to obtain a tensor field function fitting model based on deep learning, as well as gradient, divergence and curl function models, using digital twin spatiotemporal simulation data.

[0049] The soft measurement data calculation unit is used to update the deep learning-based tensor field function fitting model and the gradient, divergence and curl function models by combining the temperature, pressure, flow rate and radiation data collected on-site by the ethylene plant; and by combining the updated model and spatiotemporal interpolation method, to obtain the spatiotemporal variation of the temperature, pressure, flow rate and radiation characteristics of any point in the ethylene cracking furnace tube.

[0050] Preferably, the prediction module further includes: a visualization unit, a pattern revealing unit, an inference model building unit, an online prediction unit, and a decision-making unit;

[0051] The visualization unit is used to visualize the location, degree, and change process of coking in the furnace tube by using online measured data to drive the digital twin method.

[0052] The pattern revealing unit is used to reveal the basic characteristics and mutual influence laws of the flow, heat transfer, mass transfer and pyrolysis reaction processes inside the pyrolysis furnace tubes by using historical data of the actual production process of multiple groups of pyrolysis furnaces of the same model.

[0053] The inference model building unit is used to construct a causal relationship tree with coking and creep as the core, as well as a multi-factor regression analysis model of coking thickness of furnace tubes and a statistical inference model of coking location and thickness of furnace tubes based on local eddy currents of temperature field and pressure field, using a combination of data mining analysis and machine learning methods of pyrolysis furnace.

[0054] The online prediction unit is used to calibrate and perform soft measurement of the internal temperature and pressure fields of the furnace tube using measured operating data and deep learning models of temperature and pressure fields, so as to realize online prediction of the spatiotemporal changes of temperature and pressure fields.

[0055] The decision-making unit is used to drive the digital twin system with predictive data, combining a multi-factor regression analysis model for coking thickness and a statistical inference model for coking location to predict the coking thickness and location of the furnace tube; and combining domain expert knowledge to determine the optimal coking treatment strategy and the best time for treatment, providing auxiliary decision-making for coking treatment.

[0056] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0057] This application presents a set of targeted and practical methods for sensing coking in furnace tubes, predicting coking trends, and providing condition-based operation and maintenance support decisions. By rationally constructing a digital twin of the pyrolysis furnace and realizing virtual-real spatial data interaction between the pyrolysis furnace and the twin, and combining deep learning from multi-source data, the spatiotemporal dynamic changes of the temperature, pressure, and flow fields within the pyrolysis furnace tubes are reconstructed. This enables statistical inference, trend prediction, and prediction of the optimal timing for coking in the furnace tubes based on the inference and prediction. The method is original, novel, and advanced, and is of great significance for ensuring the safety of petrochemical production. Attached Figure Description

[0058] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0059] Figure 1 This is a schematic diagram of the digital twin construction and virtual-real space data interaction model according to an embodiment of the present invention;

[0060] Figure 2 This is a schematic diagram of the field reconstruction and spatiotemporal soft measurement process according to an embodiment of the present invention;

[0061] Figure 3 This is a schematic diagram of the furnace tube coking prediction, early warning, and auxiliary decision-making process according to an embodiment of the present invention;

[0062] Figure 4 This is a schematic diagram of a method for sensing and predicting coking in the tubes of a tubular pyrolysis furnace according to an embodiment of the present invention. Detailed Implementation

[0063] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0065] Example 1

[0066] like Figure 4As shown, this invention focuses on the critical issue of coking in furnace tubes that affects production safety. Starting with the construction of a digital twin of the pyrolysis furnace, it achieves auxiliary decision-making for coking perception and predictive maintenance in four steps through virtual-real and data-driven approaches.

[0067] Step 1: Construct a digital twin of the pyrolysis furnace and realize virtual-physical space data interaction. The specific construction method is as follows:

[0068] S1-1. Taking the pyrolysis unit as the core, starting from the pyrolysis unit and its process flow, and around its physical entity PE, virtual entity VE, digital twin data DD, security-oriented service Ss and the connection between entities CN, construct its digital twin conceptual model (PE, VE, Ss, DD, CN), integrate information data and physical data, and provide data support for all elements / the entire process;

[0069] S1-2. A bottom-up approach is adopted to establish a five-layer reference architecture for a digital twin system, comprising the physical domain, measurement and control entity, digital twin, user domain, and cross-domain functional entity. Based on the abstraction and modeling of the ethylene plant, process, and safety risks, a digital model or information transmission model of the physical entity (PE) is established. Two-way data communication between the digital twin entity (VE) and the physical entity is built, forming a digital shadow model of the ethylene plant process, and ultimately a complete digital twin model. The digital shadow model of the ethylene plant process is essentially a model that can run on a digital computer; it is a mapping of the PE entity model in the VE environment, i.e., the model form of the VE. The complete digital twin model is a digital representation of the entity's mechanism model and a mapping of the entity's intrinsic relationships within the VE environment.

[0070] S1-3. Based on the 3D virtual simulation of the ethylene cracking system, combined with a digital twin model, the data generation, exchange, and flow relationships of each stage of the ethylene process are clarified through a digital mainline, forming a digital twin model of the high-temperature cracking process inside the pipe and a virtual environment based on the model. This ensures digital compatibility and enables real-time data interaction between the real and virtual spaces of the cracking process. Specifically, the digital twin construction and the virtual-real space data interaction model are as follows: Figure 1 As shown, the system enables dynamic reproduction and simulation of the production process and safety risks in a virtual environment, as well as seamless data flow and process traceability between various production processes. Specifically, the digital simulation of the ethylene process in the VE environment involves data generated from digital twin models during the dynamic system simulation. The data exchange and flow relationships between different stages in the VE environment are the same as the data exchange process between units in the PE environment, but simpler, only involving data transfer within a digital computer simulation environment.

[0071] Step 2: Reconstruct the spatiotemporal temperature, pressure, and flow fields within the pyrolysis furnace tubes using data-driven methods. The specific construction method is as follows:

[0072] S2-1. Based on historical data of the pyrolysis furnace, especially the measured historical data of different measuring points during the in-service process of furnace tubes with different coking conditions, and integrating spatiotemporal variation soft measurement and multilayer convolutional neural networks, a spatiotemporal evolution model of the temperature field, pressure field, and flow field inside the tube is reconstructed. Using classical models such as unidirectional pressure drop and heat transfer in the convection section of the ethylene pyrolysis furnace and heat transfer of flue gas outside the tube, as well as models of temperature change and pressure transmission inside the tube, pressure drop model of two-phase flow inside the tube, heat transfer model of two-phase flow inside the tube, pressure drop model of flue gas outside the tube, heat transfer rate of the radiation chamber, heat balance equation, and model of material pyrolysis reaction process inside the tube, an expandable model set S is constructed for the temperature, pressure, and flow changes of the convection chamber of the pyrolysis furnace and the material pyrolysis reaction. Among them, the expandable model set is the basic model set formed by the above models, which can be expanded online.

[0073] S2-2. Based on the expandable model set S, numerical simulation is used to quantitatively analyze the coupling relationships between different sources and elements, such as changes in furnace tube feed characteristics (e.g., temperature, pressure, velocity, flow rate), changes in the internal and external environment of the furnace tube, and changes in convection chamber outlet material characteristics (e.g., temperature, pressure, velocity, flow rate). Among these, the quantitative analysis of the coupling relationships between various elements can use multivariate analysis, analysis of variance, and correlation analysis, whichever is selected according to the actual situation.

[0074] S2-3. A 3D mesh is used to divide the convection chamber of the pyrolysis furnace into meshes D. Based on the above-mentioned expandable model set S, the temperature change, pressure vector three-directional components, velocity vector three-directional components, and flow rate change of each node D(x,y,z;t) of mesh D are calculated with time using the mechanism model.

[0075] S2-4. Using a large amount of different feed data, different internal and external environmental data, and outlet material data obtained by simulation calculations from the furnace tube in S2-2, combined with the temperature, pressure, flow rate, and flow rate data of each node D(x,y,z;t) in the grid in S2-3, a multi-layer convolutional neural network is trained to obtain a deep learning model (CNN), denoted as CNNa;

[0076] S2-5. Considering that the characteristic variables of material temperature, pressure, flow rate, and radiation inside the pyrolysis furnace tube change with time, space, and process flow, a field description is used to represent the characteristic attributes of temperature, pressure, flow rate, and radiation at any point within the container and transmission pipeline. Taking the multi-order tensor field, composed of the spatiotemporal changes and attributes inside the convection chamber and transmission pipeline, as the object, and combining 3D simulation data to drive a digital twin, simulation data of the tensor field's characteristic attributes such as temperature, pressure, flow rate, and radiation at grid nodes (data of tensor field changes over time) are obtained. Field reconstruction and spatiotemporal soft measurement processes, such as... Figure 2 As shown.

[0077] S2-6. Using digital twin spatiotemporal extrapolation data, train multidimensional DeepAR and Spacetimeformer to obtain a deep learning-based tensor field function fitting model, as well as its gradient, divergence and curl function models, and plot the spatiotemporal vector field diagrams of temperature, pressure, flow velocity and radiation respectively.

[0078] S2-7. Using historical data such as temperature, pressure, flow rate, and radiation from the cracking furnace data warehouse, train and transfer learn the volume models of S2-4 and S2-6; update the tensor field deep learning model and the gradient, divergence, and curl fitting models by combining temperature, pressure, flow rate, and radiation data collected on-site from the ethylene plant; combine the updated model and spatiotemporal interpolation method to obtain soft measurement data of the spatiotemporal changes of characteristic attributes such as temperature, pressure, flow rate, and radiation at any point inside the ethylene cracking furnace tube.

[0079] Step 3: Inferring coking in the furnace tubes and predicting the optimal time for decoking or burning. The inference and decision support methods are as follows:

[0080] S3-1. Starting with the reconstruction of the temperature and pressure fields inside the tube based on measured data, and combined with the digital twin system, the method of driving the digital twin with online measured data is used to visualize the coking location, coking degree and change process of the furnace tube;

[0081] S3-2. Using historical data from the actual production processes of multiple groups of pyrolysis furnaces of the same model, the deep learning model CNNA is trained, tested, verified, and iteratively optimized to obtain a more accurate neural network model CNNb; by comparing and analyzing the changes in simulation results and measured data, an understanding of the distribution of material velocity, temperature, pressure, and reaction rate within the pyrolysis furnace is obtained, revealing the basic characteristics of flow, heat transfer, mass transfer, and pyrolysis reaction processes within the pyrolysis furnace tubes and their mutual influence laws.

[0082] S3-3. Using a combination of data mining analysis and machine learning from pyrolysis furnaces, a causal relationship tree with coking and creep as the core was constructed, as well as a multi-factor regression analysis model for coking thickness in furnace tubes and a statistical inference model for coking location and thickness in furnace tubes based on local eddy currents in temperature and pressure fields.

[0083] S3-4. Using measured operating data and deep learning models (CNNb) for temperature and pressure fields, calibration and soft measurement of the internal temperature and pressure fields of the furnace tube are performed; using soft measurement data of temperature and pressure changes, online prediction of the spatiotemporal changes of temperature and pressure fields is achieved based on a recurrent neural network (DA-RNN) with a two-stage attention mechanism.

[0084] S3-5. A predictive data-driven digital twin system is adopted, combining a multi-factor regression analysis model for coking thickness and a statistical inference model for coking location. High-fidelity adaptive fault-tolerant index smoothing prediction using characteristic parameters is employed, and multiple prediction models are fused at the decision-level to predict the thickness and location of coking in furnace tubes. Incorporating domain expert knowledge, the optimal coking treatment strategy and the best timing for treatment are determined. The strategy involves selecting online coking, coking burning, or furnace shutdown for cleaning based on the coking situation; the best timing is the most suitable time. This provides auxiliary decision-making support for coking treatment. The specific prediction, early warning, and auxiliary decision-making process is as follows: Figure 3 As shown.

[0085] Step 4: Testing and Verification. Using historical data replay, experiments were conducted to monitor instrument data security and trace the root causes, verifying the effectiveness and security of the monitoring method and tracing algorithm. The specific testing and verification process is as follows:

[0086] S4-1. Utilizing specific data and experimental results, combined with fault cases and domain expert knowledge, analyze and verify the causes of coking anomalies in pyrolysis furnaces; based on this, evaluate the performance and effectiveness of the monitoring algorithm, and improve and optimize the scheme, method, and adjustable parameters through feedback and iteration.

[0087] S4-2. Utilize opportunities such as furnace shutdown for coking to collect real-time data inside the furnace and tubes. Combine this with monitoring data before shutdown to conduct experimental verification of the method for localized safety hazard perception and prediction early warning of equipment. Compare the actual measurement results to verify the degree of agreement with the results inferred based on the tensor field. Through evaluation and iteration, optimize the tensor field model and improve the trend prediction algorithm.

[0088] S4-3. Connect the laboratory equipment and software platform, read DCS data, and conduct physical in-loop testing and verification based on the Matlab / Simulink simulation environment and experimental platform to comprehensively evaluate and verify the safety of the process.

[0089] Example 2

[0090] The present invention also provides a furnace tube coking sensing and prediction system for a tubular pyrolysis furnace, comprising: a construction module, an interaction module, and a prediction module;

[0091] The building blocks are used to construct a digital twin of a pyrolysis furnace;

[0092] The interaction module is used to interact with the virtual-real space data between the pyrolysis furnace and its digital twin.

[0093] The prediction module is used to reconstruct the spatiotemporal dynamic changes of temperature, pressure and flow fields inside the pyrolysis furnace tubes based on interactive virtual-real spatial data and deep learning from multi-source data. This enables the prediction of the optimal time for furnace tube decoking or burning based on statistical inference and trend prediction of coking in the furnace tubes.

[0094] In this embodiment, the construction module includes: a data integration unit, a reference architecture construction unit, and a digital twin construction unit;

[0095] The data integration unit is used to construct a digital twin conceptual model (PE, VE, Ss, DD, CN) with the pyrolysis unit as the core, starting from the pyrolysis unit and the corresponding process flow, and around the physical entity PE, virtual entity VE, digital twin data DD, security-oriented service Ss and the connection between entities CN, integrating information data and physical data.

[0096] The reference architecture building unit is used to establish a digital twin system reference architecture with a five-layer structure, based on integrated information and physical data, in a bottom-up manner, consisting of the real physical domain, measurement and control entity, digital twin, user domain, and cross-domain functional entity.

[0097] The digital twin construction unit is used to establish a digital model or information transmission model of PE based on the reference architecture of the digital twin system, on the basis of abstraction and modeling of ethylene plant, process and safety risks, to build bidirectional data communication between VE and entity, form a digital shadow model of ethylene plant process, and then form a complete digital twin model to construct a digital twin of the cracking process.

[0098] In this embodiment, the process of interacting with the virtual-real spatial data between the pyrolysis furnace and its digital twin includes:

[0099] Based on 3D virtual simulation of the ethylene cracking system and combined with a digital twin model, the data generation, exchange and flow relationships of each link in the ethylene process are cleared through the digital main line, so as to realize the dynamic reproduction and simulation of the production process and safety risks in the virtual environment, as well as the seamless flow of data and process traceability of each production process.

[0100] In this embodiment, the prediction module includes: an expandable model set S construction unit, a coupling relationship analysis unit, a change calculation unit, a deep learning model construction unit, a grid node simulation data calculation unit, a spatiotemporal vector field plotting unit, and a soft measurement data calculation unit;

[0101] The expandable model set S is constructed using the unidirectional pressure drop and heat transfer in the convection section of the ethylene cracking furnace, as well as the classical model of heat transfer of flue gas outside the tube, the model of temperature change and pressure transfer inside the tube, the model of pressure drop in the two-phase flow inside the tube, the model of heat transfer in the two-phase flow inside the tube, the model of pressure drop in the flue gas outside the tube, the heat transfer rate in the radiation chamber, the heat balance equation, and the model of the material cracking reaction process inside the tube. This constructs an expandable model set S of temperature, pressure, and flow rate changes in the convection chamber of the cracking furnace and the material cracking reaction.

[0102] The coupling relationship analysis unit is used to quantitatively analyze the coupling relationship between the changes in the characteristic quantity of furnace tube feed, the changes in the internal and external environment of the furnace tube, and the changes in the characteristic quantity of material at the convection chamber outlet, based on the expandable model set S and through numerical simulation.

[0103] The variable calculation unit is used to divide the convection chamber of the cracking furnace into a 3D grid D. Based on the expandable model set S, it calculates the temperature change, pressure vector three-directional component, velocity vector three-directional component, and flow rate change of each node D(x,y,z;t) of the grid D with time.

[0104] The deep learning model building unit is used to utilize the coupling relationship between the changes in the characteristics of the furnace tube feed, the changes in the internal and external environment of the furnace tube, and the changes in the characteristics of the material at the convection chamber outlet. Combined with the temperature, pressure, flow rate, and flow data of each node D(x,y,z;t) of the grid, a multi-layer convolutional neural network is trained to obtain a deep learning model, denoted as CNNa.

[0105] The grid node simulation data calculation unit is used to describe the temperature, pressure, flow velocity, and radiation characteristics of any point in the container and transmission pipeline using the field description. Taking the multi-order tensor field composed of the spatiotemporal changes and properties inside the convection chamber and transmission pipeline as the object, it combines 3D simulation data to drive the digital twin and obtain the simulation data of the temperature, pressure, flow velocity, and radiation characteristics of the tensor field at the grid node.

[0106] The spatiotemporal vector field plotting unit is used to train multidimensional DeepAR and Spacetimeformer using spatiotemporal extrapolation data from digital twins, obtain tensor field function fitting models based on deep learning, and gradient, divergence and curl function models, and plot spatiotemporal vector field plots of temperature, pressure, flow velocity and radiation respectively.

[0107] The soft sensor data calculation unit is used to train and transfer learn the model using historical data on temperature, pressure, flow rate, and radiation from the cracking furnace data warehouse; it updates the deep learning-based tensor field function fitting model and the gradient, divergence, and curl function models by combining temperature, pressure, flow rate, and radiation data collected on-site from the ethylene plant; and it obtains the spatiotemporal variation of the temperature, pressure, flow rate, and radiation characteristics of any point inside the ethylene cracking furnace tube using the updated model and spatiotemporal interpolation method.

[0108] In this embodiment, the prediction module further includes: a visualization unit, a pattern revealing unit, an inference model building unit, an online prediction unit, and a decision-making unit;

[0109] The visualization unit is used to reconstruct the temperature and pressure fields inside the tube based on measured data, and combined with the digital twin system, it uses the method of driving the digital twin with online measured data to visualize the coking location, coking degree and change process of the furnace tube.

[0110] The pattern revealing unit is used to train, test, verify, and iteratively optimize the deep learning model CNNa using historical data from the actual production processes of multiple groups of pyrolysis furnaces of the same model, to obtain a more accurate neural network model CNNb; by comparing and analyzing the changes in simulation results and measured data, an understanding of the distribution of material velocity, temperature, pressure, and reaction rate in the pyrolysis furnace is obtained, revealing the basic characteristics and mutual influence laws of flow, heat transfer, mass transfer, and pyrolysis reaction processes in the pyrolysis furnace tubes.

[0111] The inference model building unit is used to construct a causal relationship tree with coking and creep as the core, as well as a multi-factor regression analysis model of coking thickness in furnace tubes and a statistical inference model of coking location and thickness in furnace tubes based on local eddy currents of temperature and pressure fields, using a combination of data mining analysis and machine learning methods from cracking furnaces.

[0112] The online prediction unit is used to calibrate and perform soft measurement of the internal temperature and pressure fields of the furnace tube using measured operating data and deep learning models of temperature and pressure fields; and to realize online prediction of the spatiotemporal changes of temperature and pressure fields using soft measurement data of temperature and pressure changes based on a recurrent neural network DA-RNN with a two-stage attention mechanism.

[0113] The decision-making unit is used to drive the digital twin system with predictive data, combining a multi-factor regression analysis model for coking thickness and a statistical inference model for coking location. It uses a high-fidelity adaptive fault-tolerant index of characteristic parameters to smooth the prediction and performs decision-level fusion of multiple prediction models to predict the coking thickness and location of furnace tubes. Combined with domain expert knowledge, it determines the optimal coking treatment strategy and the best time for treatment, providing auxiliary decision-making for coking treatment.

[0114] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for sensing and predicting coking in a tube of a tubular cracking furnace, characterized in that: The following steps are involved: Build a digital twin of the cracking furnace; Interact virtual-real space data between the cracking furnace and the cracking furnace digital twin; Based on interactive virtual-real space data and combined with deep learning of multi-source data, the spatiotemporal dynamic change process of temperature field, pressure field and flow field in the cracking furnace tube is reconstructed to achieve statistical inference and trend prediction of furnace tube coking and the best time to clean or burn the furnace tube.

2. A method for sensing and predicting coking in a tubular cracking furnace according to claim 1, characterized in that: The method of constructing a digital twin of a cracking furnace includes: With the cracking device as the core, the digital twin conceptual model is constructed around the physical entity PE, virtual entity VE, digital twin data DD, security-oriented service Ss and entity connection CN, integrating information data and physical data; Based on integrated information data and physical data, a digital twin system reference architecture is established from the bottom up; Based on the digital twin system reference architecture, a digital model or information transfer model of PE is established, and two-way data communication between VE and entities is built to form a digital shadow model of the ethylene plant process, thereby forming a complete digital twin model and constructing a digital twin of the cracking process.

3. A method for sensing and predicting coking in a tubular cracking furnace according to claim 2, characterized in that: The method for interacting virtual-real space data between the cracking furnace and the cracking furnace digital twin includes: Based on the 3D virtual simulation of the ethylene cracking system and combined with the digital twin model, the digital main line clears the data generation, exchange and flow relationship of each link in the ethylene process, realizes the dynamic reproduction and deduction of the production process and safety risks in a virtual environment, as well as the seamless flow of data and process traceability of each production process.

4. A method for sensing and predicting coking in a tube of a tubular cracking furnace according to claim 3, characterized in that: The method for reconstructing the time-space dynamic change process of the temperature field, pressure field and flow field in the cracking furnace tube includes: Construct an expandable model set S of the cracking furnace convection chamber temperature, pressure, flow rate changes and material cracking reactions; Based on the expandable model set S, the coupling relationship between the change of the characteristic quantity of the furnace tube feed, the change of the internal and external environment of the furnace tube and the change of the characteristic quantity of the material at the outlet of the convection chamber is quantitatively analyzed through numerical simulation. The convection chamber of the cracking furnace is divided into a grid D using a 3D grid, and the temperature, pressure, flow rate and flow rate data of each node of the grid D are calculated based on the expandable model set S; By utilizing the coupling relationship between the changes in the characteristic quantities of the furnace tube feed, the changes in the internal and external environments of the furnace tube, and the changes in the characteristic quantities of the materials at the convection chamber outlet, combined with the temperature, pressure, flow velocity, and flow rate data of each node in the grid, a multi-layer convolutional neural network is trained to obtain a deep learning model. Use the field to describe the temperature, pressure, flow rate, and radiation characteristic attributes of any point in the container and the transmission pipeline, and obtain the simulation data of the temperature, pressure, flow rate, and radiation characteristic attributes of the tensor field at the grid nodes; Using digital twin space-time simulation data, we obtain the tensor field function fitting model based on deep learning, as well as the gradient, divergence and curl function models; Combined with the temperature, pressure, flow rate, and radiation data collected on-site in the ethylene plant, the tensor field function fitting model based on deep learning as well as the gradient, divergence, and curl function models are updated; combined with the updated model and the time-space interpolation method, the soft measurement data of the temperature, pressure, flow rate, and radiation characteristic attributes of any point in the ethylene cracking furnace tube that change with time and space are obtained.

5. A coking sensing and prediction system for a tubular cracking furnace, characterized in that: include: Building modules, interaction modules, and prediction modules; The building module is used to build a cracking furnace digital twin; The interaction module is used to interact the virtual-real space data between the cracking furnace and the cracking furnace digital twin; The prediction module is used to reconstruct the time-space dynamic change process of the temperature field, pressure field and flow field in the cracking furnace tube based on interactive virtual-real space data and combined with multi-source data deep learning, and realize the prediction of the best time for furnace tube coking or burning through statistical inference and trend prediction of furnace tube coking.

6. A furnace tube coking sensing and prediction system for a tubular cracking furnace according to claim 5, characterized in that: The building modules include: a data integration unit, a reference architecture building unit and a digital twin building unit; The data integration unit is used to build a digital twin concept model around the physical entity PE, the virtual entity VE, the digital twin data DD, the security service Ss and the entity connection CN, with the cracking device as the core, and integrate information data and physical data; The reference architecture building unit is used to establish a digital twin system reference architecture in a bottom-up manner based on integrated information data and physical data; The digital twin construction unit is used to establish a digital model or information transfer model of PE based on the digital twin system reference architecture, build two-way data communication between VE and entities, form a digital shadow model of the ethylene plant process, and then form a complete digital twin model to construct a digital twin of the cracking process.

7. A furnace tube coking sensing and prediction system for a tubular cracking furnace according to claim 6, characterized in that: The process of interacting virtual-real space data between the cracking furnace and the cracking furnace digital twin includes: Based on the 3D virtual simulation of the ethylene cracking system and combined with the digital twin model, the digital main line clears the data generation, exchange and flow relationship of each link in the ethylene process, realizes the dynamic reproduction and deduction of the production process and safety risks in a virtual environment, as well as the seamless flow of data and process traceability of each production process.

8. A furnace tube coking sensing and prediction system for a tubular cracking furnace according to claim 7, characterized in that: The prediction module includes: an expandable model set S construction unit, a coupling relationship analysis unit, a change amount calculation unit, a deep learning model construction unit, a grid node simulation data calculation unit, a time-space vector field map drawing unit and a soft measurement data calculation unit; The expandable model set S construction unit is used to construct an expandable model set S of the cracking furnace convection chamber temperature, pressure, flow rate changes and material cracking reaction; The coupling relationship analysis unit is used to quantitatively analyze the coupling relationship between the change of the characteristic quantity of the furnace tube feed, the change of the environment inside and outside the furnace tube, and the change of the characteristic quantity of the material at the convection chamber outlet through numerical simulation based on the expandable model set S; The change amount calculation unit is used to use a 3D grid to divide the cracking furnace convection chamber into a grid D, and based on the expandable model set S, calculate the temperature, pressure, flow rate, and flow data of each node of the grid D; The deep learning model building unit is used to utilize the coupling relationship between the change of the characteristic quantity of the furnace tube feed, the change of the environment inside and outside the furnace tube, and the change of the characteristic quantity of the material at the convection chamber outlet, and combine the temperature, pressure, flow rate, and flow rate data of each node of the grid to train a multi-layer convolutional neural network to obtain a deep learning model; The grid node simulation data calculation unit is used to use the field to describe the temperature, pressure, flow rate, and radiation characteristic attributes of any point in the container and the transmission pipeline, and obtain the temperature, pressure, flow rate, and radiation characteristic attribute simulation data of the tensor field at the grid node; The time-space vector field map drawing unit is used to use the digital twin space-time deduction simulation data to obtain a tensor field function fitting model based on deep learning, and a gradient, divergence and curl function model; The soft measurement data calculation unit is used to update the tensor field function fitting model based on deep learning and the gradient, divergence and curl function models in combination with the temperature, pressure, flow rate and radiation data collected on-site in the ethylene device; combined with the updated model and the time-space interpolation method, the soft measurement data of the temperature, pressure, flow rate and radiation characteristic attributes of any point in the ethylene cracking furnace tube that change with time and space are obtained.

Images