An ethylene cracker optimization method, apparatus, medium, and product

Abstract

This application relates to the field of artificial intelligence technology, specifically to an optimization method, equipment, medium, and product for an ethylene cracking unit. The method involves uniformly predicting and calculating the baseline operating data of the ethylene cracking unit to determine its key performance indicators. Based on these key performance indicators and pre-set unit constraints, a dual-objective optimization model is constructed. The dual-objective optimization model is then optimized to determine the optimal solution set for the ethylene cracking unit. Next, based on a pre-set mechanistic interpretation strategy, the optimal solution set is interpreted mechanistically to obtain the corresponding explanatory text. Based on the optimal solution set and the explanatory text, the ethylene cracking unit is optimized and adjusted to improve its operating efficiency and safety stability, while simultaneously enhancing the economic benefits and environmental performance of ethylene cracking.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to an optimization method, equipment, medium and product for an ethylene cracking unit. Background Technology

[0002] Ethylene cracking furnaces are core equipment in the petrochemical industry, and their operational optimization is crucial for improving the efficiency of the entire industrial chain and reducing energy consumption and carbon emissions. With the advancement of dual-carbon goals and the increasing market demand for high-value-added products, achieving synergistic optimization that maximizes economic benefits and minimizes carbon emissions while ensuring the safe and stable operation of the equipment has become a critical issue that the industry urgently needs to address.

[0003] Currently, traditional technologies typically rely on mechanistic models or data-driven approaches to perform single-objective optimization of ethylene cracking units. However, the ethylene cracking process itself is characterized by its complexity, involving multiple feedstocks, objectives, and constraints. This means that while ensuring safety boundaries such as tube wall temperature and furnace pressure, it is necessary to simultaneously balance maximizing efficiency and minimizing carbon emissions, while also considering the selectivity of ethylene or propylene and downstream loads. Furthermore, key operational variables in the cracking process, such as the coiloutlet temperature (COT), exhibit strong nonlinearity and high coupling with various performance indicators (such as efficiency, carbon emissions, and ethylene selectivity). This makes it difficult for traditional single-objective optimization or isolated point-based optimization methods to achieve a global and operable balance among multiple conflicting optimization objectives.

[0004] Therefore, there is an urgent need for an optimization method for ethylene cracking units to improve their operating efficiency and safety stability, while also enhancing the economic benefits and environmental performance of ethylene cracking. Summary of the Invention

[0005] This invention provides an optimization method, equipment, medium, and product for ethylene cracking units, which improves the operating efficiency and safety stability of ethylene cracking units while enhancing the economic benefits and environmental performance of ethylene cracking.

[0006] In a first aspect, this application provides a method for optimizing an ethylene cracking unit, the method comprising: The baseline operating data of the ethylene cracking unit is uniformly predicted and indexed to determine the key performance indicators of the ethylene cracking unit. The baseline operating data represents the multi-feed information, operating variable information and energy consumption information of the ethylene cracking unit, and the key performance indicators represent the unit efficiency indicators, carbon emission indicators and product output indicators of the ethylene cracking unit. A dual-objective optimization model is constructed based on the key performance indicators and preset device constraints, and optimization calculations are performed on the dual-objective optimization model to determine the optimal solution set of the ethylene cracking unit; the optimal solution set represents the non-dominated solution set among the device benefit indicators, carbon emission indicators, and product output indicators. Based on a preset mechanistic interpretation strategy, the optimized solution set is interpreted mechanistically to obtain the interpretation text corresponding to the optimized solution set. The interpretation text represents the operation variable adjustment logic corresponding to the optimized solution set. Based on the optimized solution set and the explanatory text, the ethylene cracking unit is optimized and adjusted.

[0007] Optionally, the construction of a dual-objective optimization model based on the key performance indicators and preset device constraints includes: Based on the device benefit index, a first optimization objective function corresponding to the dual-objective optimization model is constructed; the first optimization objective function is used to maximize the device benefit index. Based on the carbon emission index, a second optimization objective function corresponding to the dual-objective optimization model is constructed; the second objective function is used to minimize the carbon emission index. The bi-objective optimization model is constructed based on the first optimization objective function, the second optimization objective function, and the device constraints.

[0008] Optionally, the step of performing optimization calculations on the bi-objective optimization model to determine the optimal solution set for the ethylene cracking unit includes: Based on the baseline operating data, finite difference adjustment calculations are performed on the operating variables of each raw material to determine the marginal sensitivity of each operating variable. Based on each marginal sensitivity, target operational variables are selected from the operational variables; Based on a preset multi-objective evolutionary algorithm, and in combination with the device constraints and the objective operation variables, the bi-objective optimization model is solved to generate the optimized solution set.

[0009] Optionally, the step of performing finite difference adjustment calculations on the operating variables of each raw material based on the baseline operating condition data to determine the marginal sensitivity of each operating variable includes: Apply positive and negative perturbations of preset magnitudes to each operated variable to obtain the perturbed operated variables; The key performance indicators are updated based on the predicted operational variables after the perturbation. The marginal sensitivity is determined based on the updated key performance indicators.

[0010] Optionally, the step of performing a mechanistic interpretation on the optimized solution set based on a preset mechanistic interpretation strategy to obtain the interpretation text corresponding to the optimized solution set includes: For each representative solution in the optimized solution set, based on the adjustment direction of the operation variables corresponding to each representative solution and the corresponding marginal sensitivity value, the query request corresponding to each representative solution is determined; each representative solution represents a combination of operation variables that satisfy the device constraints and correspond to different device benefit indicators and carbon emission indicators. Based on each query request, the preset process knowledge base is searched to obtain the corresponding process knowledge entries; Based on a preset explanation template, the explanation text is generated by combining the process knowledge entries, the adjustment direction of the operational variables, and the marginal sensitivity.

[0011] Optionally, the step of uniformly predicting and calculating the baseline operating data of the ethylene cracking unit to determine the key performance indicators of the ethylene cracking unit includes: Based on the actual operating data of the ethylene cracking unit, a fuel consumption prediction model, a yield prediction model, and a steam generation prediction model are constructed. Based on the fuel consumption prediction model, yield prediction model, and steam generation prediction model, the baseline operating data are predicted and calculated to obtain the corresponding prediction results; the prediction includes at least the fuel gas consumption, product yield, steam production, and steam consumption of the ethylene cracking unit. Based on the prediction results, the key performance indicators are calculated.

[0012] Optionally, the device constraints include: operation optimization constraints, furnace group load constraints, and production load constraints; the operation optimization constraints are determined based on the raw material properties, device principle, and actual operating data of each feedstock in the ethylene cracking unit; the furnace group load constraints are determined based on the capacity parameters of the ethylene cracking unit, furnace tube thermal intensity limits, and furnace group operating load distribution relationship; and the production load constraints are determined based on the load requirements of the separation and distillation unit.

[0013] In a second aspect, this application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the ethylene cracking device optimization methods described in the first aspect above.

[0014] Thirdly, this application provides a computer storage medium storing computer program instructions, which are executed by a processor using any of the ethylene cracking apparatus optimization methods described in the first aspect above.

[0015] Fourthly, an embodiment of this application provides a computer program product, including computer program instructions, which, when executed by a processor, implement any of the ethylene cracking apparatus optimization methods described in the first aspect above.

[0016] The beneficial effects of this invention are as follows: This application provides an optimization method for an ethylene cracking unit. This method involves uniformly predicting and calculating the baseline operating data of the ethylene cracking unit to determine its key performance indicators. Based on these key performance indicators and preset unit constraints, a dual-objective optimization model is constructed. The dual-objective optimization model is then optimized to determine the optimal solution set for the ethylene cracking unit. Next, based on a preset mechanistic interpretation strategy, the optimal solution set is interpreted mechanistically to obtain the corresponding explanatory text. Based on the optimal solution set and the explanatory text, the ethylene cracking unit is optimized and adjusted to improve its operating efficiency and safety stability, while also enhancing the economic benefits and environmental performance of ethylene cracking. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0018] Figure 1 A schematic flowchart illustrating an optimization method for an ethylene cracking unit provided in an embodiment of this application; Figure 2 A schematic diagram of a Pareto solution set provided for an embodiment of this application; Figure 3 A schematic diagram illustrating an explanatory text provided for an embodiment of this application; Figure 4 A schematic diagram illustrating the explanatory framework of an ethylene cracking unit optimization method provided in the embodiments of this application; Figure 5 A flowchart illustrating another method for optimizing an ethylene cracking furnace system based on agent-assisted decision-making, provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. Unless otherwise specified, the embodiments and features in the embodiments of this application can be arbitrarily combined with each other. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than that shown here.

[0020] The terms "first" and "second" in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the term "comprising" and any variations thereof are intended to cover non-exclusive protection. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices. The term "multiple" in this application can mean at least two, for example, two, three, or more, and this application does not impose limitations.

[0021] The term "and / or" in the embodiments of this application is merely a description of the association relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0022] It is understood that the following specific embodiments of this application involve data related to ethylene cracking. When the various embodiments of this application are applied to specific products or technologies, relevant licenses or consents are required, and the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. For example, relevant volunteers can be recruited and agreements can be signed to authorize their data, thereby enabling the implementation using the data of these volunteers; alternatively, implementation can be carried out within an authorized organization, using data from members of the organization to implement the following implementation methods for data management; or, in specific implementations, the relevant data used are all simulated data, such as simulated data generated in a virtual scenario.

[0023] The embodiments of this application relate to artificial intelligence and machine learning (ML) technologies, and are primarily designed based on machine learning in artificial intelligence.

[0024] Artificial intelligence (AI) is the theory, methods, technology, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that attempts to understand the essence of intelligence and produce a new kind of intelligent machine that can react in a way similar to human intelligence. AI studies the design principles and implementation methods of various intelligent machines, enabling them to possess the functions of perception, reasoning, and decision-making.

[0025] Artificial intelligence (AI) is a comprehensive discipline encompassing a wide range of fields, including both hardware and software technologies. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies primarily include computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0026] Machine learning is a multidisciplinary field involving probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory, among others. It specifically studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence; its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and instructional learning.

[0027] Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence. Its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, and inductive learning. Artificial Neural Networks (ANNs) abstract the neural network of the human brain from an information processing perspective, establishing a simple model and forming different networks with different connection methods. A neural network is a computational model composed of a large number of interconnected nodes (or neurons). Each node represents a specific output function called the activation function. The connection between any two nodes represents a weighted value for the signal passing through that connection, called a weight. This is equivalent to the memory of the artificial neural network. The network's output varies depending on the network's connection methods, weight values, and activation functions. The network itself is usually an approximation of a certain algorithm or function in nature, or it may be an expression of a logical strategy.

[0028] The design concept of the embodiments of this application is briefly introduced below: Ethylene cracking furnaces are core equipment in the petrochemical industry, and their operational optimization is crucial for improving the efficiency of the entire industrial chain and reducing energy consumption and carbon emissions. With the advancement of dual-carbon goals and the increasing market demand for high-value-added products, achieving synergistic optimization that maximizes economic benefits and minimizes carbon emissions while ensuring the safe and stable operation of the equipment has become a critical issue that the industry urgently needs to address.

[0029] Currently, traditional technologies typically optimize ethylene cracking units based on mechanistic models or data-driven approaches, focusing on a single objective. However, the ethylene cracking process itself is characterized by its complexity, involving multiple feedstocks, objectives, and constraints. This means that while ensuring safety boundaries such as tube wall temperature and furnace pressure, it is necessary to simultaneously balance maximizing efficiency and minimizing carbon emissions, while also considering ethylene or propylene selectivity and downstream loads. Furthermore, the key operational variables in the cracking process (such as cracking temperature COT) exhibit strong nonlinearity and high coupling with various performance indicators (such as efficiency, carbon emissions, and ethylene selectivity). This makes it difficult for traditional single-objective optimization or isolated point-based optimization methods to achieve a global and operable balance among multiple conflicting optimization objectives.

[0030] In view of the above problems, embodiments of this application provide an optimization method, equipment, medium, and product for an ethylene cracking unit. This method involves uniformly predicting and calculating the baseline operating data of the ethylene cracking unit to determine its key performance indicators. Based on these key performance indicators and preset unit constraints, a dual-objective optimization model is constructed. The dual-objective optimization model is then optimized to determine the optimal solution set for the ethylene cracking unit. Next, based on a preset mechanistic interpretation strategy, the optimal solution set is interpreted mechanistically to obtain the corresponding explanatory text. Based on the optimal solution set and the explanatory text, the ethylene cracking unit is optimized and adjusted to improve its operating efficiency and safety stability, while also enhancing the economic benefits and environmental performance of ethylene cracking.

[0031] Furthermore, this application constructs a complete closed loop from unified evaluation, multi-objective optimization, mechanism explanation to execution decision-making, systematically solving the optimization challenges of ethylene cracking units. This method avoids ambiguity of multiple indicators through a unified calculation benchmark, directly addresses the core contradiction between efficiency and carbon emissions through a dual-objective optimization model, and provides a feasible trade-off. This application also makes the optimization results more transparent and trustworthy through mechanistic explanation, greatly improving the acceptability and executability of the optimization scheme in the field. Moreover, under strict unit constraints, this application ensures that the entire optimization and adjustment process remains within safe boundaries, achieving a synergistic improvement in multiple dimensions of ethylene cracking unit performance, including economy, environmental protection, safety, and operational feasibility.

[0032] Furthermore, this application embodiment can perform a one-time forward calculation on the operating parameters of multiple feedstocks (naphtha (NAP), heavy vacuum gas oil (HVGO), liquefied petroleum gas (LPG), and mixed C2) under historical and current operating conditions based on a unified prediction and index calculation module. This yields key indicators such as profit, carbon emissions, superheated steam recovery (SS), desuperheating steam consumption (DS), and ethylene or propylene yield and production, thereby forming an initial candidate set of feasible operating conditions. This application also utilizes finite difference calculations to assess the marginal sensitivity of key operating variables (COT, dilution to hydrogen carbon ratio (DHR), and feedstock (FEED)) for each feedstock, identifying feasible solutions exhibiting local advantages in the dual objectives (profit maximization and carbon emission minimization) as high-value solutions. Next, the Nondominated Sorting Genetic Algorithm II (NSGA-II) multi-objective optimization is used to iteratively search and update within the constraints of device safety and boundary conditions. Combined with historical feasible solutions, warm start and crowding maintenance are performed to gradually approach the Pareto front. The representative solutions and sensitivity evidence produced in each iteration are used for the Retrieval-Augmented Generation (RAG) of the mechanistic interpretation engine to ensure the consistency and traceability of "variable direction - index response - boundary proximity". Finally, a set of Pareto front solutions reflecting the optimal trade-off between profit and carbon emissions is obtained. The optimal operating parameters can be selected on the front according to specific production goals (such as prioritizing the increase of ethylene yield or the reduction of carbon emissions), and executable parameter tuning suggestions in the form of arrow templates are output. Thus, compared with traditional optimization methods that rely solely on single-point experience, the embodiments of this application balance convergence speed and interpretability in a closed loop of unified caliber calculation, sensitivity measurement, NSGA-II optimization, RAG interpretation, and result placement, providing a practical and feasible cost reduction, efficiency improvement, and emission reduction implementation plan for ethylene cracking units.

[0033] Please refer to Figure 1 The following is a flowchart illustrating an optimization method for an ethylene cracking unit provided in an embodiment of this application. The specific implementation process of this method is as follows: Step 101: Perform unified prediction and index calculation on the baseline operating data of the ethylene cracking unit to determine the key performance indicators of the ethylene cracking unit.

[0034] In this embodiment of the application, a unified prediction and calculation framework will be used to transform the multidimensional and heterogeneous baseline operating data of the ethylene cracking unit into key performance indicators that are consistent in scope and can be quantified and compared.

[0035] Specifically, the baseline operating data in this application embodiment is a comprehensive digital representation of the real-time and historical operating status of the equipment. This may include, but is not limited to: multi-feed information reflecting the diversity and properties of raw materials (e.g., the composition and flow rate of each feed stream), operating variable information set or adjusted by the operator (e.g., cracking temperature COT, dilution ratio, and residence time), and energy consumption information reflecting energy and material consumption (e.g., fuel consumption and steam consumption). Based on a pre-established and validated process mechanism model or a high-precision data-driven model, this application will map the above information into key performance indicators that can be used for decision analysis. These include equipment benefit indicators related to economic gains (e.g., profit margin or high-value product output), carbon emission indicators corresponding to environmental requirements (e.g., carbon emissions per unit of product or total carbon emissions), and product yield indicators related to production planning and downstream balance (e.g., ethylene yield, propylene yield, and total terpene yield). This establishes a unified and objective quantitative benchmark for subsequent optimization, ensuring that all optimization activities are conducted within the same evaluation system.

[0036] In one possible implementation, this application embodiment will construct a fuel consumption prediction model, a yield prediction model, and a steam generation prediction model based on the actual operating data of the ethylene cracking unit. The constructed models will be used to predict and calculate the baseline operating data to obtain the corresponding prediction results, such as the fuel gas consumption, product yield, steam production, and steam consumption of the ethylene cracking unit. Based on the prediction results, key performance indicators will be calculated.

[0037] Specifically, this application embodiment will predict the yield vector, fuel consumption, and steam recovery in one go based on the model manager, and solidify ethylene and propylene into corresponding indicators, compatible with both "percentage" and "mass fraction" metrics. For example, it will collect industrial data on the actual operation of the ethylene cracking unit, including but not limited to the flow rates of multiple feedstocks (naphtha, hydrotreated tail oil, liquefied petroleum gas, and mixed C2), furnace operating variables (furnace tube outlet temperature, vapor-to-hydrocarbon ratio), and fuel gas consumption. Based on the heat transfer in the convection / radiation section and the conservation of furnace energy, it will establish a fuel gas consumption data-driven prediction model using the fuel gas composition and lower heating value. Simultaneously, based on waste heat recovery and the thermodynamic balance of the steam system, it will establish an SS steam generation prediction model through the nonlinear relationship between actual industrial data of SS steam generation and furnace operating variables. Furthermore, based on the cracking reaction network and furnace energy balance, it will construct a yield mechanism model using feedstock flow rates and cracking depth characterization (furnace tube outlet temperature and vapor-to-hydrocarbon ratio), establish a mapping relationship from feedstock to product fraction vectors, and solidify ethylene or propylene output as corresponding indicators.

[0038] In one possible implementation, the key performance indicators in the embodiments of this application include, but are not limited to, profit, carbon emissions, ultra-high pressure steam (SS), steam consumption (DS), and the yield and production of ethylene or propylene.

[0039] Step 102: Construct a dual-objective optimization model based on key performance indicators and preset device constraints.

[0040] In this embodiment of the application, a dual-objective optimization model will be constructed based on key performance indicators and device constraints to discover the optimal dual-objective trade-off scheduling scheme under physical constraints.

[0041] Specifically, the dual-objective optimization model in this application uses the unit benefit index and carbon emission index as independent optimization objectives (usually set as maximizing unit benefit and minimizing carbon emission), while incorporating the product output index as an important boundary condition or sub-objective into the optimization framework. Unit constraints characterize the insurmountable safety and process boundaries of the ethylene cracking unit and are a core component of the model. They can be expressed by a series of mathematical inequalities or equations, such as the upper limit of tube bundle wall temperature to ensure the lifespan of the cracking furnace tubes, the range of furnace pressure to maintain system stability, the flue gas temperature limit to protect downstream equipment, and the steam and feedwater load windows that the utility system can provide.

[0042] In one possible implementation, this application embodiment will construct a first optimization objective function corresponding to the dual-objective optimization model based on the device benefit index, used to maximize the device benefit index, and construct a second optimization objective function based on the carbon emission index, used to minimize the carbon emission index. Thus, a dual-objective optimization model is constructed through the first optimization objective function, the second optimization objective function, and preset device constraints.

[0043] Specifically, the device constraints in this application embodiment may include, but are not limited to, operational optimization constraints, furnace group load constraints, and production load constraints. The operational optimization constraints are determined based on the feedstock properties, device principles, and actual operating data of the ethylene cracking unit. The furnace group load constraints are determined based on the capacity parameters of the ethylene cracking unit, furnace tube thermal intensity limits, and the load distribution relationship of the furnace group. The production load constraints are determined based on the load requirements of the separation and distillation unit. Thus, this application will construct a dual-objective optimization model that maximizes profit and minimizes carbon emissions within the constraints of device safety and boundary conditions.

[0044] Step 103: Perform optimization calculations on the dual-objective optimization model to determine the optimal solution set for the ethylene cracking unit.

[0045] In this embodiment, the constructed bi-objective optimization model can be optimized using multi-objective evolutionary algorithms such as NSGA-II. The resulting set of optimal solutions can be a Pareto optimal set. Each solution in the set represents a specific trade-off point between plant efficiency and carbon emissions under given constraints, at which further improvement is not possible without compromising the other objective, while the product output indicator is also at a relatively optimal level. Therefore, this set of optimal solutions objectively reveals the fundamental trade-off between economic benefits and environmental protection, representing a non-dominated solution among plant efficiency indicators, carbon emission indicators, and product output indicators.

[0046] In one possible implementation, the embodiments of this application will perform finite difference adjustment calculations on the operating variables of each raw material based on baseline operating condition data, determine the marginal sensitivity of each operating variable, and select target operating variables from each operating variable based on each marginal sensitivity. Then, based on a preset multi-objective evolutionary algorithm, combined with device constraints and target operating variables, the dual-objective optimization model will be solved to generate an optimized solution set.

[0047] Specifically, in this embodiment of the application, finite difference adjustment refers to making small unit fluctuations in the operating variables of different raw materials to recalculate the model results indicators, and then calculating the marginal sensitivity based on the change in the indicators.

[0048] In one possible implementation, the embodiments of this application will apply positive and negative perturbations of preset magnitude to each operated variable to obtain the perturbed operated variables, and perform prediction calculations based on the perturbed operated variables to update the key performance indicators, thereby determining the marginal sensitivity based on the updated key performance indicators.

[0049] Specifically, in order to quantify the marginal impact of each operational variable on key indicators, this application embodiment uses the finite difference method to adjust the variables of each component (such as NAP, HVGO, LPG, C2) by a unit, thereby calculating the sensitivity matrix of indicators such as Profit, Carbon, SS, DS, Ethylene (C2H4), and Propylene (C3H6). Let the decision vector for the i-th component be:

[0050] The device-level index vector is:

[0051] right The amplitude of the Jth component is The positive tug, and the amplitude is The negative swivel, the average value of the two-way differential, and the sensitivity is defined as:

[0052] in, It is marginal sensitivity. To be applied to The tiny fluctuations on For receiving the model The index function.

[0053] Based on the above sensitivity assessment, the selection of the dominant lever is derived.

[0054] In one possible implementation, embodiments of this application may employ NSGA-II for multi-objective optimization, including optimization of carbon emissions and optimization to maximize benefits. The objective function for multi-objective optimization is:

[0055] in, Representing the Taiwan-type pyrolysis furnace, Represents the type of feed. The yield of each product, This is the feed load for a single cracking furnace. for Price factors, This refers to fuel gas consumption. For feed rate, To dilute the amount of steam used, This refers to the output of ultra-high pressure steam.

[0056]

[0057] in, for Carbon content, For feed rate, The amount of product. This refers to the consumption of fuel gas. for carbon emission factors, This represents the amount of electricity consumed.

[0058] In conclusion, The objective function for profitability is calculated by multiplying product output by the corresponding price factor and subtracting costs such as fuel gas and raw materials. Using the carbon emission objective function, the CO2 produced by coking, as well as the carbon emissions caused by fuel combustion, steam, and electricity, are calculated through mass conservation.

[0059] For details, please refer to Figure 2 The diagram shown is a schematic representation of a Pareto solution set provided in an embodiment of this application. Figure 2 The Pareto front solution obtained after multi-objective real-time optimization is shown, and the relationship between carbon emissions and the overall benefits of the furnace group is explained. The Pareto front solution is calculated based on the following operating conditions: NAP COT: 840 s℃ DHR: 0.55 FEED: 29 t / h, HVGO COT: 780 s℃ DHR: 0.82 FEED: 29 t / h, LPG COT: 849 s℃ DHR: 0.40 FEED: 28 t / h, C2 COT: 850 s℃ DHR: 0.35 FEED: 25 t / h, population size 120, generation 80. The representative solution will be selected from the Pareto solution set generated: (1) maximizing benefits ( (2) Minimize carbon emissions (3) and the compromise solution, and normalized coordinates to the ideal point ( , The solution is obtained by minimizing the Euclidean distance of , while eliminating boundary endpoints to select thematic solutions that satisfy the diversity of production strategies.

[0060] Step 104: Based on the preset mechanistic interpretation strategy, perform mechanistic interpretation on the optimized solution set to obtain the interpretation text corresponding to the optimized solution set.

[0061] In this embodiment, the optimized solution set presented in numerical form is given operational semantics that can be understood by human operation experts, thereby realizing the conversion between mathematical models and on-site operations.

[0062] Specifically, the mechanistic interpretation strategy in this application is based on rules or templates of process mechanism knowledge, primarily used for reverse analysis and translation of each non-dominated solution in the solution set. This strategy analyzes the key reasons leading to the specific performance of this solution in terms of efficiency and carbon emissions. For example, it identifies that an increase in cracking temperature dominates the increase in ethylene yield, thus improving efficiency, but simultaneously leading to increased fuel consumption and carbon emissions; or that adjusting the dilution ratio optimizes product distribution, maintaining efficiency while reducing emissions. Thus, this strategy can transform causal analysis into structured natural language descriptions, forming explanatory text. This explanatory text clearly indicates the specific operational variable adjustment logic from the current baseline operating conditions to the optimized solution. For example, the explanatory text could be: "To achieve this optimization point, it is recommended to increase the cracking temperature A by X degrees Celsius and simultaneously decrease the dilution ratio of feedstock B by Y. This operation is expected to increase the ethylene yield by Z%, but will lead to an increase in fuel consumption by W%."

[0063] In one possible implementation, this embodiment of the application will, for each representative solution in the optimized solution set, determine the corresponding query request based on the adjustment direction of the corresponding operational variables and the corresponding marginal sensitivity value of each representative solution. Each representative solution represents a combination of operational variables that satisfies the device constraints and corresponds to different device benefit indicators and carbon emission indicators. Next, based on each query request, a preset process knowledge base is searched to obtain the corresponding process knowledge entries. Then, based on a preset explanation template, and combining the process knowledge entries, the adjustment direction of the operational variables, and the marginal sensitivity, an explanation text is generated.

[0064] Specifically, this application embodiment can build an optimized auxiliary decision-making agent based on the LangChain framework, and register a toolchain for the coordination of unfair functions. It includes the following modules and coordinates them in a toolchain orchestration manner: (1) LLM provider and message orchestration: Through local or remote large language model interface (e.g., including system / human message templates and temperature and text length control), the intent of user instructions is parsed and structured to form a quadruple of "raw material - variable - indicator - target".

[0065] (2) RAG Searcher (Vector Index): An embedding library is constructed for pyrolysis mechanism, equipment boundary, safety warning, material distribution attributes, and operating process. Top-k search is performed based on similarity s, and filtering is performed using a threshold τ.

[0066] in, It is the final, filtered set of knowledge items returned by the retrieval process. The i-th document or knowledge fragment in the knowledge base (embedded library), For similarity function, It is the predicted similarity threshold. For query vector, This represents a specific task or function instruction. Represents the current input data or state. This represents the optimization goal, and the context refers to the operation scenario, supplementary information, or dialogue records.

[0067] (3) Unified calculation and sensitivity tools: Yield, Fuelsum, SS, DS and sensitivity are obtained in a single forward pass.

[0068] (4) NSGA-II is preferred to generate the Pareto set P, and when computing power / dependency is limited, it falls back to the baseline neighborhood weighted scan, where the knee point is normalized to the ideal point as:

[0069] (5) Mechanistic explanation engine: integrates material fingerprints, variable direction, sensitivity evidence and boundary proximity to generate explanatory text for arrow templates; (6) Historical database and front-end interface: Persist representative solutions, key indicators and explanatory texts for front-end card-based display and traceability comparison.

[0070] In one possible implementation, the mechanism explanation in the embodiments of this application can be implemented using the RAG paradigm, specifically including: (1) Knowledge base: Based on third-party and self-built knowledge bases (pyrolysis mechanism, equipment operation procedures, material distribution fingerprint and boundary and safety prompts), a vector index is constructed.

[0071] (2) Intent recognition and retrieval: The Agent analyzes the scene intent based on the user and system prompt, extracts the triple of raw material-variable-indicator, and retrieves the k pieces of evidence with the highest similarity (e.g., preferably k=3). 5).

[0072] (3) Evidence fusion and interpretation generation: The retrieved fragments are spliced ​​with numerical evidence (sensitivity, boundary proximity, representative solution index) to generate a text interpretation of the arrow template.

[0073] (4) De-illusion and consistency control: Source labeling, repeatability / length control and terminology calibration are performed on the generated results to ensure consistency with the calculation scope of the model and to avoid exaggerating the role of the method; finally, the material-level explanation and overall summary are output and archived.

[0074] For details, please refer to Figure 3 The image shown is a schematic diagram of an explanatory text provided in an embodiment of this application. Figure 3 This paper demonstrates the mechanism of the arrow template displayed on the front-end page after the process of adjusting the operation of four raw materials in this application.

[0075] Step 105: Optimize and adjust the ethylene cracking unit based on the optimized solution set and interpretation text.

[0076] In this embodiment, control instructions that can directly guide or assist production adjustments are generated based on the quantitative optimization results of the optimized solution set and the qualitative explanations in the explanatory text. This step is not simply recommending a numerical solution, but rather allowing operators or advanced control systems to make the final decision based on multiple feasible trade-offs provided by the optimized solution set, as well as the adjustment logic and potential impact behind each solution as described in the explanatory text.

[0077] Specifically, this application can select the most suitable optimization solution based on the current production and operation strategy (such as prioritizing efficiency or mandatory emission reduction), and strictly follow the operation variable adjustment logic (including adjustment object, direction and magnitude) detailed in the corresponding explanatory text to accurately adjust the actual operation parameters such as the controller setpoint of the ethylene cracking unit, thereby ensuring that the optimization scheme can be implemented safely, orderly and understandably, and transforming the upstream model calculation results into changes in the actual operating status of the downstream unit, thereby optimizing the operation of the ethylene cracking unit.

[0078] In one possible implementation, the embodiments of this application can associate and store parameters such as the material distribution conditions, operating variables, and key performance indicators corresponding to the optimized solution set, as well as explanatory text, to form a traceable decision version, and control the front-end interface to display it visually, such as displaying performance comparison charts between different optimized solutions, listing or graphically representing the operating variable adjustment suggestions, and displaying the explanatory text in a region.

[0079] Specifically, this application embodiment can utilize a single-page application (SPA) structure based on HyperText Markup Language 5 (HTML5), Cascading Style Sheets Level 3 (CSS3), and native JavaScript. The visualization engine ECharts is employed to provide graphical representations of indicators, component comparisons, and sensitivity bar charts, supporting offline static resource loading and responsive redrawing. The front-end interface can include functional blocks such as overall indicator cards, component comparison tables, sensitivity visualizations, mechanism explanation views, safety alert views, and interactive dialog areas, and supports incremental refresh in dependency order after representative solutions are updated.

[0080] In one possible implementation, the results of this application can be stored at the single-furnace level. For example, for each raw material / furnace, the multi-objective optimization results obtained after meeting the corresponding constraints are written into a relational database (preferably MySQL) with "furnace number / raw material / time stamp (or batch number)" as the primary key. The stored fields include at least: baseline and representative solution operation variables (COT, DHR, FEED), key indicator set (Profit, Carbon, SS, DS, C2H4, C3H6) and feasibility flags.

[0081] In one possible implementation, the embodiments of this application can be summarized and stored at the furnace group level. For example, the group-level representative solution and overall indicators (including total ethylene or propylene production, group-level ultra-high pressure steam recovery or steam consumption SS / DS, and group-level profit or carbon emission Profit / Carbon) formed under the premise of satisfying the overall constraints of the furnace group are written into the summary table with "device / timestamp (or scheme type)" as the primary key.

[0082] In one possible implementation, embodiments of this application may also archive the explanatory text and the chain of evidence. For example, the agent's mechanistic explanatory text and its evidentiary elements may be archived together. These evidentiary elements include, but are not limited to: Top-k retrieval document identifiers, component fingerprint identifiers, variable direction and marginal sensitivity vectors, boundary proximity, and representative solution variable settings. The explanatory text and evidence may be stored in a mixed format of structured fields and JSON to support subsequent tracing and comparison.

[0083] In one possible implementation, consistency and idempotency: the stored procedures of this application embodiment employ database transaction commit and idempotent write strategies (e.g., controlled by primary keys and version numbers), and perform recovery processing for data that has been repeatedly written or partially interrupted. Specifically, the indexes and retrieval systems stored in this application can establish joint indexes based on time, furnace number, raw materials, scheme type, and target focus, and support condition-based historical scheme retrieval and comparative analysis. The front-end can also interact with the database through back-end services, such as Representational State Transfer (REST), JavaScript Object Notation (JSON), or equivalent interfaces. The front-end does not directly connect to the database, thus ensuring data consistency and clear permission boundaries.

[0084] In one possible implementation, refer to Figure 4 The diagram illustrates the explanatory framework of an agent-assisted decision-making optimization method for an ethylene cracking unit provided in this application. The framework, from bottom to top, includes a model layer, an optimization layer, and an intelligent decision-making layer. At the bottom layer, a predictive model for key performance indicators of the cracking process is constructed to achieve high-precision prediction of process variables such as product yield, fuel gas consumption, and steam generation (SS). The middle layer is the optimization solution layer, responsible for establishing the mathematical description of the multi-objective optimization problem and calling corresponding optimization algorithms to achieve global optimization of operating parameters. The upper layer is the agent decision-making and scheduling layer, which comprehensively calls various tool modules to perform data entry, optimization calculations, and result storage, thereby realizing full-process optimization and autonomous decision-making of the ethylene cracking furnace system.

[0085] In one possible implementation, refer to Figure 5 The diagram shows a flowchart of another agent-assisted decision-making optimization method for an ethylene cracking furnace system provided in this application. This application collects baseline operating conditions for multiple feedstocks and uniformly calculates key indicators, including at least: profit, carbon emissions, ultra-high pressure steam (SS), steam consumption (DS), and ethylene or propylene yield and output. For each feedstock, finite difference adjustments are made to the cracking furnace operating variables to calculate their marginal sensitivity to the key indicators. Next, a dual-objective model for maximizing profit and minimizing carbon emissions is constructed within the constraints of equipment safety and boundary conditions. The dual optimization objectives are maximizing ethylene cracking benefits and minimizing carbon emissions. Candidate solutions (e.g., Pareto front solution sets) are generated using NSGA-II, and representative solutions are selected based on knee points or thematic objectives. Then, the Agent-based mechanistic interpretation engine is invoked to generate feedstock-level interpretations and overall summaries based on raw material attribute knowledge, variable change direction, sensitivity evidence, and boundary proximity. Explanatory text in operation-oriented arrow templates is output, and the feedstock operating conditions, overall indicators, and explanation text corresponding to the representative solutions are saved and used for front-end display.

[0086] Please see Figure 6As shown, based on the same technical concept, this application also provides a computer device 60. In one embodiment, this computer device can be a device specifically for optimizing an ethylene cracking unit, or it can be a device for overall control and energy storage scheduling. The computer device is as follows... Figure 6 As shown, it includes a memory 601, a communication module 603, and one or more processors 602.

[0087] The memory 601 is used to store computer programs executed by the processor 602. The memory 601 may mainly include a program storage area and a data storage area. The program storage area may store the operating system and programs required to run instant messaging functions, etc.; the data storage area may store various instant messaging information and operation instruction sets, etc.

[0088] Memory 601 may be volatile memory, such as random-access memory (RAM); memory 601 may also be non-volatile memory, such as read-only memory, flash memory, hard disk drive (HDD), or solid-state drive (SSD); or memory 601 may be any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto. Memory 601 may be a combination of the above-described memories.

[0089] Processor 602 may include one or more central processing units (CPUs) or digital processing units, etc. Processor 602 is used to implement the above-described ethylene cracking apparatus optimization method when calling the computer program stored in memory 601.

[0090] The communication module 603 is used to communicate with the ethylene cracking unit and the industrial control system.

[0091] This application embodiment does not limit the specific connection medium between the memory 601, communication module 603, and processor 602 described above. This application embodiment... Figure 6 The memory 601 and the processor 602 are connected via a bus 604, and the bus 604 is in Figure 6 The diagram uses thick lines to describe the connections between other components; these are for illustrative purposes only and should not be considered limiting. The 604 bus can be divided into address bus, data bus, control bus, etc. For ease of description, Figure 6 It is described using only a thick line, but does not indicate that there is only one bus or one type of bus.

[0092] The memory 601 stores a computer storage medium, which stores computer-executable instructions. The computer-executable instructions are used to implement the ethylene cracking apparatus optimization method of the embodiments of this application, and the processor 602 is used to execute the ethylene cracking apparatus optimization method of the above embodiments.

[0093] Based on the same inventive concept, embodiments of this application also provide a storage medium storing a computer program that, when run on a computer, causes the computer to perform the steps in the ethylene cracking apparatus optimization method according to various exemplary embodiments of this application described above.

[0094] In some possible implementations, various aspects of the ethylene cracking unit optimization method provided in this application can also be implemented in the form of a computer program product, which includes a computer program that, when run on a computer device, causes the computer device to perform the steps in the ethylene cracking unit optimization method according to various exemplary embodiments of this application described above. For example, the computer device can perform the steps of the various embodiments.

[0095] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0096] The program product of the embodiments of this application may employ a portable compact disc read-only memory (CD-ROM) and include a computer program, and may run on a computer device. However, the program product of this application is not limited thereto. In this application, the readable storage medium may be any tangible medium that contains or stores a program, and the computer program included therein may be used by or in conjunction with a command execution system, apparatus, or device.

[0097] A readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying a readable computer program. This propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of sending, propagating, or transmitting a program for use by or in conjunction with a command execution system, apparatus, or device.

[0098] Computer programs contained on readable media may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0099] Computer programs for performing the operations of this application can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages.

[0100] It should be noted that although several units or sub-units of the device have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of this application, the features and functions of two or more units described above can be embodied in one unit. Conversely, the features and functions of one unit described above can be further divided and embodied by multiple units.

[0101] Furthermore, although the operations of the method of this application are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0102] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0103] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0104] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for optimizing an ethylene cracking unit, characterized in that, The method includes: The baseline operating data of the ethylene cracking unit is uniformly predicted and indexed to determine the key performance indicators of the ethylene cracking unit. The baseline operating data represents the multi-feed information, operating variable information and energy consumption information of the ethylene cracking unit, and the key performance indicators represent the unit efficiency indicators, carbon emission indicators and product output indicators of the ethylene cracking unit. A dual-objective optimization model is constructed based on the key performance indicators and preset device constraints, and optimization calculations are performed on the dual-objective optimization model to determine the optimal solution set of the ethylene cracking unit; the optimal solution set represents the non-dominated solution set among the device benefit indicators, carbon emission indicators, and product output indicators. Based on a preset mechanistic interpretation strategy, the optimized solution set is interpreted mechanistically to obtain the interpretation text corresponding to the optimized solution set. The interpretation text represents the operation variable adjustment logic corresponding to the optimized solution set. Based on the optimized solution set and the explanatory text, the ethylene cracking unit is optimized and adjusted.

2. The method as described in claim 1, characterized in that, The construction of a dual-objective optimization model based on the key performance indicators and preset device constraints includes: Based on the device benefit index, a first optimization objective function corresponding to the dual-objective optimization model is constructed; the first optimization objective function is used to maximize the device benefit index. Based on the carbon emission index, a second optimization objective function corresponding to the dual-objective optimization model is constructed; the second objective function is used to minimize the carbon emission index. The bi-objective optimization model is constructed based on the first optimization objective function, the second optimization objective function, and the device constraints.

3. The method as described in claim 1, characterized in that, The optimization calculation of the bi-objective optimization model to determine the optimal solution set of the ethylene cracking unit includes: Based on the baseline operating data, finite difference adjustment calculations are performed on the operating variables of each raw material to determine the marginal sensitivity of each operating variable. Based on each marginal sensitivity, target operational variables are selected from the operational variables; Based on a preset multi-objective evolutionary algorithm, and in combination with the device constraints and the objective operation variables, the bi-objective optimization model is solved to generate the optimized solution set.

4. The method as described in claim 3, characterized in that, The step of performing finite difference adjustment calculations on the operating variables of each raw material based on the baseline operating condition data to determine the marginal sensitivity of each operating variable includes: Apply positive and negative perturbations of preset magnitude to each operated variable to obtain the perturbed operated variables; The key performance indicators are updated based on the predicted operational variables after the perturbation. The marginal sensitivity is determined based on the updated key performance indicators.

5. The method as described in claim 1, characterized in that, The method based on a preset mechanistic interpretation strategy performs a mechanistic interpretation on the optimized solution set to obtain the interpretation text corresponding to the optimized solution set, including: For each representative solution in the optimized solution set, based on the adjustment direction of the operation variables corresponding to each representative solution and the corresponding marginal sensitivity value, the query request corresponding to each representative solution is determined; each representative solution represents a combination of operation variables that satisfy the device constraints and correspond to different device benefit indicators and carbon emission indicators. Based on each query request, the preset process knowledge base is searched to obtain the corresponding process knowledge entries; Based on a preset explanation template, the explanation text is generated by combining the process knowledge entries, the adjustment direction of the operational variables, and the marginal sensitivity.

6. The method as described in claim 1, characterized in that, The process of uniformly predicting and calculating the baseline operating data of the ethylene cracking unit to determine the key performance indicators of the ethylene cracking unit includes: Based on the actual operating data of the ethylene cracking unit, a fuel consumption prediction model, a yield prediction model, and a steam generation prediction model are constructed. Based on the fuel consumption prediction model, yield prediction model, and steam generation prediction model, the baseline operating data are predicted and calculated to obtain the corresponding prediction results; the prediction includes at least the fuel gas consumption, product yield, steam production, and steam consumption of the ethylene cracking unit. Based on the prediction results, the key performance indicators are calculated.

7. The method as described in claim 1, characterized in that, The constraints of the unit include: operation optimization constraints, furnace group load constraints, and production load constraints. The operation optimization constraints are determined based on the raw material properties, unit principle, and actual operating data of each feedstock in the ethylene cracking unit. The furnace group load constraints are determined based on the capacity parameters of the ethylene cracking unit, furnace tube thermal intensity limits, and furnace group operating load distribution relationships. The production load constraints are determined based on the load requirements of the separation and distillation unit.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

9. A computer storage medium storing computer program instructions thereon, characterized in that, When executed by a processor, the computer program instructions implement the steps of the method according to any one of claims 1 to 7.

10. A computer program product comprising computer program instructions, characterized in that, When executed by a processor, the computer program instructions implement the steps of the method according to any one of claims 1 to 7.

Images