Variable optimization method for an absorption stabilization system and control device

Abstract

The application discloses a kind of variable optimization method and control equipment of absorption stabilization system, it is related to industrial control technical field.The method includes: obtaining the current working condition data of absorption stabilization system;Current working condition data is input to working condition identification model and is identified processing, and current working condition mode is obtained;Based on current working condition mode and current working condition data, call net benefit prediction model and obtain net benefit prediction value;After determining variable value range, variable value range is input to process constraint relationship model, and process constraint parameter is obtained;Based on process constraint parameter, when target function meets net benefit prediction value, the system variable optimization solution set is determined;According to system variable optimization solution set, the process parameter of absorption stabilization system is adjusted.The application can be solved according to different working condition mode, optimization variable is not solved by neural network algorithm under the condition of satisfying process constraint condition, and reliable variable optimization result is obtained, and the net benefit of absorption stabilization system is improved.

Description

Technical Field

[0001] This invention relates to the field of industrial control technology, and in particular to a variable optimization method and control device for an absorber-stabilized system. Background Technology

[0002] Absorption stabilization systems are a process technology used in energy and chemical production processes such as petrochemicals and coal chemicals to separate gases and liquids in order to improve product quality and production efficiency. The main function of an absorption stabilization system is to absorb and stabilize various components in a gas mixture to achieve high-quality product output and efficient process control. The operation of an absorption stabilization system involves the optimization and control of multiple variables in the operation and production process, such as temperature, pressure, and flow rate. The optimization and control of these variables have a crucial impact on the stability and efficiency of the production process.

[0003] Traditional absorption stabilization systems have historically relied heavily on operator experience for process optimization and control. To reduce dependence on subjective experience and minimize instability and errors caused by manual operation, most related technologies employ neural network algorithms to solve for optimal control variables, thereby achieving automated control of the absorption stabilization system and increasing net production efficiency.

[0004] However, due to the limitations of computation speed, neural network algorithms often overlook the differences in net benefit output of absorption stabilization systems under different operating conditions, as well as the process constraints that optimization variables need to meet (such as the content of relevant chemical components, physical property limitations, etc.), resulting in unreliable variable optimization results and preventing absorption stabilization systems from producing ideal net benefits. Summary of the Invention

[0005] To address the aforementioned technical problems and deficiencies, the purpose of this invention is to provide a variable optimization method and control device for an absorption stabilization system. This method can solve for the optimization variables without using neural network algorithms, thereby obtaining reliable variable optimization results and improving the net efficiency of the absorption stabilization system, while meeting process constraints under different operating conditions.

[0006] To achieve the above objectives, in a first aspect, the present invention provides a variable optimization method for an absorption stabilization system, comprising: acquiring current operating condition data of the absorption stabilization system; inputting the current operating condition data into an operating condition identification model for identification processing to obtain a current operating condition mode, the current operating condition mode being used to characterize the current operating status of the absorption stabilization system; based on the current operating condition mode and the current operating condition data, calling a net benefit prediction model for predictive analysis to obtain a predicted net benefit value; after determining the range of values ​​for the system variables to be optimized in the objective function, inputting the range of values ​​for the variables into a process constraint relationship model to obtain process constraint parameters, the objective function being used to characterize the mapping relationship between the system net benefit and the system variables in the mechanism model of the absorption stabilization system; based on the process constraint parameters, determining the optimal solution set of the system variables when the objective function meets the predicted net benefit value; and adjusting the process parameters of the absorption stabilization system according to the optimal solution set of the system variables.

[0007] Through the above embodiments, after obtaining the current operating condition data of the absorption stabilization system, the current operating condition mode can be accurately identified through the operating condition identification model, which helps to understand the system's operating status in real time. Next, combining the current operating condition mode and operating condition data, a net benefit prediction model is used for predictive analysis, which can quickly obtain the predicted value of net benefit, providing an important basis for decision-making. After determining the value range of the system variables to be optimized, the process constraint parameters are obtained through the process constraint relationship model, further clarifying the mapping relationship between system variables and system net benefit. Based on these process constraint parameters, the optimized solution set of system variables when the objective function meets the predicted value of net benefit can be solved. Finally, the process parameters of the absorption stabilization system are adjusted according to these optimized solution sets. This achieves reliable variable optimization results without using neural network algorithms to solve for the optimized variables under different operating conditions and while satisfying process constraints, thereby improving the net benefit of the absorption stabilization system.

[0008] In one embodiment, the step of determining the optimal solution set of system variables when the objective function meets the predicted net benefit value based on the process constraint parameters includes: after determining the predicted net benefit value as the maximum value of the objective function, determining the key parameters of the flying mouse algorithm, including population size, number of iterations, flight speed, and step size; generating an initial population based on the variable value range and the process constraint parameters; and performing multiple iterative updates on the initial population according to the key parameters of the algorithm to obtain the optimal solution set of system variables when the objective function reaches its maximum value.

[0009] Through the above embodiments, this invention employs an improved flying mouse algorithm, demonstrating powerful search capabilities in global optimization problems, particularly in constraint handling, where it is extremely efficient and can quickly find the optimal solution. Compared to traditional neural network algorithms, the flying mouse algorithm in this embodiment outperforms traditional algorithms in terms of diversity preservation, convergence speed, and resource consumption, and can efficiently find the optimal solution.

[0010] In one embodiment, the step of adjusting the process parameters of the absorption stabilization system based on the optimal solution set of the system variables includes: conducting an environmental assessment on the optimal solution set of the system variables to obtain an environmental assessment result for the absorption stabilization system; determining the optimal parameters of the system variables from the optimal solution set of the system variables based on the environmental assessment result; and adjusting the process parameters of the absorption stabilization system based on the optimal parameters of the system variables.

[0011] By employing the above embodiments and conducting environmental assessments on the optimized solution set of system variables, a comprehensive understanding of the environmental impact of the absorption stabilization system under different parameter configurations can be achieved. Based on the environmental assessment results, optimized parameters for system variables that meet environmental standards can be identified, thereby ensuring that the system operates efficiently while minimizing adverse environmental impacts. Adjusting the process parameters of the absorption stabilization system based on these optimized parameters can further improve the system's environmental performance, achieving a win-win situation for both economic benefits and environmental protection. This process provides strong support for the sustainable development of enterprises and demonstrates the possibility of synergistic development between environmental protection and economic benefits.

[0012] In one embodiment, before obtaining the current operating condition data of the absorption stabilization system, the method further includes: obtaining multiple historical operating condition data of the absorption stabilization system; performing preliminary classification of the historical operating condition data using a clustering algorithm to obtain operating condition pattern clustering results; after the operating condition pattern clustering results are verified and evaluated, accurately labeling each of the historical operating condition data to obtain labeled data; determining effective feature data for distinguishing different operating condition patterns based on the labeled data; and training a supervised learning model based on the effective feature data to obtain an operating condition recognition model.

[0013] Using the above embodiments, historical operating condition data of the absorbing stability system is utilized to perform preliminary classification of the data through a clustering algorithm, resulting in clustering results for operating condition patterns. After verifying and evaluating the clustering results, the historical operating condition data is accurately labeled to form labeled data. Next, effective feature data distinguishing different operating condition patterns is determined based on the labeled data, and a supervised learning model is trained based on this feature data to finally obtain the operating condition identification model. This process improves the accuracy and efficiency of operating condition identification, providing important support for the optimized control of the absorbing stability system.

[0014] In one embodiment, after training the supervised learning model based on the effective feature data to obtain the working condition identification model, the method further includes: constructing a regression model using a machine learning algorithm; determining the identification result output by the working condition identification model as the target working condition mode; determining target working condition data in the historical working condition data according to the target working condition mode, the target working condition data including the operating parameters affecting net benefit in the absorption stabilization system and the maximum net benefit under the target working condition mode; and training the regression model based on the target working condition data to obtain a net benefit prediction model.

[0015] This embodiment utilizes machine learning algorithms to construct a regression model that can accurately predict the net benefits of an absorptive stable system. First, a target operating condition mode is determined using an operating condition identification model. Corresponding target operating condition data is then found in historical data, including operating parameters affecting net benefits and the maximum net benefit under that mode. Next, the regression model is trained based on this target operating condition data to obtain a net benefit prediction model. This model can accurately predict net benefits under different operating conditions, providing crucial decision support for system optimization and maximizing economic benefits.

[0016] In one embodiment, the regression model includes an XGBoost model, a random forest model, and a support vector machine model. The step of training the regression model based on the target operating condition data to obtain a net benefit prediction model includes: dividing the target operating condition data into a training set, a validation set, and a test set; training the XGBoost model, the random forest model, and the support vector machine model respectively based on the training set to obtain a hybrid ensemble model; validating and evaluating the hybrid ensemble model using the validation set to obtain an evaluation result; adjusting the parameters of the hybrid ensemble model based on the evaluation result; testing and evaluating the hybrid ensemble model using the test set to obtain the generalization ability of the ensemble model; and confirming that the training of the hybrid ensemble model is complete when the generalization ability meets the set criteria, thus obtaining the net benefit prediction model.

[0017] By employing the above embodiments, and dividing the target operating condition data into training, validation, and test sets, multiple machine learning models can be effectively trained and evaluated. The control device uses the training set to train XGBoost, Random Forest, and Support Vector Machine models respectively, thereby constructing a hybrid ensemble model. The validation set is used to validate and evaluate the hybrid ensemble model, obtaining preliminary performance evaluation results. Based on these results, the control device adjusts the parameters of the hybrid ensemble model to optimize its performance. Subsequently, the test set is used to test and evaluate the adjusted hybrid ensemble model to determine its generalization ability. When the generalization ability meets the set criteria, the control device confirms that the hybrid ensemble model training is complete, obtaining a model for predicting net benefits. This process ensures the accuracy and reliability of the model, providing strong support for the optimized control of absorbable stable systems.

[0018] In one embodiment, after the step of acquiring multiple historical operating condition data of the absorption stabilization system, the method further includes: preprocessing the historical operating condition data to obtain preprocessed historical operating condition data, wherein the preprocessing includes invalid data deletion processing, missing value imputation processing, smoothing processing, data type conversion, and normalization processing.

[0019] The above embodiments help to eliminate data noise, reduce data dimensionality, and unify data format, providing accurate and reliable basic data for subsequent data analysis and model training, and ensuring the stability and accuracy of the model.

[0020] Secondly, the present invention provides a control device, comprising: an acquisition module, an identification module, a prediction module, an input module, a determination module, and an adjustment module. The acquisition module acquires current operating condition data of an absorption stabilization system; the identification module inputs the current operating condition data into an operating condition identification model for identification processing to obtain a current operating condition mode, which characterizes the current operating status of the absorption stabilization system; the prediction module, based on the current operating condition mode and the current operating condition data, calls a net benefit prediction model for predictive analysis to obtain a predicted net benefit value; the input module, after determining the range of values ​​for the system variables to be optimized in the objective function, inputs the range of values ​​for these variables into a process constraint relationship model to obtain process constraint parameters, whereby the objective function characterizes the mapping relationship between the system net benefit and the system variables in the mechanism model of the absorption stabilization system; the determination module, based on the process constraint parameters, determines the optimal solution set of system variables when the objective function conforms to the predicted net benefit value; and the adjustment module adjusts the process parameters of the absorption stabilization system according to the optimal solution set of the system variables.

[0021] The control device of this invention can achieve the technical effects of the above method, which will not be elaborated here.

[0022] Thirdly, the present invention provides a control device, including a processor and a memory, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the above-described method is implemented.

[0023] The control device of this invention can achieve the technical effects of the above method, which will not be elaborated here.

[0024] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the above-described method.

[0025] The storage medium of this invention can achieve the technical effects of the above method, which will not be elaborated here.

[0026] One or more technical solutions provided in the embodiments of the present invention have at least the following technical effects or advantages:

[0027] 1. After acquiring the current operating condition data of the absorption stabilization system, the current operating mode can be accurately identified through the operating condition identification model, which helps to understand the system's operating status in real time. Then, combining the current operating mode and operating data, a net benefit prediction model is used for predictive analysis, which can quickly obtain the predicted value of net benefit, providing an important basis for decision-making. After determining the value range of the system variables to be optimized, the process constraint parameters are obtained through the process constraint relationship model, further clarifying the mapping relationship between system variables and system net benefit. Based on these process constraint parameters, the optimal solution set of system variables when the objective function meets the predicted net benefit value can be solved. Finally, the process parameters of the absorption stabilization system are adjusted according to these optimal solution sets. This achieves reliable variable optimization results without using neural network algorithms to solve for the optimization variables under different operating modes and while satisfying process constraints, thereby improving the net benefit of the absorption stabilization system.

[0028] 2. The improved flying mouse algorithm exhibits powerful search capabilities in global optimization problems, especially in constraint handling, where it is extremely efficient and can quickly find the optimal solution. Compared to traditional neural network algorithms, the flying mouse algorithm in this embodiment performs better in terms of diversity preservation, convergence speed, and resource consumption, and can efficiently find the optimal solution.

[0029] 3. Using historical operating condition data of the absorption stabilization system, a clustering algorithm is used to initially classify the data, obtaining clustering results for operating condition patterns. After verifying and evaluating the clustering results, the historical operating condition data is accurately labeled to form labeled data. Next, based on the labeled data, effective feature data distinguishing different operating condition patterns is determined. Then, a supervised learning model is trained based on these feature data, ultimately obtaining the operating condition identification model. This process improves the accuracy and efficiency of operating condition identification, providing important support for the optimized control of the absorption stabilization system. Attached Figure Description

[0030] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention. It is obvious that the drawings described below are merely some embodiments of the invention, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. In the drawings:

[0031] Figure 1 This is a flowchart of a variable optimization method for an absorbing stable system according to an embodiment of the present invention;

[0032] Figure 2This is a flowchart of another variable optimization method for an absorbing stable system according to an embodiment of the present invention;

[0033] Figure 3 This is a functional block diagram of a control device according to an embodiment of the present invention;

[0034] Figure 4 This is a schematic diagram of the computer system architecture of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0035] The terminology used in the following embodiments of the present invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in the specification and appended claims of the present invention, the singular expressions “a,” “an,” “the,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “or” as used in the present invention refers to and includes any or all possible combinations of one or more of the listed items. Hereinafter, the terms “first” and “second” are used for descriptive purposes only to distinguish technical features and should not be construed as implying relative importance or implicitly indicating the number of indicated technical features. In the description of the embodiments of the present invention, unless otherwise stated, “a plurality” means two or more. The embodiments of the present invention are described in detail below.

[0036] Absorption stabilization systems are generally used in the post-treatment process of catalytic cracking units in the chemical refining industry. Their purpose is to utilize absorption and distillation principles to separate rich gas and crude gasoline in the oil-gas separator at the top of the fractionation tower into dry gas (below C2), liquefied petroleum gas (C3, C4), and stabilized gasoline with acceptable vapor pressure. An absorption stabilization system typically includes an absorption tower, a desorption tower, a fractionation tower, and a stabilization tower.

[0037] The operation of an absorption stabilization system involves the optimization and control of multiple variables in the operation and production process, such as temperature, pressure, and flow rate. The optimized control of these variables affects the output of dry gas, liquefied gas, and condensate, as well as the consumption of electricity, water, and steam, ultimately impacting the net benefit output of the absorption stabilization system.

[0038] To improve the net efficiency of the absorption stabilization system and reduce reliance on manual operation, a chemical plant used a neural network algorithm to solve the multivariate optimization problem of the absorption stabilization system. The specific process is as follows:

[0039] First, data related to the absorption stabilization system are collected from the chemical plant's historical database.

[0040] Then, select an appropriate neural network model based on the complexity of the problem, such as a multilayer perceptron (MLP), convolutional neural network (CNN), or recurrent neural network (RNN). Simultaneously, determine the number of layers, the number of neurons in each layer, activation functions, and other hyperparameters of the neural network.

[0041] Next, the neural network model is trained and optimized.

[0042] Finally, the variables to be optimized are input into the model, and the model will output the net benefit prediction value and the corresponding optimization variable parameters.

[0043] In the above scheme, the neural network model requires a large amount of computing resources to solve the optimization variables. In order to improve the speed of calculation, the differences in the net benefit output of the absorption stabilization system under different operating conditions and the process constraints that the optimization variables need to meet (such as the content of relevant chemical components, physical property limitations, etc.) are often ignored. This leads to unreliable variable optimization results, making it impossible for the absorption stabilization system of the chemical plant to stably produce the ideal net benefit.

[0044] Therefore, this invention provides a variable optimization scheme for an absorption stabilization system. Under different operating conditions and while meeting process constraints, it can obtain reliable variable optimization results without resorting to neural network algorithms, thereby improving the net efficiency of the absorption stabilization system. This scheme obtains current operating data, uses an operating condition identification model to determine the current operating mode, and then combines this with a net efficiency prediction model to predict the system's net efficiency, providing a more accurate understanding of the system's operating status and potential optimization space. This scheme also considers process constraints, ensuring that the optimization results meet actual production requirements and avoiding the shortcomings of neglecting constraints in related technologies. By adjusting process parameters based on the optimized solution set of system variables, this scheme can steadily improve the net efficiency and production efficiency of the absorption stabilization system, bringing substantial economic and environmental benefits to industries such as energy and chemical engineering.

[0045] The following specific embodiments will illustrate in detail how the present invention achieves the technical effects of the above-mentioned solution, so as to solve the deficiencies and defects in related technologies.

[0046] This invention provides a variable optimization method for an absorbing stable system, such as... Figure 1 As shown, it includes the following steps:

[0047] Step 101: Obtain the current operating condition data of the absorption stabilization system.

[0048] Specifically, the control equipment uses various sensors and monitoring instruments to monitor and collect data on the operating status of the absorption stabilization system in real time. This data includes, but is not limited to, physical or chemical parameters such as temperature, pressure, flow rate, liquid level, and component concentration, which comprehensively reflect the current operating status of the system.

[0049] Step 102: Input the current working condition data into the working condition recognition model for recognition processing to obtain the current working condition mode.

[0050] The current operating condition mode characterizes the current operating status of the absorption stabilization system. An operating condition mode refers to the operating state and performance of the absorption stabilization system under set operating conditions. Different operating condition modes result in different net benefits for the absorption stabilization system; therefore, accurately identifying the operating condition mode is a crucial step in ensuring the efficient operation of the absorption stabilization system and improving its net benefits.

[0051] Specifically, operating modes can be categorized according to different devices and application scenarios. In some embodiments, operating modes may include the following:

[0052] Normal operating mode: Under normal conditions, the absorption stabilization system can effectively absorb and stabilize various components in the mixed gas, ensuring product quality and production stability.

[0053] Low-load mode: When the flow rate of feed gas entering the system decreases, the system may enter a low-load mode. This may lead to a decrease in the concentration of hydrocarbon gases in the absorber, requiring corresponding adjustments to the operating parameters.

[0054] High-load mode: Conversely, if the feed gas flow rate increases, the system may enter a high-load mode. In this mode, the concentration of hydrocarbon gases in the absorber increases, requiring corresponding adjustments to operating parameters to maintain stable system operation.

[0055] Optimize operating conditions: To improve the efficiency and production effectiveness of the absorption stabilization system, optimization control strategies can be adopted, such as adjusting operating parameters and changing raw material ratios. By optimizing operating conditions, product quality and yield can be further improved.

[0056] Fault-prone mode: When the absorption stabilization system malfunctions or experiences an accident, emergency measures are required, such as shutdown for maintenance and emission of harmful gases. In this mode, the normal operation of the system will be severely affected, and restoration needs to be completed as soon as possible.

[0057] Seasonal patterns: Under certain climatic conditions, such as summer and winter, the operating conditions of the system may be affected by temperature variations. Absorption stabilization systems may need to adjust operating parameters to adapt to different seasonal demands.

[0058] On the other hand, in the above steps, the working condition identification model is a supervised learning model, employing algorithms such as random forest and support vector machine. Specifically, the training method for the working condition identification model is as follows:

[0059] Cluster analysis involves applying clustering algorithms (such as K-means and DBSCAN) to perform preliminary classification of unlabeled data, analyze the characteristics of each cluster, and identify possible operating conditions.

[0060] Combining mechanistic model validation and data annotation, the clustering results are submitted to domain experts for validation using the mechanistic model. Experts evaluate the clustering results based on their experience and knowledge, confirming or adjusting the operational condition classifications. Based on the experts' feedback, the data is accurately labeled.

[0061] Based on the labeled data, select the most effective features to distinguish different operating conditions. Perform feature engineering, such as creating derived features and performing feature transformations.

[0062] Supervised learning models are trained using labeled data, such as random forests, support vector machines, or neural networks. Model parameters are then optimized using methods like cross-validation to ensure good generalization ability.

[0063] Model validation and evaluation involves assessing the model's performance on independent test datasets, including metrics such as accuracy and recall. In-depth analysis of the model's predictions is conducted to ensure they reflect the characteristics of real-world operating conditions.

[0064] Model deployment and real-time application: Deploy the trained model to the production environment for real-time condition identification. Monitor the model's performance in real time to ensure its accuracy and reliability.

[0065] In this step, the control equipment's operating condition identification enables a better understanding and optimization of the absorption stabilization system's operational status. The operating condition identification model takes current operating condition data as input and processes it to obtain the current operating condition pattern, which characterizes the current operational status of the absorption stabilization system. Through operating condition identification, the system's operating state can be understood more accurately, potential problems and optimization opportunities can be identified, thus providing an important foundation for subsequent net benefit prediction and system variable optimization. Simultaneously, operating condition identification also helps improve system stability and reliability, reduce operating costs, and increase production efficiency and product quality.

[0066] Step 103: Based on the current operating mode and the current operating data, call the net benefit prediction model to perform predictive analysis and obtain the net benefit prediction value.

[0067] The core function of the net benefit prediction model is to predict the maximum net benefit achievable by the absorption stability system under current operating conditions. It fully utilizes historical and real-time data, employing machine learning and regression analysis techniques to accurately capture the relationship between changes in operating data and net benefits under the current operating conditions. The net benefit prediction model not only considers the system's direct benefits but also fully assesses indirect benefits such as environmental factors and resource consumption, striving to provide decision-makers with comprehensive and accurate information. In this way, control equipment can anticipate potential benefit fluctuations, providing a scientific basis for subsequent adjustments to process parameters.

[0068] In this embodiment, the construction and training of the net benefit prediction model can be carried out by dividing historical data under various working conditions into training, testing, and validation sets. By leveraging the advantages of different models in handling different types of data, a hybrid ensemble model (including XGBoost, Random Forest, and Support Vector Machine models) can be established to improve overall prediction accuracy. Specifically, this includes the following:

[0069] The models are trained separately using current working condition data, including XGboost, Random Forest, and Support Vector Machine models, while optimizing their respective parameters to ensure optimal performance.

[0070] The overall model performance is optimized by adjusting the ensemble strategy and parameters. After the model is trained, the ensemble model is evaluated using a validation set and adjusted to improve performance. Finally, a final evaluation is performed on the test set to confirm the model's generalization ability.

[0071] After deploying the integrated model to the production environment, its performance is continuously monitored and adjusted as needed. An update threshold is set so that the model is automatically updated when a new operating condition occurs; users can also manually update the operating condition model.

[0072] Step 104: After determining the range of values ​​for the system variables to be optimized in the objective function, input the range of values ​​for these variables into the process constraint relationship model to obtain the process constraint parameters.

[0073] The objective function characterizes the mapping relationship between the net benefit of the absorption stabilization system and the system variables in the mechanistic model. The range of variable values ​​can be set according to actual needs, conforming to the current operating condition. Afterward, the process constraint relationship model can output accurate and appropriate process constraint parameters.

[0074] In this embodiment, the mechanistic model of the absorption stabilization system is a mathematical representation of the gas and liquid separation process. This model describes the operating rules of the absorption stabilization system, as well as the interrelationships and influences among the various variables within the system. By establishing a mechanistic model of the absorption stabilization system, we can gain a deeper understanding of the system's essence and internal mechanisms, providing a scientific basis for optimizing and controlling the system's operation.

[0075] The construction of mechanistic models requires consideration of various physical and chemical laws governing the absorption stabilization system, as well as the interactions between its components. For example, in petrochemical or coal chemical industries, absorption stabilization systems need to be designed according to different process requirements and feedstock characteristics to achieve optimal absorption and separation effects under specific conditions. Mechanistic models can simulate system performance under different operating conditions and optimize operating parameters to obtain the best process results.

[0076] In one embodiment, the objective function of the mechanistic model can be expressed as follows:

[0077] Z=Z1+Z2+Z3+Z4-C1-C2-C3-C4;

[0078] Where Z represents the net benefit of the absorption stabilization system, Z1 represents the dry gas benefit, Z2 represents the liquefied gas benefit, Z3 represents the gasoline benefit, Z4 represents the condensate benefit, C1 represents the electricity consumption cost, C2 represents the water consumption cost, C3 represents the medium-pressure steam consumption cost, and C4 represents the low-pressure steam consumption cost.

[0079] The multiple system variables of an absorption stable system can affect the magnitude of the values ​​of each variable in the objective function, and thus affect the magnitude of the net benefit.

[0080] On the other hand, process constraint parameters can describe the key quality characteristics of a product or the limitations of the process, such as the content of certain chemical components, physical property limitations, etc. In some examples, process constraint parameters may include the constraint range of the content of C3 or higher components in dry gas, the constraint range of the content of components with less than C3 in liquefied petroleum gas (LPG), the constraint range of the content of C5 or higher components in LPG, and the constraint range of the saturated vapor pressure of stable gasoline.

[0081] In this embodiment, historical data from each operating mode can be used to establish regression models between optimization variables and process constraint parameters. Alternatively, a hybrid ensemble model can be used, with a validation set used to evaluate the ensemble model, and an update threshold set. When the update threshold is exceeded, the model is automatically updated.

[0082] This embodiment determines the system variables related to the net benefit of the absorption stabilization system based on the mechanistic model, calculates the correlation between the relevant system variables and their impact on the net benefit, eliminates multicollinearity, and divides the relevant system variables into optimizable variables and non-optimizable variables. Among them, the optimizable variables are used as modeling variables to build a mathematical model between the optimizable variables and the net benefit, while the non-optimizable variables participate in data preprocessing to ensure the rationality of the data and the quality of the solution during the modeling and optimization process.

[0083] For example, optimizable variables include the supplemental absorbent flow rate, the reflux ratio at the top of the stabilizer, the bottom temperature of the desorber, the bottom temperature of the stabilizer, the top pressure of the stabilizer, and the top pressure of the reabsorption tower. Non-optimizable variables include the feedstock rate, the dry gas exit flow rate, the stabilized gasoline exit flow rate, the reflux flow rate at the top of the stabilizer, the condensate flow rate, the deethaned gasoline flow rate, the flow rates in absorber 1, absorber 2, absorber 3, absorber 4, fractionation tower 1, fractionation tower 2, the reboiler steam consumption at the bottom of the desorber, the extraction temperature in absorber 1, the return temperature in absorber 1, the extraction temperature in absorber 2, the return temperature in absorber 2, the extraction temperature in absorber 3, the return temperature in absorber 3, the extraction temperature in absorber 4, the return temperature in absorber 4, the extraction temperature in fractionation tower 2, the return temperature from the stabilizer in fractionation tower 2, the extraction temperature in fractionation tower 1, and the return temperature from the desorber in fractionation tower 1.

[0084] Step 105: Based on the process constraint parameters, determine the optimal solution set of system variables when the objective function meets the predicted net benefit value.

[0085] Specifically, this embodiment can employ algorithms such as the flying mouse algorithm, particle swarm optimization algorithm, genetic algorithm, simulated annealing algorithm, ant colony optimization algorithm, and artificial fish swarm algorithm to find the optimal solution. The flying mouse algorithm will be used as an example for detailed explanation below.

[0086] The Flying Squirrel Algorithm is a heuristic optimization algorithm inspired by the dynamic foraging strategy and efficient gliding movement of flying squirrels. In the Flying Squirrel Algorithm, the initial population is the starting point for evolutionary computation; it represents a randomly generated set of individuals. Each individual can be considered a potential solution in the problem space, and the entire initial population covers a broad region of the problem space. The Flying Squirrel Algorithm gradually explores and optimizes the solution space through fitness evaluation and evolutionary operations on the initial population. Specifically, the algorithm evaluates each individual based on a fitness function, selecting individuals with higher fitness for reproduction and evolution. Evolutionary operations can include crossover, mutation, and other operations to generate new offspring individuals. Through multiple rounds of iteration and evolution, the Flying Squirrel Algorithm can gradually approach the optimal solution or a near-optimal solution to the problem.

[0087] In some embodiments, this step specifically includes the following:

[0088] (1) The above net benefit prediction value is determined as the maximum value of the objective function.

[0089] (2) Determine the key parameters of the flying mouse algorithm, which include population size, number of iterations, flight speed, and step size. Population size refers to the size of the population searching for solutions in the flying mouse algorithm. The choice of population size affects both the algorithm's performance and convergence speed. Generally, a larger population size results in a larger solution space, but it also increases computational complexity and time cost.

[0090] Flight speed refers to the speed at which the flying mouse moves while searching the solution space. The choice of flight speed also affects the performance and convergence speed of the algorithm. If the flight speed is too fast, it may result in an excessively large solution space, increasing computational complexity and time cost; if the flight speed is too slow, it may result in an excessively small solution space, making it difficult to find the optimal solution.

[0091] Step size refers to the distance the flying mouse moves each time it searches the solution space. The choice of step size also affects the algorithm's performance and convergence speed. If the step size is too large, it may lead to an excessively large solution space, increasing computational complexity and time cost; if the step size is too small, it may lead to an excessively small solution space, making it difficult to find the optimal solution.

[0092] (3) Based on the range of values ​​of the variable and the process constraint parameters, an initial population is generated. The initial population is a set of solutions randomly generated within the given range of variables and process constraints. Then, using this set of solutions as the starting point, the population continuously evolves in subsequent iterations to find the optimal solution.

[0093] (4) Based on the key parameters of the algorithm, the initial population is iteratively updated multiple times to obtain the optimized solution set of system variables when the objective function reaches its maximum value. Specifically:

[0094] The initial population is iteratively updated based on a flight and search strategy using a flying mouse algorithm. This includes simulating the gliding behavior of a flying mouse and using information from the current optimal solution to guide the search direction.

[0095] In each iteration, the algorithm adjusts its search strategy based on the current population state and key parameters to explore the solution space more efficiently. Within each generation, superior individuals are selected for the next generation based on their fitness. This can be achieved through methods such as roulette wheel selection or tournament selection to ensure that superior genes are preserved and passed on to the next generation.

[0096] New offspring individuals are generated through crossover and mutation operations. Crossover combines the superior genes of parent individuals, while mutation introduces a degree of randomness, increasing population diversity and preventing the algorithm from getting trapped in local optima.

[0097] Repeat the above iterative process until the termination condition is met. The termination condition can be reaching the maximum number of iterations, finding the optimal solution that meets the accuracy requirements, or the population fitness no longer significantly improving, etc.

[0098] When the algorithm terminates, it outputs the optimal solution set found so far. In the context of an absorbing stable system, this means that a set of system variable values ​​has been found such that the maximum value of the objective function is equal to or close to the predicted net benefit value, while also satisfying the process constraints.

[0099] This embodiment employs an improved flying mouse algorithm, demonstrating powerful search capabilities in global optimization problems, particularly excelling in constraint handling and rapidly finding the optimal solution. Compared to traditional neural network algorithms, the flying mouse algorithm in this embodiment outperforms traditional algorithms in terms of diversity preservation, convergence speed, and resource consumption, efficiently finding the optimal solution.

[0100] Step 106: Adjust the process parameters of the absorption stabilization system based on the optimized solution set of the system variables.

[0101] Specifically, the control equipment analyzes the optimized solution set to extract variable values ​​related to the process parameters of the absorption stabilization system. These variables involve key process parameters such as temperature, pressure, flow rate, and concentration.

[0102] Based on these variable values, the control equipment then begins to adjust the actual process parameters of the absorption stabilization system. This typically involves communication and control with field equipment, such as adjusting valve openings, changing heater power, and adjusting pump speeds via a PLC (Programmable Logic Controller) or DCS (Distributed Control System).

[0103] During the adjustment of process parameters, the control equipment needs to monitor the system's operating status and performance indicators in real time. This can be achieved through communication with field sensors and actuators, ensuring that the system operates stably according to the requirements of the optimized solution set.

[0104] If a deviation is found between the actual operating state and the requirements of the optimized solution set, the control equipment will activate the closed-loop control mechanism, including adjusting the parameters of the control algorithm, recalculating the optimized solution set, or taking other corrective measures to ensure that the system always operates in the optimal state.

[0105] The control equipment can also record the process and results of adjusting process parameters for subsequent analysis and optimization. This data can be used to evaluate the effectiveness of the current control strategy and to make further adjustments or improvements as necessary.

[0106] Throughout the process, the control equipment needs to ensure the safety and reliability of the system. This may include functions such as fault diagnosis, preventing over-adjustment, and ensuring safe shutdown in emergency situations.

[0107] In one embodiment, this step may further include:

[0108] (1) An environmental assessment was conducted on the system variable optimization solution set to obtain the environmental assessment results for the absorption stabilization system.

[0109] Specifically, the control equipment needs to conduct an environmental assessment of the optimized solution set of this system variables. The assessment process may involve simulating or predicting the environmental impact corresponding to the values ​​of each system variable, such as the type and quantity of emissions, energy consumption, and resource utilization efficiency. By comparing the environmental impact of different solutions, the environmental performance of each solution can be evaluated. Alternatively, the optimized solution set of system variables can be input into an environmental assessment model to obtain the environmental assessment results. The training method for building the environmental assessment model is as follows:

[0110] Determine the assessment indicators: It is necessary to clearly define the indicators used to assess environmental performance. These indicators may include the types and quantities of emissions, energy consumption, resource utilization efficiency, etc. Selecting appropriate assessment indicators is key to establishing an effective assessment model.

[0111] Data Collection: To build and validate the evaluation model, relevant data needs to be collected. This may include historical data, real-time monitoring data, experimental data, etc. This data will be used to train the model and verify its accuracy.

[0112] Model Building: Based on the selected assessment indicators and collected data, an environmental assessment model can be built. The model can be based on statistical methods, machine learning algorithms, or other mathematical models. The purpose of the model is to take the optimized solution set of system variables as input and output the corresponding environmental assessment results.

[0113] Model validation and calibration: After the model is built, it needs to be validated and calibrated. This can be achieved by training and testing the model using known data to ensure its accuracy and reliability.

[0114] Inputting the optimal solution set of system variables: Once the model is built and validated, the optimal solution set of system variables can be input into the model. These solution sets represent different schemes for the values ​​of system variables, and through model calculations, the environmental assessment results corresponding to each scheme can be obtained.

[0115] Results Analysis and Interpretation: The environmental assessment results output by the model are analyzed and interpreted. This helps decision-makers understand the impact of different system variable values ​​on environmental performance and provides a basis for subsequent optimization and decision-making.

[0116] (2) Based on the environmental assessment results, determine the system variable optimization parameters from the system variable optimization solution set.

[0117] Specifically, through in-depth analysis and comparison of the evaluation results, the control equipment can clearly determine the degree and merits of the environmental impact of each system variable value scheme. Based on this, the control equipment will select those solutions with smaller environmental impact and compliance with environmental protection standards from the set of optimal solutions for the system variables, and determine them as the optimal parameters for the system variables.

[0118] (3) Adjust the process parameters of the absorption stabilization system according to the system variable optimization parameters.

[0119] When adjusting the process parameters of an absorption stabilization system, the control equipment operates based on optimized parameters of the determined system variables. These optimized parameters reflect the best system configuration under environmental and economic considerations, not only helping to improve the economic efficiency of the absorption stabilization system but also ensuring that the system operation meets environmental protection requirements, achieving a harmonious balance between economic benefits and environmental protection. Through precise commands and real-time feedback mechanisms, the control equipment works closely with field equipment to ensure that various process parameters, such as temperature, pressure, and flow rate, are precisely adjusted according to the requirements of the optimized parameters. This process aims to ensure that the absorption stabilization system can operate efficiently, stably, and environmentally friendly under the new parameter settings, thereby improving overall performance and optimizing resource utilization.

[0120] In this step, the control equipment can ensure that the economic benefits of the absorption and stabilization system are guaranteed while its operation meets environmental protection standards, achieving a win-win situation for both economy and environmental protection.

[0121] To improve the performance of the models and algorithms involved in this embodiment, the models and algorithms need to be trained and optimized before being deployed to a real production environment or during use. For example... Figure 2 As shown, the specific process is as follows:

[0122] Step 201: Establish a mechanistic model of the absorption stable system.

[0123] The mechanistic model has been explained in detail above and will not be repeated here.

[0124] Step 202: Obtain multiple historical operating condition data of the absorption stabilization system.

[0125] Step 203: Preprocess the historical operating condition data to obtain preprocessed historical operating condition data.

[0126] The preprocessing includes invalid data deletion, missing value imputation, smoothing, data type conversion, and normalization, which are used to clean and validate these historical operating data.

[0127] Specifically, in the data cleaning and validation process, the historical operating condition data dataset is first meticulously reviewed to identify and process invalid data. This includes identifying obviously unreasonable or erroneous data points, such as temperature or pressure readings exceeding physical limits, which may be due to instrument malfunction, entry errors, or transmission problems. Once such invalid data is identified, it can be either deleted or replaced with reasonable estimates. Next, for missing values ​​in the dataset, the pattern and cause of the missing values ​​are first assessed, and then an appropriate imputation method is selected based on the nature and characteristics of the data, choosing one of linear interpolation, multiple interpolation, or model-based prediction methods. The linear interpolation formula is:

[0128]

[0129] Where y1 and y2 are the values ​​of two known points, x1 and x2 are the positions of these two points, and x is the position where the value to be estimated is needed.

[0130] In addition, smoothing the data is also crucial to reduce random fluctuations and noise. This can be achieved through techniques such as moving averages and exponential smoothing, which helps to reveal the true trends and patterns in the data, especially important when conducting time series analysis.

[0131] For a time series Xt, the n-period moving average can be expressed as:

[0132]

[0133] Among them, MA t It is the moving average at time point t.

[0134] The formula for exponential smoothing is:

[0135] S t =αX t +(1-α)S t-1

[0136] Where α is the smoothing constant (0 < α ≤ 1), S t X is the smoothed value at time t. t It is the actual observed value at time t.

[0137] Simultaneously, outlier detection is necessary. Statistical methods or algorithms such as box plot analysis or isolated forests are used to identify outliers, and decisions are made based on their potential impact and causes to determine whether to adjust, replace, or delete these outliers. Next, data type conversion and normalization are performed, including converting categorical data to numerical data, timestamps to usable time features, and normalization or standardization to eliminate the influence of different units and scales, especially when preparing for subsequent machine learning modeling.

[0138] The most common method is min-max normalization, and its formula is:

[0139]

[0140] Where X is the original value, X' is the normalized value, and X' is the normalized value. min and X max These are the minimum and maximum values ​​of the variable, respectively.

[0141] Finally, in the final stage of data cleaning and validation, a comprehensive assessment of data quality is conducted to ensure the integrity, consistency, and usability of the dataset. This includes data integrity checks, consistency checks (such as the continuity of time series data), and an evaluation of the representativeness of the data.

[0142] Step 204: Perform preliminary classification of the historical operating condition data using a clustering algorithm to obtain the operating condition pattern clustering results.

[0143] Specifically, clustering algorithms (such as K-means and DBSCAN) are used to perform preliminary classification of unlabeled data, analyze the characteristics of each cluster, and identify possible operating conditions.

[0144] Step 205: After the clustering results of the working condition pattern are verified and evaluated, the working condition patterns of each historical working condition data are accurately labeled to obtain labeled data.

[0145] Specifically, combining mechanistic model validation and data annotation, the clustering results are submitted to domain experts for validation using the mechanistic model. Experts evaluate the clustering results based on their experience and knowledge, confirming or adjusting the operational condition classifications. Based on the experts' feedback, the data is accurately labeled, resulting in labeled data.

[0146] Step 206: Determine the effective feature data for distinguishing different operating conditions based on the labeled data.

[0147] Specifically, based on the labeled data, the most effective features for distinguishing different operating conditions are selected. Feature engineering can be performed, such as creating derived features and performing feature transformations.

[0148] Step 207: Train the supervised learning model based on the effective feature data to obtain the working condition recognition model.

[0149] Specifically, supervised learning models, such as random forests, support vector machines, or neural networks, are trained using labeled data. Model parameters are then optimized using methods such as cross-validation to ensure the model has good generalization ability.

[0150] Step 208: Evaluate the model's performance on an independent test dataset, including metrics such as accuracy and recall. Conduct in-depth analysis of the model's prediction results to ensure they conform to the characteristics of actual working conditions.

[0151] Step 209, Model Deployment and Real-Time Application: Deploy the trained model to the production environment for real-time operational condition recognition. Monitor the model's performance in real time to ensure its accuracy and reliability.

[0152] Step 210: Construct a regression model using machine learning algorithms.

[0153] Among them, the regression models include the XGboost model, the random forest model, and the support vector machine model.

[0154] Step 211: Determine the recognition result output by the working condition recognition model as the target working condition mode.

[0155] Step 212: Determine the target operating condition data from the historical operating condition data according to the target operating condition mode.

[0156] The target operating condition data includes the operating parameters affecting net benefits in the absorption stabilization system, and the maximum net benefits under the target operating condition mode.

[0157] Specifically, when determining the target operating condition data, the control equipment utilizes historical operating condition data as a reference. By analyzing and comparing data points in the historical operating condition data that match the target operating condition pattern, the control equipment can identify and extract historical operating condition data that conforms to the target operating condition pattern. This data becomes an important basis for adjusting the process parameters of the absorption stabilization system, ensuring that the system can achieve optimal operating status and performance under the target operating condition. The accurate analysis and application of historical data by the control equipment provides reliable data support for the optimization and stable operation of the system.

[0158] Step 213: Train the regression model based on the target working condition data to obtain the net benefit prediction model.

[0159] Specifically, using current operating data, we train XGboost, Random Forest, and Support Vector Machine models separately, while optimizing their respective parameters to ensure optimal performance.

[0160] In one embodiment, this step includes:

[0161] (1) Divide the target working condition data into training set, validation set and test set.

[0162] (2) Train the XGboost model, the random forest model and the support vector machine model respectively based on the training set to obtain the hybrid ensemble model.

[0163] (3) The hybrid ensemble model is validated and evaluated based on the validation set, and the evaluation results are obtained.

[0164] (4) Adjust the parameters of the hybrid integrated model based on the evaluation results.

[0165] (5) The hybrid ensemble model is tested and evaluated using the test set to obtain the generalization ability of the ensemble model.

[0166] Generalization ability refers to the model's ability to correctly understand and predict new, unseen data.

[0167] (6) When the generalization ability meets the set standard, the training of the hybrid ensemble model is confirmed to be completed, and the net benefit prediction model is obtained.

[0168] Step 214: Establish a process constraint relationship model.

[0169] Specifically, using historical data under the current operating condition, regression models are established between optimization variables and process constraint parameters. A hybrid ensemble model is also used, and the ensemble model is evaluated using a validation set. An update threshold is set, and the model is automatically updated when the threshold is exceeded. Specifically, the following steps are included:

[0170] (1) Clarify the process: First, it is necessary to understand the process of the absorption stabilization system in detail, including the working principle of each piece of equipment, the flow path of materials and energy, and the key operating parameters.

[0171] (2) Identify constraints: Based on the characteristics of the process, identify the key constraints that affect system performance and net benefits. These constraints may involve material balance, energy balance, equipment performance limitations, etc.

[0172] (3) Establish a mathematical model: Based on the defined constraints, establish a mathematical model of the process constraint relationship using mathematical methods. This model should be able to accurately describe the mathematical relationship between system variables and process constraints.

[0173] (4) Verification and Calibration: The accuracy and reliability of the process constraint relationship model are verified by comparing it with actual operating data. If deviations are found in the model, it needs to be calibrated and adjusted to improve its prediction accuracy.

[0174] (5) Continuous optimization: As system operation data accumulates, the process constraint relationship model can be updated and optimized regularly to adapt to changes in system performance and new process requirements.

[0175] Step 215: Solve for the optimal solution using the flying mouse algorithm.

[0176] Specifically, including:

[0177] (1) Improve the flying mouse algorithm to adapt to the current optimization objective by setting or adjusting key parameters of the flying mouse algorithm, such as population size, number of iterations, flight speed, step size, etc. Integrate an effective method for handling constraints, such as the penalty function method, to ensure that the algorithm complies with process constraints during the search process.

[0178] (2) The flying mouse algorithm is used to solve the problem. An initial population is randomly generated within a given range of variables and process constraints. Through the iterative process of the flying mouse algorithm, the population is continuously updated to find a solution that satisfies the process constraints and maximizes the net benefit. The algorithm performance is monitored during the iteration process, and parameters or strategies are continuously adjusted. Finally, after the algorithm converges, the optimal solution is extracted, which is the parameter combination that maximizes the net benefit under the condition of satisfying the process constraints.

[0179] (3) Result verification and application: Evaluate the optimal parameter combination found by the algorithm to ensure its effectiveness and feasibility in actual processes. Formulate operational recommendations or adjust the process flow based on the optimal solution.

[0180] Step 216: Record the optimization results and update the model.

[0181] This step specifically includes:

[0182] (1) Threshold update strategy: Based on the changes in optimization results, an accuracy or performance threshold is set. When the model's prediction accuracy or optimization result falls below this threshold, an update is triggered. Once it is detected that the improvement in optimization results is no longer significant, or the model's prediction accuracy drops to a certain level, the model is automatically retrained or its parameters are adjusted, including adding more training data or adjusting the model's structure and hyperparameters.

[0183] (2) A proactive update strategy, which does not rely on performance metrics but evaluates the model periodically. For example, a comprehensive check is performed on the model after a certain number of iterations or time intervals. During periodic evaluations, even if the model performance does not fall below a threshold, potential new trends or changes in data distribution are taken into account, and the model is proactively optimized and adjusted. This update aims to maintain the model's sensitivity and adaptability to new data and changing environments.

[0184] The variable optimization method for the absorption-stabilized system provided in this embodiment has the following advantages:

[0185] (1) Improve optimization efficiency

[0186] By integrating mechanistic models with machine learning (such as XGBoost), this invention can more effectively process complex data, improving the accuracy of prediction and optimization. This method can reveal complex relationships hidden within the data, thereby enhancing the efficiency and quality of the optimization process. One of the significant advantages of this invention in the field of industrial catalytic cracking technology is its highly efficient automated computing capabilities and rapid solution speed. By combining advanced mechanistic models and data-driven machine learning methods, this invention can automatically handle the entire process from data preprocessing to final optimization decision-making. This automation reduces reliance on manual operation and improves the overall efficiency and accuracy of the process.

[0187] (2) Enhance the adaptability to process constraints

[0188] This invention fully considers the process constraints of actual chemical production processes to ensure the practicality and operability of the optimization results. By precisely setting and managing the range of process parameters, this invention can ensure safety and stability while improving yield and efficiency.

[0189] (3) Global Optimization Solution Capability: Employing an improved flying mouse algorithm, this invention demonstrates powerful search capabilities in global optimization problems, particularly excelling in constraint handling and rapidly finding the optimal solution. Compared to traditional algorithms such as genetic algorithms and neural networks, the flying mouse algorithm outperforms traditional algorithms in terms of diversity preservation, convergence speed, and resource consumption, enabling it to efficiently find the optimal solution.

[0190] (4) Practicality and scalability

[0191] This invention is not only applicable to current chemical production problems, but also has good scalability. It can be applied to other industrial processes, especially those involving complex systems and scenarios requiring strict constraints.

[0192] (5) Enhanced decision support

[0193] By integrating data-driven models and optimization algorithms, this invention provides a powerful decision support tool. It helps engineers and managers make more accurate and efficient operational decisions, optimize resource allocation, and improve overall production efficiency.

[0194] (6) Highly efficient operating condition identification

[0195] By employing clustering algorithms, this invention first identifies different operating condition patterns in an unsupervised manner, which helps to discover potential process states and trends in complex datasets. Clustering algorithms can perform preliminary classification and pattern recognition of data without explicit labels. Combined with a supervised learning model, this invention further improves the accuracy of operating condition identification. Once the preliminary operating condition classification is obtained through clustering, the supervised learning model can use labeled data for precise pattern recognition and classification. This method enables the model to perform excellently in terms of accuracy and robustness. The combination of clustering and supervised learning fully leverages the advantages of data-driven methods, namely, discovering new insights through data exploration while achieving accurate prediction and classification through model learning. This method is particularly suitable for complex and multivariate industrial environments. The method of this invention can dynamically adapt to changes in process conditions. As new data is continuously input, the clustering and supervised learning models can be updated and adjusted to reflect new operating conditions and environmental changes.

[0196] This invention also provides a control device for executing the methods provided in the above embodiments, such as... Figure 3 As shown, it includes an acquisition module 301, an identification module 302, a prediction module 303, an input module 304, a determination module 305, and an adjustment module 306.

[0197] The acquisition module 301 is used to acquire the current operating condition data of the absorption stabilization system;

[0198] The identification module 302 is used to input the current operating condition data into the operating condition identification model for identification processing to obtain the current operating condition mode, which is used to characterize the current operating status of the absorption stabilization system.

[0199] The prediction module 303 is used to call the net benefit prediction model to perform predictive analysis based on the current working condition mode and the current working condition data, and obtain the net benefit prediction value.

[0200] The input module 304 is used to input the range of values ​​of the system variables to be optimized in the objective function into the process constraint relationship model after determining the range of values ​​of the variables, so as to obtain the process constraint parameters. The objective function is used to characterize the mapping relationship between the net benefit of the system and the system variables in the mechanism model of the absorption stabilization system.

[0201] The determination module 305 is used to determine the optimal solution set of system variables when the objective function meets the predicted net benefit value, based on the process constraint parameters.

[0202] The adjustment module 306 is used to adjust the process parameters of the absorption stabilization system based on the optimized solution set of the system variables.

[0203] In one embodiment, the determining module 305 is specifically used to: after determining the net benefit prediction value as the maximum value of the objective function, determine the key parameters of the flying mouse algorithm, including the population size, number of iterations, flight speed and step size; generate an initial population based on the range of variable values ​​and the process constraint parameters; and perform multiple iterative updates on the initial population according to the key parameters of the algorithm to obtain the optimized solution set of system variables when the objective function reaches its maximum value.

[0204] In one embodiment, the adjustment module 306 is specifically used for: performing an environmental assessment on the optimized solution set of the system variables to obtain an environmental assessment result for the absorption stabilization system; determining the optimized parameters of the system variables from the optimized solution set of the system variables based on the environmental assessment result; and adjusting the process parameters of the absorption stabilization system based on the optimized parameters of the system variables.

[0205] In one embodiment, the control device further includes a training module, which is used to: acquire multiple historical operating condition data of the absorption stabilization system; perform preliminary classification of the historical operating condition data using a clustering algorithm to obtain operating condition pattern clustering results; after the operating condition pattern clustering results are verified and evaluated, accurately label the operating condition patterns of each historical operating condition data to obtain labeled data; determine effective feature data for distinguishing different operating condition patterns based on the labeled data; and train a supervised learning model based on the effective feature data to obtain an operating condition recognition model.

[0206] In one embodiment, the training module can also be used to: construct a regression model using a machine learning algorithm; determine the identification result output by the working condition identification model as the target working condition mode; determine the target working condition data in the historical working condition data according to the target working condition mode, the target working condition data including the working parameters affecting the net benefit in the absorption stabilization system and the maximum net benefit under the target working condition mode; and train the regression model based on the target working condition data to obtain a net benefit prediction model.

[0207] In one embodiment, the training module can also be used to: divide the target working condition data into a training set, a validation set, and a test set; train the XGboost model, the random forest model, and the support vector machine model respectively based on the training set to obtain a hybrid ensemble model; validate and evaluate the hybrid ensemble model according to the validation set to obtain an evaluation result; adjust the parameters of the hybrid ensemble model based on the evaluation result; test and evaluate the hybrid ensemble model through the test set to obtain the generalization ability of the ensemble model; when the generalization ability meets the set standard, confirm that the training of the hybrid ensemble model is complete, and obtain the net benefit prediction model.

[0208] In one embodiment, the training module is further configured to: preprocess the historical operating condition data to obtain preprocessed historical operating condition data, wherein the preprocessing includes invalid data deletion, missing value imputation, smoothing, data type conversion, and normalization.

[0209] The control device provided in the embodiments of the present invention can execute the methods provided in the above embodiments, and therefore can also achieve the beneficial effects produced by the methods in the above embodiments, which will not be repeated here.

[0210] The control device in this embodiment of the invention is an electronic device. Figure 4 A schematic diagram of the architecture of an electronic device suitable for implementing embodiments of the present invention is shown.

[0211] It should be noted that, Figure 4 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0212] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by instructions (computer programs), or by instructions (computer programs) controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor. The electronic device of this embodiment includes a storage medium and a processor, wherein the storage medium stores multiple instructions that can be loaded by the processor to execute any step of the method provided in the embodiments of the present invention.

[0213] Specifically, the storage medium and the processor are electrically connected directly or indirectly to enable data transmission or interaction. For example, these components can be electrically connected to each other via one or more signal lines. The storage medium stores computer-executable instructions that implement data access control methods, including at least one software functional module that can be stored in the storage medium in the form of software or firmware. The processor executes various functional applications and data processing by running the software program and module stored in the storage medium. The storage medium can be, but is not limited to, Random Access Memory (RAM), Read-Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), etc. The storage medium stores the program, and the processor executes the program after receiving the execution instructions.

[0214] Furthermore, the software programs and modules within the aforementioned storage medium may also include an operating system, which may include various software components or drivers for managing system tasks (e.g., memory management, storage device control, power management, etc.) and can communicate with various hardware or software components to provide an operating environment for other software components. The processor can be an integrated circuit chip with signal processing capabilities. The aforementioned processor can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc., which can implement or execute the methods, steps, and logic flowcharts disclosed in this embodiment. The general-purpose processor can be a microprocessor or any conventional processor.

[0215] Since the instructions stored in the storage medium can execute the steps in any of the methods provided in the embodiments of the present invention, the beneficial effects of any of the methods provided in the embodiments of the present invention can be achieved, as detailed in the preceding embodiments, and will not be repeated here.

[0216] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A variable optimization method for an absorbing stable system, characterized in that, include: Obtain the current operating condition data of the absorption stabilization system; The current operating condition data is input into the operating condition identification model for identification processing to obtain the current operating condition mode, which is used to characterize the current operating status of the absorption and stabilization system. Based on the current operating condition mode and the current operating condition data, the net benefit prediction model is invoked to perform predictive analysis and obtain the net benefit prediction value. After determining the range of values ​​for the system variables to be optimized in the objective function, the range of values ​​is input into the process constraint relationship model to obtain the process constraint parameters. The objective function is used to characterize the mapping relationship between the net benefit of the system and the system variables in the mechanism model of the absorption stabilization system. Based on the process constraint parameters, determining the optimal solution set of system variables when the objective function meets the predicted net benefit value includes: after determining the predicted net benefit value as the maximum value of the objective function, determining the key parameters of the flying mouse algorithm, including population size, number of iterations, flight speed, and step size; generating an initial population based on the variable value range and the process constraint parameters; and iteratively updating the initial population according to the key parameters to obtain the optimal solution set of system variables when the objective function reaches its maximum value. The process parameters of the absorption stabilization system are adjusted based on the optimized solution set of the system variables.

2. The method according to claim 1, characterized in that, The step of adjusting the process parameters of the absorption stabilization system based on the optimized solution set of the system variables includes: An environmental assessment is performed on the optimized solution set of the system variables to obtain the environmental assessment results for the absorption stabilization system. Based on the environmental assessment results, the system variable optimization parameters are determined from the system variable optimization solution set; The process parameters of the absorption stabilization system are adjusted based on the system variable optimization parameters.

3. The method according to any one of claims 1 to 2, characterized in that, Before the step of acquiring the current operating condition data of the absorption stabilization system, the method further includes: Acquire multiple historical operating condition data of the absorption stabilization system; The historical operating condition data is initially classified using a clustering algorithm to obtain the operating condition pattern clustering results. After the clustering results of the working condition patterns are verified and evaluated, the working condition patterns of each of the historical working condition data are accurately labeled to obtain labeled data; Based on the labeled data, determine the effective feature data used to distinguish different operating conditions; The supervised learning model is trained based on the effective feature data to obtain the working condition recognition model.

4. The method according to claim 3, characterized in that, After the step of training the supervised learning model based on the effective feature data to obtain the working condition identification model, the method further includes: Build a regression model using machine learning algorithms; The identification result output by the working condition identification model is determined as the target working condition mode; The target operating condition data is determined from the historical operating condition data according to the target operating condition mode. The target operating condition data includes the operating parameters that affect the net benefit in the absorption stabilization system, and the maximum net benefit under the target operating condition mode. The regression model is trained based on the target operating condition data to obtain a net benefit prediction model.

5. The method according to claim 4, characterized in that, The regression model includes the XGBoost model, the random forest model, and the support vector machine model; the step of training the regression model based on the target working condition data to obtain the net benefit prediction model includes: The target operating condition data is divided into a training set, a validation set, and a test set; The XGboost model, the random forest model, and the support vector machine model are trained separately based on the training set to obtain a hybrid ensemble model; The hybrid ensemble model is validated and evaluated based on the validation set to obtain the evaluation results; The parameters of the hybrid ensemble model are adjusted based on the evaluation results; The hybrid ensemble model is tested and evaluated using the test set to obtain its generalization ability. When the generalization ability meets the set criteria, the training of the hybrid ensemble model is confirmed to be complete, and the net benefit prediction model is obtained.

6. The method according to claim 3, characterized in that, Following the step of acquiring multiple historical operating condition data of the absorption stabilization system, the method further includes: The historical operating condition data is preprocessed to obtain preprocessed historical operating condition data. The preprocessing includes invalid data deletion, missing value imputation, smoothing, data type conversion, and normalization.

7. A control device, characterized in that, include: The acquisition module is used to acquire the current operating condition data of the absorption stabilization system; The identification module is used to input the current operating condition data into the operating condition identification model for identification processing to obtain the current operating condition mode, which is used to characterize the current operating status of the absorption and stabilization system. The prediction module is used to call the net benefit prediction model to perform predictive analysis based on the current operating mode and the current operating data, and obtain the net benefit prediction value. The input module is used to input the range of values ​​of the system variables to be optimized in the objective function into the process constraint relationship model to obtain process constraint parameters. The objective function is used to characterize the mapping relationship between the net benefit of the system and the system variables in the mechanism model of the absorption stabilization system. The determination module is used to determine the optimal solution set of system variables when the objective function meets the predicted net benefit value based on the process constraint parameters. This includes: after determining the predicted net benefit value as the maximum value of the objective function, determining the key parameters of the flying mouse algorithm, including population size, number of iterations, flight speed, and step size; generating an initial population based on the variable value range and the process constraint parameters; and iteratively updating the initial population multiple times according to the key parameters to obtain the optimal solution set of system variables when the objective function reaches its maximum value. The adjustment module is used to adjust the process parameters of the absorption stabilization system based on the optimized solution set of the system variables.

8. A control device, characterized in that, It includes a processor and a memory, wherein the memory stores a computer program, which, when executed by the processor, implements the method of any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 6.

Images