Method for improving sealing performance of fluorine lining guide pipe based on intelligent optimization algorithm

Abstract

The invention discloses a fluorine lining conduit sealing performance improvement method based on an intelligent optimization algorithm, and the method comprises the following steps: collecting and preprocessing related data of a fluorine lining conduit, and generating a multi-dimensional input data set; establishing a hybrid digital twinborn model to obtain a sealing performance prediction result; constructing an improved water circulation algorithm optimization framework, and generating optimization control parameters; on the basis of an improved water circulation algorithm optimization framework, a search update result is formed; obtaining a candidate parameter sealing performance result and a next-generation candidate parameter set by using a hybrid digital twinborn model; the optimal parameter set is applied to the manufacturing and assembling process of the fluorine lining guide pipe, and a manufacturing and assembling result is generated; and obtaining an updated optimal parameter set based on a manufacturing and assembling result and data acquired in real time. According to the method, a mixed digital twinning and improved water circulation algorithm is adopted, and the sealing performance of the fluorine lining guide pipe is dynamically optimized.

Description

Technical Field

[0001] This invention relates to the field of fluoropolymer-lined conduit sealing optimization technology, and in particular to a method for improving the sealing performance of fluoropolymer-lined conduits based on intelligent optimization algorithms. Background Technology

[0002] Existing fluoropolymer-lined conduit sealing design and manufacturing primarily rely on finite element analysis, experimental data, and human experience for parameter determination. Common methods include sealing structure simulation calculations based on finite element contact mechanics, multi-objective optimization design based on response surfaces, and swarm intelligence algorithms such as genetic algorithms, particle swarm optimization, or standard water circulation algorithms for parameter optimization.

[0003] These methods can improve the rationality of sealing groove geometry, gasket compression ratio, and heat treatment process parameters to a certain extent, and can be combined with experiments to verify and predict leakage rate, critical failure pressure, and fatigue life. However, existing technologies generally suffer from long calculation cycles, poor adaptability to changes in operating conditions, and insufficient coupling between parameter optimization results and actual production environment.

[0004] Finite element models and traditional data-driven models often struggle to reflect dynamic changes during production and assembly processes, as well as temperature and pressure fluctuations during operation, leading to a decrease in prediction accuracy over time. Conventional genetic algorithms, particle swarm optimization algorithms, and standard water cycle algorithms have limited search efficiency and convergence speed when dealing with multi-objective, multi-constraint, and uncertain design spaces. Furthermore, they lack physical information constraints specific to the sealing failure mechanism, making them prone to local optima and low feasible solution ratios.

[0005] Meanwhile, most existing technologies remain at the single optimization stage of design and manufacturing, lacking a real-time feedback mechanism for manufacturing assembly results and operation monitoring data, and are unable to continuously correct models and process parameters during long-term equipment operation.

[0006] Therefore, how to provide a method for improving the sealing performance of fluorinated lined conduits based on intelligent optimization algorithms is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0007] One objective of this invention is to propose a method for improving the sealing performance of fluorinated lined conduits based on intelligent optimization algorithms. This invention comprehensively applies a hybrid digital twin model and an improved water circulation algorithm. By collecting multi-dimensional input data in real time and constructing a predictive model that integrates physical mechanisms and data-driven approaches, it can accurately predict leakage rate, critical failure pressure, and fatigue life. Guided by uncertainty, it performs dynamic optimization and iterative updates of the candidate parameter set, achieving full-process adaptive optimization of sealing groove geometry, gasket compression ratio, and heat treatment process parameters. This method can continuously collect temperature, pressure, and leakage monitoring data during production and operation and feed them back to the model, achieving closed-loop control of design, manufacturing, assembly, and operation. It possesses advantages such as high prediction accuracy, high parameter optimization efficiency, strong long-term operational stability, and good production consistency.

[0008] A method for improving the sealing performance of fluorinated lined conduits based on an intelligent optimization algorithm according to an embodiment of the present invention includes the following steps:

[0009] Collect relevant data on fluorine-lined catheters and preprocess them to generate a multi-dimensional input dataset.

[0010] A hybrid digital twin model is built based on a multi-dimensional input dataset, and is trained and updated in real time to obtain sealing performance prediction results;

[0011] An improved water circulation algorithm optimization framework is constructed based on the sealing performance prediction results. A multi-objective optimization function is set and a dynamic weight decomposition strategy is introduced to generate optimized control parameters.

[0012] In the improved water cycle algorithm optimization framework, the sealing performance prediction results guide the search strategy for rainfall and evaporation, and form the search update results.

[0013] The candidate parameter set is initialized using optimized control parameters and search update results, and then input into the hybrid digital twin model to obtain the sealing performance results of the candidate parameters and the next generation candidate parameter set;

[0014] The optimal parameter set is determined based on the sealing performance results of the candidate parameters and applied to the manufacturing and assembly process of fluoropolymer-lined conduits to generate manufacturing and assembly results.

[0015] Real-time data is collected during the production operation phase and input into the hybrid digital twin model along with the manufacturing and assembly results, outputting an updated set of optimal parameters.

[0016] Optionally, the process of collecting and preprocessing relevant data from the fluorinated liner to generate a multi-dimensional input dataset specifically includes:

[0017] By using geometric measurement devices, torque sensors, pressure sensors, and temperature sensors arranged on the fluoropolymer-lined conduit, the geometric structural parameters, gasket compression, assembly torque, fluoroplastic physical property parameters, and operating environment parameters of the fluoropolymer-lined conduit are collected synchronously.

[0018] The collected geometric parameters, gasket compression, assembly torque, fluoroplastic physical properties, and operating environment parameters are combined according to their respective time sequences to construct the original data matrix, and the corresponding time series is established with the collection time as the index.

[0019] The original data matrix in the time series is processed to unify the dimensions. A normalization method based on the mean and standard deviation of each measurement dimension is used to convert parameters with different dimensions and numerical ranges into dimensionless data and generate a normalized data matrix.

[0020] Based on the normalized data matrix, outlier determination is performed for each measurement dimension. By calculating the deviation of each data point from the mean of the corresponding dimension and comparing it with the preset threshold range, all data points that deviate from the mean by more than the threshold range are removed, and a data matrix after removing outlier data is generated.

[0021] Based on the data matrix after removing outliers, a timestamp corresponding to the collection time is added to each data set, and a multidimensional feature mapping is established according to the dimensions required for the sealing performance analysis of fluorinated lined conduits to obtain a multidimensional input dataset.

[0022] Optionally, the step of establishing a hybrid digital twin model based on a multi-dimensional input dataset, training and updating it in real time to obtain the sealing performance prediction result specifically includes:

[0023] The multi-dimensional input dataset is divided into historical experimental dataset and online sensor dataset. The historical experimental dataset is a matrix of experimental and simulation data accumulated over a long period of time, and the online sensor dataset is a matrix of online data collected in real time and preprocessed.

[0024] Based on historical test datasets, a finite element contact mechanics model and a fluid-structure interaction model were established to calculate the stress and fluid pressure changes of fluorinated lined conduits under different working conditions, and to obtain a physical mechanism feature matrix that includes the geometric contact pressure distribution, gasket compression stress field, and temperature and pressure coupled stress field.

[0025] The physical mechanism feature matrix and historical test dataset are used together as training input to construct a deep neural network model and set a loss function with the square error between the predicted value and the measured sealing performance index as the target. The network parameters are continuously updated iteratively through supervised learning until the loss function converges, and the parameters of the deep neural network model with initial prediction ability are obtained.

[0026] The online sensing dataset and the initial neural network model parameters are input into the established deep neural network model. Combined with the real-time calculation results of the finite element contact mechanics model and the fluid-structure interaction model, the network parameters are gradually updated in a gradient manner. This allows the neural network model to maintain dynamic adjustment of parameters and improve accuracy as the data input changes continuously, forming a real-time updated hybrid digital twin model.

[0027] The system utilizes a real-time updated hybrid digital twin model to perform inference calculations on a multi-dimensional input dataset, outputting a set of sealing performance prediction results that include predicted leakage rate, predicted critical failure pressure, uncertainty, and predicted fatigue life.

[0028] Optionally, the improvement of the water circulation algorithm optimization framework based on the sealing performance prediction results, the setting of a multi-objective optimization function and the introduction of a dynamic weight decomposition strategy to generate optimization control parameters specifically include:

[0029] Using the set of sealing performance prediction results as optimization input, a multi-objective optimization function is constructed, which includes four objectives: leakage rate, critical failure pressure, fatigue life, and manufacturing cost. Weight coefficients that can be dynamically adjusted with optimization iterations are set for leakage rate, critical failure pressure, fatigue life, and manufacturing cost. The parameter vectors of three types of candidate design schemes, namely sealing groove geometry, gasket compression ratio, and heat treatment process parameters, are included as variables in the multi-objective optimization function.

[0030] Based on the multi-objective optimization function, a set of constraints is established. The assembly torque range, the strength limit of fluoroplastic materials, and the manufacturing tolerance requirements are set as inequality constraints, while the geometric relationship and physical consistency of the sealing structure are set as equality constraints.

[0031] An adaptive penalty function is constructed based on the set of constraints. By introducing a penalty coefficient that is dynamically adjusted with the number of iterations into the comprehensive optimization objective function, a square-form penalty value is applied to candidate design schemes that violate inequality constraints such as assembly torque range, fluoroplastic material strength limit, and manufacturing tolerance requirements, as well as equality constraints that violate the geometric relationship and physical consistency of the sealing structure. The penalty function is then added to the multi-objective optimization function to form the comprehensive optimization objective function.

[0032] In the improved water cycle algorithm framework, based on the multi-objective optimization function, the set of constraints, and the comprehensive optimization objective function, an initial set of candidate design parameters is randomly generated within the allowable parameter range and regarded as a set of water droplets. The initial comprehensive optimization objective value of each candidate design is calculated. Based on the magnitude of the initial comprehensive optimization objective value, the candidate design schemes with higher comprehensive performance are selected to form a river set, and the remaining candidate design schemes are combined to form a stream set. A hierarchical structure between rivers, streams, and water droplets is established.

[0033] In the iterative process of the improved water cycle algorithm, guided by the uncertainty in the set of sealing performance prediction results, a global dominant search is performed on the candidate design schemes in the river set, a local follow-up search is performed on the candidate design schemes in the stream set, and evaporation and rainfall operations are triggered on the candidate design schemes with low comprehensive optimization objective values ​​to generate new candidate design schemes in the parameter space. At the same time, the parameter update step size is corrected by combining the contact stress gradient and thermal expansion coefficient.

[0034] The process continuously performs operations such as the collection of river collections, the migration of stream collections, and the generation of new candidate design schemes by evaporation and rainfall. After each iteration, the comprehensive optimization target value is calculated. When the difference between the comprehensive optimization target values ​​of two consecutive iterations is less than the preset convergence threshold or the set maximum number of iterations is reached, the iteration process ends, the final optimization control parameters are output, and the optimization control parameters are used as the input for the next step of search and update.

[0035] Optionally, in the improved water cycle algorithm optimization framework, the search strategy for rainfall and evaporation based on the sealing performance prediction results, to form the search update results, specifically includes:

[0036] Within the established multi-objective optimization framework of the improved water cycle algorithm, the uncertainty of the sealing performance prediction result set serves as the real-time search guide. Rainfall operations are performed on parameter space regions with high uncertainty to generate new candidate design schemes, while evaporation operations are performed on parameter space regions with low uncertainty to enhance the local search of existing candidate design schemes.

[0037] Calculate the contact stress gradient and thermal expansion coefficient of each candidate design scheme at the current iteration time. Then, based on the original iteration step size, substitute the contact stress gradient and thermal expansion coefficient into the step size correction formula for dynamic correction to obtain the corrected step size factor.

[0038] The parameter vector of each candidate design scheme is iteratively updated using the corrected step size factor. The parameter vector of the previous iteration step, the corrected step size factor, the contact stress gradient, the coefficient of thermal expansion, and the uncertainty of the sealing performance prediction result are substituted into the parameter update formula to calculate the parameter vector of the next iteration step.

[0039] After each iteration, based on the uncertainty in the latest comprehensive optimization target value and sealing performance prediction results, the parameter vectors of all candidate design schemes are re-evaluated and classified. Candidate design schemes with high comprehensive performance are assigned to the river set, and candidate design schemes with potential improvement space are assigned to the stream set. The distribution and proportion of each set are adjusted according to the classification results.

[0040] The process continuously executes uncertainty-guided rainfall and evaporation operations, step size corrections based on contact stress gradients and thermal expansion coefficients, and parameter iterative updates of candidate design schemes until the comprehensive optimization objective function meets the preset convergence conditions or reaches the set maximum number of iterations, and then outputs the search update results.

[0041] The search update results are used as input for the next step of generating a new generation of candidate design parameters. The search update results include the final optimal parameter vector of each candidate design, the corresponding comprehensive optimization target value, the distribution and classification information of the final river and stream set, and the uncertainty measurement results of the sealing performance prediction.

[0042] Optionally, the step of initializing the candidate parameter set using optimized control parameters and search update results, inputting it into the hybrid digital twin model, and obtaining the sealing performance results of the candidate parameters and the next-generation candidate parameter set specifically includes:

[0043] Based on the optimized control parameters and the search update results, an initial set of candidate parameters is randomly generated within the set parameter range, including three types of variables: sealing groove geometry, gasket compression rate, and heat treatment process parameters. Each candidate parameter set is assigned a unique number and an initial iteration step to construct an initial search solution set.

[0044] Each initial set of candidate parameters is input into a real-time updated hybrid digital twin model. The sealing performance prediction value of the candidate parameter set is calculated using the current deep neural network model parameters and physical mechanism feature matrix. The sealing performance prediction results, including the predicted leakage rate, the predicted critical failure pressure, and the predicted fatigue life, are output.

[0045] Substitute the sealing performance prediction results of each candidate parameter set into the comprehensive optimization objective function, and combine the weight coefficients of each objective of the multi-objective optimization function with the inequality constraints including the assembly torque range, the strength limit of fluoroplastic materials, and the manufacturing tolerance requirements, as well as the equality constraints of the geometric relationship and physical consistency of the sealing structure, to calculate and record the comprehensive optimization objective value of each candidate parameter set under the initial iteration.

[0046] Based on the comprehensive optimization target value, all candidate parameter sets are classified. The candidate parameter set with the highest comprehensive performance evaluation value is assigned to the river set, the candidate parameter set with the intermediate comprehensive performance evaluation value is assigned to the stream set, and the candidate parameter set with the lowest comprehensive performance evaluation value is subjected to evaporation operation. A new candidate parameter set is randomly generated in the parameter space to simulate rainfall operation, forming a hierarchical structure and a new candidate parameter distribution.

[0047] In each iteration, the modified step size factor is used to combine the candidate parameter vector of the previous iteration step with the uncertainty of the prediction results of contact stress gradient, thermal expansion coefficient and sealing performance. The new candidate parameter vector is calculated according to the parameter update formula, and all candidate parameter sets in the river set and stream set are updated synchronously.

[0048] Repeat the iterative process of predicting the sealing performance of the candidate parameter set, calculating the comprehensive optimization target, hierarchical classification and parameter updating, until the change of the comprehensive optimization target between two consecutive iterations is less than the preset convergence threshold or the set maximum number of iterations is reached, and output the final next-generation candidate parameter set.

[0049] Optionally, the step of determining the optimal parameter set based on the sealing performance results of candidate parameters and applying it to the manufacturing and assembly process of fluoropolymer-lined conduits to generate manufacturing and assembly results specifically includes:

[0050] After each iteration, the sealing performance results of the candidate parameters obtained in the current iteration are input into the comprehensive optimization objective function. Combining the inequality constraints such as the weight coefficients of each objective in the multi-objective optimization function, the assembly torque range, the strength limit of fluoroplastic materials, and manufacturing tolerance requirements, as well as the equality constraints of the geometric relationship and physical consistency of the sealing structure, the current comprehensive optimization objective value of each candidate parameter set is calculated. The current comprehensive optimization objective value is then compared with the comprehensive optimization objective value corresponding to the previous iteration to obtain the difference in comprehensive optimization objective values.

[0051] The convergence condition is determined based on the difference between the comprehensive optimization target values. When the difference between the comprehensive optimization target values ​​of all candidate parameter sets is less than the preset convergence accuracy threshold or the number of iterations reaches the set maximum number of iterations, the optimization process is determined to have converged.

[0052] After the optimization process is determined to be converged, the candidate parameter set with the best comprehensive optimization objective value is selected from all candidate parameter sets. The final values ​​of the sealing groove geometry, gasket compression ratio and heat treatment process parameters are extracted to form the optimal parameter set and recorded.

[0053] The optimal set of parameters is applied to the manufacturing and assembly process of fluorine-lined conduits, and production operations including heat treatment of the lining, precision machining of the sealing groove, and control of bolt preload are performed to form a manufacturing and assembly result. The manufacturing and assembly result includes finished product structural parameters, process execution parameters, key sealing performance indicators, and traceability information.

[0054] Optionally, the step of collecting real-time data during the production operation phase and inputting it along with the manufacturing and assembly results into the hybrid digital twin model to output an updated set of optimal parameters specifically includes:

[0055] After the fluorine-lined conduit enters the production and operation stage, temperature data, pressure data and leakage monitoring data are continuously collected by sensors arranged in the pipe body and connection parts during the operation. The collected operation monitoring data is then associated and stored with the optimal parameter set and process execution parameters recorded in the manufacturing and assembly results.

[0056] The operational monitoring data and the manufacturing assembly results are input into the hybrid digital twin model, and the parameters and physical mechanism characteristics of the deep neural network part in the hybrid digital twin model are adaptively updated.

[0057] The sealing performance was recalculated using the updated hybrid digital twin model, resulting in a sealing performance prediction that includes the latest predicted leakage rate, the latest predicted critical failure pressure, and the latest predicted fatigue life. The latest uncertainty in the sealing performance prediction results was also evaluated.

[0058] When the latest uncertainty exceeds the preset threshold, the improved water cycle algorithm is triggered to enter a new round of optimization iteration, and the existing candidate parameter set is regenerated and filtered.

[0059] In the new optimization iteration process, the candidate parameter set is grouped and updated according to the latest uncertainty, and the parameter vector of the candidate parameter set is dynamically adjusted in combination with the contact stress gradient, thermal expansion coefficient and the corrected step size factor, so that the parameter update direction is more in line with the sealing performance optimization requirements under real-time working conditions.

[0060] After the iteration reaches the preset convergence condition or the maximum number of iterations, the updated optimal parameter set is output and replaced by the original optimal parameter set for the production, manufacturing and operation maintenance of fluorinated lined catheters.

[0061] The beneficial effects of this invention are:

[0062] This invention constructs a hybrid digital twin model integrating finite element contact mechanics, fluid-structure interaction, and deep neural networks. This model enables real-time prediction and dynamic updating of leakage rate, critical failure pressure, and fatigue life of fluorinated lined conduits, allowing for continuous quantitative evaluation of sealing performance during the design, manufacturing, and operation phases. The hybrid digital twin model can receive and adaptively correct temperature, pressure, and leakage data from production assembly and in-service operation in real time, effectively solving the problems of traditional finite element analysis and single data-driven models failing to reflect real-time operating condition changes and exhibiting a decline in prediction accuracy over time.

[0063] The core innovation of this invention lies in proposing and applying an improved water cycle algorithm, which deeply integrates swarm search mechanisms with physical principles. The improved water cycle algorithm uses the uncertainty in the sealing performance prediction results as the search guide. During iteration, it performs rainfall operations to generate new candidate solutions in high-uncertainty regions and evaporation operations to strengthen the local search in low-uncertainty regions. It also incorporates contact stress gradient and thermal expansion coefficient into the step size correction model, achieving adaptive adjustment of candidate parameters towards the optimal solution. The algorithm simultaneously introduces dynamic weight decomposition and an adaptive penalty mechanism, balancing leakage rate minimization, critical failure pressure maximization, fatigue life extension, and manufacturing cost control in multi-objective, multi-constraint optimization. Compared with traditional genetic algorithms, particle swarm optimization, or standard water cycle algorithms, this method possesses higher global search capability and convergence speed under complex constraints, and improves the stability and robustness of feasible solutions.

[0064] This invention further establishes a closed-loop control process covering data acquisition, model prediction, parameter optimization, manufacturing assembly, and operational feedback. By continuously collecting real-time monitoring data during the production operation phase and triggering further iterations of the improved water circulation algorithm, an updated set of optimal parameters is output and used for process correction, achieving long-term dynamic optimization of sealing groove geometry, gasket compression ratio, and heat treatment process parameters. This closed-loop optimization mechanism ensures stable sealing performance and batch consistency of the fluoropolymer-lined conduit throughout its entire life cycle, providing a verifiable technical foundation for industrial continuous production. Attached Figure Description

[0065] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0066] Figure 1 This is a flowchart of a method for improving the sealing performance of fluorinated lined conduits based on intelligent optimization algorithms, as proposed in this invention.

[0067] Figure 2 This is a schematic diagram of the hybrid digital twin model construction and real-time updating in the method for improving the sealing performance of fluorinated lined conduits based on intelligent optimization algorithms proposed in this invention.

[0068] Figure 3 This is a flowchart illustrating the operation of the improved water circulation algorithm for the proposed method of enhancing the sealing performance of fluorinated lined conduits based on intelligent optimization algorithms during the iterative update of the candidate parameter set. Detailed Implementation

[0069] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0070] refer to Figure 1-3 A method for improving the sealing performance of fluorinated lined conduits based on intelligent optimization algorithms includes the following steps:

[0071] Collect relevant data on fluorine-lined catheters and preprocess them to generate a multi-dimensional input dataset.

[0072] A hybrid digital twin model is built based on a multi-dimensional input dataset, and is trained and updated in real time to obtain sealing performance prediction results;

[0073] An improved water circulation algorithm optimization framework is constructed based on the sealing performance prediction results. A multi-objective optimization function is set and a dynamic weight decomposition strategy is introduced to generate optimized control parameters.

[0074] In the improved water cycle algorithm optimization framework, the sealing performance prediction results guide the search strategy for rainfall and evaporation, and form the search update results.

[0075] The candidate parameter set is initialized using optimized control parameters and search update results, and then input into the hybrid digital twin model to obtain the sealing performance results of the candidate parameters and the next generation candidate parameter set;

[0076] The optimal parameter set is determined based on the sealing performance results of the candidate parameters and applied to the manufacturing and assembly process of fluoropolymer-lined conduits to generate manufacturing and assembly results.

[0077] Real-time data is collected during the production operation phase and input into the hybrid digital twin model along with the manufacturing and assembly results, outputting an updated set of optimal parameters.

[0078] In this embodiment, the process of collecting relevant data from the fluorinated liner catheter and preprocessing it to generate a multi-dimensional input dataset specifically includes:

[0079] By using geometric measurement devices, torque sensors, pressure sensors, and temperature sensors arranged on the fluoropolymer-lined conduit, the geometric structural parameters, gasket compression, assembly torque, fluoroplastic physical property parameters, and operating environment parameters of the fluoropolymer-lined conduit are collected synchronously.

[0080] The collected geometric parameters, gasket compression, assembly torque, fluoroplastic physical properties, and operating environment parameters are combined according to their respective time sequences to construct an original data matrix containing all measurement dimensions, and a corresponding time series is established with the collection time as the index.

[0081] The original data matrix in the time series is processed to unify the dimensions. A normalization method based on the mean and standard deviation of each measurement dimension is adopted to convert parameters with different dimensions and numerical ranges into dimensionless data within the same standardized range, thereby generating a normalized data matrix.

[0082] Based on the normalized data matrix, outlier determination is performed for each measurement dimension. By calculating the deviation of each data point from the mean of the corresponding dimension and comparing it with the preset threshold range, all data points that deviate from the mean by more than the threshold range are removed, and a data matrix after removing outlier data is generated.

[0083] Based on the data matrix after removing outliers, a timestamp corresponding to the collection time is added to each data set, and a multidimensional feature mapping is established according to the dimensions required for the sealing performance analysis of fluorinated lined conduits. The processed data is then organized into a complete multidimensional input dataset.

[0084] In this embodiment, the step of establishing a hybrid digital twin model based on a multi-dimensional input dataset, training and updating it in real time to obtain the sealing performance prediction result specifically includes:

[0085] The multi-dimensional input dataset is divided into historical experimental dataset and online sensor dataset. The historical experimental dataset is a matrix of experimental and simulation data accumulated over a long period of time, and the online sensor dataset is a matrix of online data collected in real time and preprocessed.

[0086] Based on historical test datasets, a finite element contact mechanics model and a fluid-structure interaction model were established to calculate the stress and fluid pressure changes of fluorinated lined conduits under different working conditions, and to obtain a physical mechanism feature matrix that includes the geometric contact pressure distribution, gasket compression stress field, and temperature and pressure coupled stress field.

[0087] The physical mechanism feature matrix and historical test dataset are used together as training input to construct a deep neural network model and set a loss function with the square error between the predicted value and the measured sealing performance index as the target. The network parameters are continuously updated iteratively through supervised learning until the loss function converges, and the parameters of the deep neural network model with initial prediction ability are obtained.

[0088] The online sensing dataset and initial neural network model parameters are input into the established deep neural network model. Combined with the real-time calculation results of the finite element contact mechanics model and the fluid-structure interaction model, the network parameters are gradually updated in gradient. This allows the neural network model to maintain dynamic adjustment of parameters and improve accuracy as the data input changes continuously, thus forming a hybrid digital twin model that can be updated in real time.

[0089] By using a real-time updated hybrid digital twin model to perform inference calculations on a multi-dimensional input dataset, the output includes a set of sealing performance prediction results containing predicted leakage rate, predicted critical failure pressure, uncertainty, and predicted fatigue life. This set of results serves as input data for constructing an improved water circulation algorithm optimization framework, providing real-time and accurate performance evaluation basis for optimization iterations.

[0090] In this embodiment, the improvement of the water circulation algorithm optimization framework based on the sealing performance prediction results, the setting of a multi-objective optimization function and the introduction of a dynamic weight decomposition strategy to generate optimization control parameters specifically include:

[0091] Using the set of sealing performance prediction results as optimization input, a multi-objective optimization function is constructed, which includes four objectives: leakage rate, critical failure pressure, fatigue life, and manufacturing cost. Weight coefficients that can be dynamically adjusted with optimization iterations are set for leakage rate, critical failure pressure, fatigue life, and manufacturing cost. The parameter vectors of three types of candidate design schemes, namely sealing groove geometry, gasket compression ratio, and heat treatment process parameters, are included as variables in the multi-objective optimization function to calculate the evaluation value of each candidate design scheme in terms of comprehensive performance.

[0092] Based on the multi-objective optimization function, a set of constraints is established. The assembly torque range, the strength limit of fluoroplastic materials, and the manufacturing tolerance requirements are set as inequality constraints, while the geometric relationship and physical consistency of the sealing structure are set as equality constraints. This ensures that the candidate design schemes meet the sealing performance requirements in terms of geometric dimensions, material strength, and assembly accuracy.

[0093] An adaptive penalty function is constructed based on the set of constraints. By introducing a penalty coefficient that is dynamically adjusted with the number of iterations into the comprehensive optimization objective function, a squared penalty value is applied to candidate design schemes that violate inequality constraints such as assembly torque range, fluoroplastic material strength limit, and manufacturing tolerance requirements, as well as equality constraints that violate the geometric relationship and physical consistency of the sealing structure. The penalty function is then added to the multi-objective optimization function to form the comprehensive optimization objective function, which is used to comprehensively evaluate the feasibility and performance of each group of candidate design schemes during iterative calculations.

[0094] In the improved water cycle algorithm framework, based on the multi-objective optimization function, the set of constraints, and the comprehensive optimization objective function, an initial set of candidate design parameters is randomly generated within the allowed parameter range and regarded as a set of water droplets. The initial comprehensive optimization objective value of each candidate design is calculated. Based on the magnitude of the initial comprehensive optimization objective value, the candidate design schemes with higher comprehensive performance are selected to form a river set, and the remaining candidate design schemes are combined to form a stream set. A hierarchical structure is established between rivers, streams, and water droplets to provide an initial distribution basis for performing parameter updates and global optimization in the search space.

[0095] In the iterative process of the improved water cycle algorithm, guided by the uncertainty in the set of sealing performance prediction results, a global dominant search is performed on candidate design schemes in the river set, and a local follow-up search is performed on candidate design schemes in the stream set. Evaporation and rainfall operations are triggered for candidate design schemes with low comprehensive optimization objective values ​​to generate new candidate design schemes in the parameter space. At the same time, the parameter update step size is corrected by combining contact stress gradient and thermal expansion coefficient to achieve dynamic adjustment of candidate design scheme parameters towards the direction of optimal comprehensive performance.

[0096] ;

[0097] in, For the first During the nth iteration The parameter vector of each candidate design scheme includes three types of parameters: sealing groove geometry, gasket compression ratio, and heat treatment process parameters. No. During the nth iteration The parameter vector of each candidate design scheme, For the first The step size factor for each iteration is used to control the magnitude of parameter updates. In time The contact stress gradient below, derived from calculations using the finite element contact mechanics model, is used to reflect the direction of force changes in the sealed structure. In time The coefficient of thermal expansion, derived from the thermal properties of fluoroplastic materials, is used to describe the effect of temperature changes on the dimensions of the sealing structure. For the first The uncertainty weighting coefficient for each iteration is used to control the proportion of uncertainty's influence in parameter updates. In time The uncertainty of the set of sealing performance prediction results described below is output by the hybrid digital twin model and is used to reflect the confidence level of the sealing performance prediction.

[0098] The process continuously performs operations such as the collection of river collections, the migration of stream collections, and the generation of new candidate design schemes by evaporation and rainfall. After each iteration, the comprehensive optimization target value is calculated. When the difference between the comprehensive optimization target values ​​of two consecutive iterations is less than the preset convergence threshold or the set maximum number of iterations is reached, the iteration process ends, the final optimization control parameters are output, and the optimization control parameters are used as the input for the next step of search and update.

[0099] In this embodiment, the step of guiding the search strategy for rainfall and evaporation based on the sealing performance prediction results and forming the search update results within the improved water cycle algorithm optimization framework specifically includes:

[0100] In the multi-objective optimization framework established by the improved water cycle algorithm, the uncertainty of the sealing performance prediction result set is used as the real-time search guide. Rainfall operation is performed on the parameter space region with high uncertainty to generate new candidate design schemes, and evaporation operation is performed on the parameter space region with low uncertainty to enhance the local search of existing candidate design schemes. Thus, the ratio of global search to local search of candidate design schemes is dynamically adjusted in each iteration.

[0101] Calculate the contact stress gradient and thermal expansion coefficient of each candidate design scheme at the current iteration time. Then, based on the original iteration step size, substitute the contact stress gradient and thermal expansion coefficient into the step size correction formula for dynamic correction, and obtain the corrected step size factor to guide the next parameter update.

[0102] The contact stress gradient is obtained by using a finite element contact mechanics model to determine the pressure distribution of the sealing groove and gasket contact surface under given geometric parameters and load conditions, and the partial derivative of the pressure field with respect to the parameter vector of the candidate design scheme is obtained to obtain the contact pressure change rate caused by each parameter.

[0103] The coefficient of thermal expansion is calculated by combining the fluoroplastic material database with the current temperature field. The thermal expansion characteristics of the material are determined based on the thermal deformation rate per unit initial length at the measured current temperature. The overall coefficient of thermal expansion is obtained by weighting by area or volume in the case of multiple material interfaces.

[0104] The parameter vector of each candidate design scheme is iteratively updated using the corrected step size factor. The parameter vector of the previous iteration step, the corrected step size factor, the contact stress gradient, the coefficient of thermal expansion, and the uncertainty of the sealing performance prediction result are substituted into the parameter update formula to calculate the parameter vector of the next iteration step, so as to realize the dynamic adjustment of the candidate design scheme in the direction of optimal comprehensive performance.

[0105] After each iteration, based on the uncertainty in the latest comprehensive optimization target value and sealing performance prediction results, the parameter vectors of all candidate design schemes are re-evaluated and classified. Candidate design schemes with high comprehensive performance are assigned to the river set, and candidate design schemes with potential improvement space are assigned to the stream set. The distribution and proportion of each set are adjusted according to the classification results to provide a dynamic grouping basis for the next round of rainfall, evaporation and parameter updates.

[0106] The process continuously executes uncertainty-guided rainfall and evaporation operations, step size corrections based on contact stress gradients and thermal expansion coefficients, and parameter iterative updates of candidate design schemes until the comprehensive optimization objective function meets the preset convergence conditions or reaches the set maximum number of iterations, and then outputs the search update results.

[0107] The search update results are used as input for the next step of generating a new generation of candidate design parameter sets. The search update results include the final optimal parameter vector of each candidate design (including sealing groove geometry, gasket compression ratio and heat treatment process parameters), the corresponding comprehensive optimization target value, the distribution classification information of the final river and stream set, and the uncertainty measurement results of the sealing performance prediction.

[0108] In this embodiment, the process of initializing the candidate parameter set using optimized control parameters and search update results, inputting it into the hybrid digital twin model, and obtaining the sealing performance results of the candidate parameters and the next-generation candidate parameter set specifically includes:

[0109] Based on the optimized control parameters and the search update results, an initial set of candidate parameters, including three types of variables such as sealing groove geometry, gasket compression ratio and heat treatment process parameters, is randomly generated within the set parameter range. Each candidate parameter set is assigned a unique number and an initial iteration step to construct an initial search solution set for sealing performance prediction and optimization iteration.

[0110] Each initial set of candidate parameters is input into a real-time updated hybrid digital twin model. The sealing performance prediction value of the candidate parameter set is calculated using the current deep neural network model parameters and physical mechanism feature matrix. The sealing performance prediction results, including the predicted leakage rate, the predicted critical failure pressure, and the predicted fatigue life, are output.

[0111] Substitute the sealing performance prediction results of each candidate parameter set into the comprehensive optimization objective function, and combine the weight coefficients of each objective of the multi-objective optimization function with the inequality constraints including the assembly torque range, the strength limit of fluoroplastic materials, and the manufacturing tolerance requirements, as well as the equality constraints of the geometric relationship and physical consistency of the sealing structure, to calculate and record the comprehensive optimization objective value of each candidate parameter set under the initial iteration.

[0112] Based on the comprehensive optimization objective value, all candidate parameter sets are classified. The candidate parameter set with the highest comprehensive performance evaluation value is assigned to the river set, and the candidate parameter set with the intermediate comprehensive performance evaluation value is assigned to the stream set. The candidate parameter set with the lowest comprehensive performance evaluation value is subjected to evaporation operation, and a new candidate parameter set is randomly generated in the parameter space to simulate rainfall operation, thereby forming a hierarchical structure and a new candidate parameter distribution for iterative search.

[0113] In each iteration, the modified step size factor is used to combine the candidate parameter vector of the previous iteration with the uncertainty of the contact stress gradient, thermal expansion coefficient and sealing performance prediction results. The new candidate parameter vector is calculated according to the parameter update formula. All candidate parameter sets in the river set and the stream set are updated synchronously to achieve continuous optimization of candidate parameters along the direction of optimal comprehensive performance.

[0114] Repeat the iterative process of predicting the sealing performance of the candidate parameter set, calculating the comprehensive optimization target, hierarchical classification and parameter updating, until the change of the comprehensive optimization target between two consecutive iterations is less than the preset convergence threshold or the set maximum number of iterations is reached, and output the final next-generation candidate parameter set.

[0115] In this embodiment, the step of determining the optimal parameter set based on the sealing performance results of candidate parameters and applying it to the manufacturing and assembly process of fluoropolymer-lined conduits to generate manufacturing and assembly results specifically includes:

[0116] After each iteration, the sealing performance results of the candidate parameters obtained in the current iteration are input into the comprehensive optimization objective function. Combining the inequality constraints such as the weight coefficients of each objective in the multi-objective optimization function, the assembly torque range, the strength limit of fluoroplastic materials, and manufacturing tolerance requirements, as well as the equality constraints of the geometric relationship and physical consistency of the sealing structure, the current comprehensive optimization objective value of each candidate parameter set is calculated. The current comprehensive optimization objective value is then compared with the comprehensive optimization objective value corresponding to the previous iteration to obtain the difference in comprehensive optimization objective values.

[0117] The convergence condition is determined based on the difference between the comprehensive optimization target values. When the difference between the comprehensive optimization target values ​​of all candidate parameter sets is less than the preset convergence accuracy threshold or the number of iterations reaches the set maximum number of iterations, the optimization process is determined to have converged.

[0118] After the optimization process is determined to be converged, the candidate parameter set with the best comprehensive optimization objective value is selected from all candidate parameter sets. The final values ​​of the sealing groove geometry, gasket compression ratio and heat treatment process parameters are extracted to form the optimal parameter set and recorded.

[0119] The optimal set of parameters is applied to the manufacturing and assembly process of fluoropolymer-lined conduits, including production operations such as liner heat treatment, sealing groove finishing, and bolt preload control, to form a manufacturing and assembly result. This result includes finished product structural parameters (verified sealing groove geometry, gasket compression ratio, and bolt preload), process execution parameters (heat treatment temperature, holding time, and cooling rate), key sealing performance indicators (actually measured contact stress distribution, leakage rate, critical failure pressure, and fatigue life), and traceability information (corresponding production time, equipment number, operator, and batch). This ensures the final manufacturing quality, sealing performance, and consistency of subsequent batch production of fluoropolymer-lined conduits.

[0120] In this embodiment, the step of collecting real-time data during the production operation phase and inputting it along with the manufacturing and assembly results into the hybrid digital twin model to output an updated set of optimal parameters specifically includes:

[0121] After the fluorine-lined conduit enters the production and operation stage, temperature data, pressure data and leakage monitoring data are continuously collected by sensors arranged in the pipe body and connection parts during the operation. The collected operation monitoring data is then associated and stored with the optimal parameter set and process execution parameters recorded in the manufacturing and assembly results.

[0122] The operation monitoring data and the manufacturing and assembly results are input into the hybrid digital twin model. The parameters and physical mechanism characteristics of the deep neural network part in the hybrid digital twin model are adaptively updated so that the hybrid digital twin model can reflect the changes in sealing performance under the current actual operating conditions.

[0123] The sealing performance was recalculated using the updated hybrid digital twin model, resulting in a sealing performance prediction that includes the latest predicted leakage rate, the latest predicted critical failure pressure, and the latest predicted fatigue life. The latest uncertainty in the sealing performance prediction results was also evaluated.

[0124] When the latest uncertainty exceeds the preset threshold, the improved water cycle algorithm is triggered to enter a new round of optimization iteration, and the existing candidate parameter set is regenerated and filtered.

[0125] In the new optimization iteration process, the candidate parameter set is grouped and updated according to the latest uncertainty, and the parameter vector of the candidate parameter set is dynamically adjusted in combination with the contact stress gradient, thermal expansion coefficient and the corrected step size factor, so that the parameter update direction is more in line with the sealing performance optimization requirements under real-time working conditions.

[0126] After the iteration reaches the preset convergence condition or the maximum number of iterations, the updated optimal parameter set is output and replaced by the original optimal parameter set for the production, manufacturing and operation and maintenance of fluoropolymer-lined conduits, providing the latest process parameters and control basis for continuously improving sealing performance.

[0127] Example 1:

[0128] To verify the feasibility of this invention in practice, it was applied to the optimization and renovation project of the sealing system of a fluorine-lined conduit in a chemical production enterprise. The fluorine-lined conduits in this enterprise operate under high temperature, high pressure, and corrosive media conditions for extended periods. Traditional design methods relying on finite element simulation and human experience are insufficient to cope with frequent temperature fluctuations and pressure pulsations, resulting in a high failure rate of seals at some pipe connections. On average, multiple leakage incidents occur annually, increasing maintenance costs and production safety risks. The existing production process's sealing parameters are typically updated over a period exceeding six months, and during long-term equipment operation, the leakage rate gradually increases to more than double the initial level, failing to guarantee the continuous stability of sealing performance.

[0129] In this embodiment, temperature, pressure, and leakage monitoring sensors are deployed to continuously collect data on the geometric parameters of the fluoropolymer-lined conduit, gasket compression, assembly torque, physical properties of the fluoroplastics, and operating environment. Real-time predictions are then performed using a hybrid digital twin model constructed according to this invention. This model integrates finite element contact mechanics, fluid-structure interaction, and deep neural network algorithms, allowing for parameter updates after each production run to continuously predict leakage rate, critical failure pressure, and fatigue life. Based on this, an improved water circulation algorithm proposed in this invention is used to perform uncertainty-driven rainfall and evaporation searches, combined with contact stress gradient and thermal expansion coefficient for step size correction, achieving dynamic iterative optimization of the sealing groove geometry, gasket compression rate, and heat treatment process parameters. After multiple iterations, the system can complete the entire parameter search and update within two hours without increasing computational resources, significantly shorter than the approximately twelve hours required by the original particle swarm optimization algorithm.

[0130] During three consecutive months of production verification, the average leakage rate of the fluorinated lined conduit optimized using this invention decreased from 0.18% to 0.05%, the critical failure pressure increased from 12.5 MPa to 14.8 MPa, and fatigue life improved by approximately 35%. Under the same raw materials and production rhythm, the annual seal failure rate decreased by approximately 70%, and maintenance costs decreased by more than 50%. In operation monitoring, the system can automatically capture abnormal fluctuations in temperature and pressure and promptly trigger iterative optimization to ensure that sealing parameters continuously match actual operating conditions, achieving closed-loop control of design, manufacturing, and operation. The application of this invention not only extends the maintenance cycle of the conduit but also improves the safety and economy of the production line, demonstrating the operability and long-term effectiveness of the proposed method in industrial environments.

[0131] Table 1. Key data record table for improving the sealing performance of fluorinated lined conduits according to the present invention.

[0132] Runtime Average leakage rate (%) Critical failure pressure (MPa) Average fatigue life (hours) Annual seal failure rate (%) Annual maintenance cost (ten thousand yuan) Optimize computation time (hours) Baseline before modification 0.18 12.5 8000 12.0 60 12 One month after the renovation 0.07 14.6 10300 4.5 30 2 Two months after the renovation 0.05 14.8 10850 3.8 28 2 3 months after the renovation 0.05 14.8 10900 3.5 27 2

[0133] As can be seen from Table 1 above, the present invention significantly improves several key indicators of the sealing performance of fluorinated lined conduits compared to the original method. Firstly, regarding the average leakage rate, the original value was 0.18%, while after applying the method of the present invention, it decreased to 0.07% within one month, further decreased to 0.05% after two months, and remained stable in the third month, with an overall reduction of nearly 72%. This result indicates that, through the synergistic optimization of the hybrid digital twin model and the improved water circulation algorithm, the sealing groove geometry, gasket compression ratio, and heat treatment process parameters can be matched to changes in operating conditions in real time during production, effectively suppressing leakage.

[0134] Secondly, regarding the critical failure pressure, the level before the modification was 12.5 MPa. After applying the method of this invention, it increased to 14.6 MPa within one month and stabilized at 14.8 MPa after two months, with an overall improvement of approximately 18%. This indicates that the present invention has achieved a stable effect in improving the pressure resistance of the sealing structure and can enhance the durability of the conduit under high-pressure conditions. At the same time, the average fatigue life increased from approximately 8,000 hours to 10,900 hours, an improvement of approximately 36%, indicating that the fatigue resistance of the sealing structure was significantly enhanced through dynamic optimization based on uncertainty and real-time step size correction.

[0135] In terms of operational reliability, this invention reduces the annual seal failure rate from 12.0% before the modification to 3.5% three months later, a decrease of approximately 71%. Maintenance costs are reduced from approximately 600,000 yuan per year to 270,000 yuan, a reduction of more than 50%. In addition, the optimization calculation time is shortened from approximately 12 hours in the traditional method to 2 hours, improving the response speed of production scheduling and process adjustments.

[0136] In summary, by employing the hybrid digital twin model and improved water circulation algorithm proposed in this invention, temperature, pressure, and leakage data can be continuously collected and iteratively optimized in actual production, achieving rapid convergence and long-term dynamic correction of sealing parameters. The results demonstrate that this invention not only achieves significant improvements in sealing performance but also effectively reduces maintenance costs and enhances operational safety and production efficiency, providing strong technical support for the long-term stable operation of fluorinated lined conduits.

[0137] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for improving the sealing performance of fluorinated lined conduits based on intelligent optimization algorithms, characterized in that, Includes the following steps: Collect relevant data on fluorine-lined catheters and preprocess them to generate a multi-dimensional input dataset. A hybrid digital twin model is built based on a multi-dimensional input dataset, and is trained and updated in real time to obtain sealing performance prediction results; An improved water circulation algorithm optimization framework is constructed based on the sealing performance prediction results. A multi-objective optimization function is set and a dynamic weight decomposition strategy is introduced to generate optimized control parameters. In the improved water cycle algorithm optimization framework, the sealing performance prediction results guide the search strategy for rainfall and evaporation, and form the search update results. The candidate parameter set is initialized using optimized control parameters and search update results, and then input into the hybrid digital twin model to obtain the sealing performance results of the candidate parameters and the next generation candidate parameter set; The optimal parameter set is determined based on the sealing performance results of the candidate parameters and applied to the manufacturing and assembly process of fluoropolymer-lined conduits to generate manufacturing and assembly results. Real-time data is collected during the production operation phase and input into the hybrid digital twin model along with the manufacturing and assembly results, outputting an updated set of optimal parameters.

2. The method for improving the sealing performance of fluorinated lined conduits based on intelligent optimization algorithms according to claim 1, characterized in that, The preprocessing includes dimensional unification, outlier removal, timestamp annotation, and multidimensional feature mapping.

3. The method for improving the sealing performance of fluorinated lined conduits based on intelligent optimization algorithms according to claim 1, characterized in that, The process of establishing a hybrid digital twin model based on a multi-dimensional input dataset, training and updating it in real time to obtain sealing performance prediction results specifically includes: The multi-dimensional input dataset is divided into historical test dataset and online sensing dataset. Based on the historical test dataset, a finite element contact mechanics model and a fluid-structure interaction model are established to calculate the stress and fluid pressure changes of the fluorinated liner under different working conditions and obtain the physical mechanism feature matrix. Based on the physical mechanism feature matrix and historical experimental dataset, a deep neural network model is constructed and a loss function is set. The network parameters are continuously updated iteratively through supervised learning methods to obtain the parameters of the deep neural network model. The online sensing dataset and the initial neural network model parameters are input into the deep neural network model, and the network parameters are gradually updated by combining the real-time calculation results of the finite element contact mechanics model and the fluid-structure interaction model to form a real-time updated hybrid digital twin model. The system utilizes a real-time updated hybrid digital twin model to perform inference calculations on a multi-dimensional input dataset, outputting a set of sealing performance prediction results that include predicted leakage rate, predicted critical failure pressure, uncertainty, and predicted fatigue life.

4. The method for improving the sealing performance of fluorinated lined conduits based on intelligent optimization algorithms according to claim 1, characterized in that, The improved water circulation algorithm optimization framework constructed based on the sealing performance prediction results, which sets a multi-objective optimization function and introduces a dynamic weight decomposition strategy to generate optimization control parameters specifically includes: Based on the set of sealing performance prediction results, a multi-objective optimization function is constructed that includes four objectives: leakage rate, critical failure pressure, fatigue life and manufacturing cost, and weight coefficients are set for each of the four objectives. Based on the multi-objective optimization function, a set of constraints is established. The assembly torque range, the strength limit of fluoroplastic materials and the manufacturing tolerance requirements are set as inequality constraints, and the geometric relationship and physical consistency of the sealing structure are set as equality constraints. An adaptive penalty function is constructed based on the set of constraints. By introducing a penalty coefficient that is dynamically adjusted with the number of iterations into the comprehensive optimization objective, a square-form penalty value is applied to the candidate design scheme that violates the constraints, and then added to the multi-objective optimization function to form the comprehensive optimization objective function. In the improved water cycle algorithm framework, based on the multi-objective optimization function, the set of constraints, and the comprehensive optimization objective function, an initial candidate design scheme parameter set is generated and regarded as a water droplet set; Calculate the initial comprehensive optimization target value for each candidate design scheme, and divide the candidate design schemes into river set and stream set based on the initial comprehensive optimization target value; In the iterative process of the improved water cycle algorithm, guided by the uncertainty in the set of sealing performance prediction results, a global dominant search is performed on the candidate design schemes in the river set, a local follow-up search is performed on the candidate design schemes in the stream set, and evaporation and rainfall operations are triggered on the candidate design schemes with low comprehensive optimization objective values ​​to generate new candidate design schemes. The process continuously executes the collection of river collections, the migration of stream collections, and the generation of new candidate design schemes through evaporation and rainfall. When the difference between the comprehensive optimization target values ​​of two consecutive iterations is less than the preset convergence threshold or the set maximum number of iterations is reached, the iteration process ends and the optimization control parameters are output.

5. The method for improving the sealing performance of fluorinated lined conduits based on intelligent optimization algorithms according to claim 1, characterized in that, In the improved water cycle algorithm optimization framework, the search strategy guided by the sealing performance prediction results to form the search update results specifically includes: In the improved water cycle algorithm optimization framework, the uncertainty of the sealing performance prediction result set is used as the real-time search guide. Rainfall operation is performed on the parameter space region of uncertainty to generate new candidate design schemes, and evaporation operation is performed on the parameter space region of uncertainty to enhance the local search of existing candidate design schemes. Calculate the contact stress gradient and thermal expansion coefficient of each candidate design scheme at the current iteration time, and dynamically correct the original iteration step size to obtain the corrected step size factor. The parameter vector of each candidate design scheme is iteratively updated using the corrected step size factor, and the parameter vector of the next iteration step is calculated. After each iteration, all candidate design schemes are re-evaluated and classified based on the latest comprehensive optimization objective value and uncertainty. The system continuously performs uncertainty-guided rainfall and evaporation operations, step size corrections based on contact stress gradients and thermal expansion coefficients, and parameter iterative updates of candidate design schemes until the comprehensive optimization objective meets the preset convergence conditions or reaches the set maximum number of iterations, and then outputs the search update results.

6. The method for improving the sealing performance of fluorinated lined conduits based on intelligent optimization algorithms according to claim 1, characterized in that, The process of initializing the candidate parameter set using optimized control parameters and search update results, inputting it into the hybrid digital twin model, and obtaining the sealing performance results of the candidate parameters and the next-generation candidate parameter set specifically includes: Based on the optimized control parameters and search update results, an initial set of candidate parameters is randomly generated within the set parameter range, and a unique number and initial iteration step are assigned to each set of candidate parameters. Each initial set of candidate parameters is input into a real-time updated hybrid digital twin model to calculate the predicted sealing performance value of the candidate parameter set and output the predicted sealing performance result. Substitute the sealing performance prediction results of each candidate parameter set into the comprehensive optimization objective function, combine with the multi-objective optimization function, calculate and record the comprehensive optimization objective value of each candidate parameter set under the initial iteration; Based on the comprehensive optimization target value, all candidate parameter sets are classified to obtain river set and stream set. The candidate parameter set with the lowest comprehensive performance evaluation value is subjected to evaporation operation and a new candidate parameter set is generated. In each iteration, the modified step size factor is used to combine the candidate parameter vector of the previous iteration step with the contact stress gradient, thermal expansion coefficient and uncertainty to calculate a new candidate parameter vector, and the entire candidate parameter set in the river set and stream set is updated synchronously. Repeat the iterative process of predicting the sealing performance of the candidate parameter set, calculating the comprehensive optimization target, hierarchical classification and parameter updating, until the change of the comprehensive optimization target between two consecutive iterations is less than the preset convergence threshold or the set maximum number of iterations is reached, and output the final next-generation candidate parameter set.

7. The method for improving the sealing performance of fluorinated lined conduits based on intelligent optimization algorithms according to claim 1, characterized in that, The process of determining the optimal parameter set based on the sealing performance results of candidate parameters and applying it to the manufacturing and assembly process of fluoropolymer-lined conduits to generate manufacturing and assembly results specifically includes: After each iteration, the sealing performance result of the candidate parameters obtained in the current iteration is input into the comprehensive optimization objective function to calculate the current comprehensive optimization objective value and compare it with the comprehensive optimization objective value corresponding to the previous iteration to obtain the difference of the comprehensive optimization objective value. When the difference between the comprehensive optimization target values ​​of all candidate parameter sets is less than the preset convergence accuracy threshold or the number of iterations reaches the set maximum number of iterations, the optimization process is determined to have converged. After determining that the optimization process has converged, the set of candidate parameters with the optimal comprehensive optimization objective value is selected from all candidate parameter sets to form the optimal parameter set. The optimal set of parameters is applied to the manufacturing and assembly process of fluorine-lined conduits to execute production operations and form the final manufacturing and assembly results.

8. The method for improving the sealing performance of fluorinated lined conduits based on intelligent optimization algorithms according to claim 1, characterized in that, The process of collecting real-time data during the production operation phase and inputting it along with the manufacturing and assembly results into the hybrid digital twin model to output an updated set of optimal parameters specifically includes: After the fluorine-lined conduit enters the production and operation stage, operation monitoring data is continuously collected by sensors placed in the conduit body and connection parts. The operation monitoring data and manufacturing assembly results are input into the hybrid digital twin model, and the parameters and physical mechanism characteristics of the deep neural network part in the hybrid digital twin model are adaptively updated. The sealing performance was recalculated using the updated hybrid digital twin model to obtain the latest sealing performance prediction results, and the latest uncertainty was evaluated. When the latest uncertainty exceeds a preset threshold, the improved water cycle algorithm is triggered to enter a new round of optimization iteration; In the new optimization iteration process, the candidate parameter set is grouped and updated according to the latest uncertainty, and the parameter vector of the candidate parameter set is dynamically adjusted. After the iteration reaches the preset convergence condition or the maximum number of iterations, the updated optimal parameter set is output and replaced by the original optimal parameter set for the production, manufacturing and operation maintenance of fluorinated lined catheters.

Images