A method and device for optimizing a diagonal prestressed airport pavement tension strategy

Abstract

The present application relates to a kind of oblique prestressed airport pavement tension strategy optimization method and device, belong to prestressed construction technical field, including the following steps:1: obtain engineering input parameter;2: establish parameterized finite element model;3: the tension construction process is constructed as sequential decision problem;4: construct reinforcement learning environment;5: in each decision step of reinforcement learning environment, select step-by-step tension type action;6: obtain the structural response data after the current decision step;7: carry out constraint check and reward calculation, determine the tension strategy satisfying constraint;8: repeat the execution step 5 to 7, obtain all candidate tension strategy;9: output the tension strategy satisfying structural safety constraint and construction constraint condition and corresponding structural response check result.The present application can make reinforcement learning training process directly according to finite element structural response decision-making, to improve the safety of tension strategy, engineering interpretability and applicability.

Description

Technical Field

[0001] This invention belongs to the field of prestressed construction technology, specifically relating to an optimization method for tensioning inclined prestressed airport pavement and an optimization device for tensioning inclined prestressed airport pavement. Background Technology

[0002] Airport pavement slabs need to withstand aircraft wheel loads, temperature gradients, differences in base support, construction residual stress, and repeated loads during operation over long periods. For airport pavement structures with high load-bearing requirements, large slab dimensions, or those requiring improved crack resistance, prestressing technology can improve the stress state of the pavement slab, reduce the risk of tensile stress concentration, and improve structural durability. Inclined prestressed airport pavement slabs are a pavement structure in which prestressing tendons are arranged obliquely relative to the slab edge or the direction of main load transfer. Their prestress transfer path differs from ordinary orthogonal reinforcement methods, and the interaction between prestressing tendon groups is more pronounced.

[0003] During the tensioning construction of inclined prestressed runway slabs, prestressing tendons are typically tensioned in groups, and multi-stage tensioning methods can be employed. Each selection of which group of prestressing tendons to tension, the target tension level, or the magnitude of the tension increment will alter the stress field, displacement field, bottom contact state, stress state of the prestressing tendons, and local stress state in the anchorage zone. Due to the inclined arrangement, different groups of prestressing tendons are spatially coupled, and different tensioning sequences and staged tensioning methods may lead to different stress peaks, edge warping displacements, and local stress concentrations in the anchorage zone at different construction stages.

[0004] Existing tensioning strategies typically rely on engineering experience, the principle of symmetrical tensioning, or a small number of finite element simulations. While these methods have a certain degree of engineering feasibility, they are difficult to systematically consider the stage-by-stage coupling effect between the inclined prestressing tendon groups during the gradual tensioning process. They are also difficult to evaluate the local adverse responses in all construction steps. Furthermore, if the finite element simulation directly exhaustively enumerates all prestressing tendon group tensioning sequences and tensioning level combinations, the number of calculations will increase rapidly with the number of prestressing tendon groups and tensioning levels, making it difficult to obtain a tensioning scheme that balances safety and construction efficiency throughout the entire process within a controllable strategy space.

[0005] Furthermore, existing tensioning strategy optimizations often focus only on the final stress state or a single safety indicator, lacking a unified evaluation of stress safety, stress uniformity, displacement control, anchorage zone safety, and process efficiency throughout the entire construction process. For strategies generated by intelligent algorithms, if the state and constraints cannot be updated directly based on the finite element solution results at each decision step, inconsistencies between the training path and the actual construction stress path can easily arise.

[0006] Therefore, there is an urgent need for a tensioning strategy that can meet the multi-objective constraints of the entire construction process and has engineering interpretability. Summary of the Invention

[0007] The purpose of this invention is to address the problems existing in the prior art by providing an optimization method for the tensioning strategy of inclined prestressed airport pavement. This method solves the problems that existing inclined prestressed airport pavement tensioning strategies mainly rely on experience, have high computational costs for full-combination finite element search, have difficulty accurately representing the structural response of the construction step with simplified models, have a disconnect between reinforcement learning state transition and the actual finite element stress process, and have difficulty taking into account the multi-objective constraints of the entire construction process.

[0008] The technical solution is as follows: An optimization method for tensioning oblique prestressed airport pavement includes the following steps: Step 1: Obtain the engineering input parameters for the inclined prestressed airport pavement slab. The engineering input parameters include pavement slab parameters, base course parameters, prestressed tendon group parameters, tensioning level parameters, boundary conditions, and construction constraints. Step 2: Based on the engineering input parameters, establish a parametric finite element model of the inclined prestressed airport runway slab. The parametric finite element model is used to simulate the stress field, displacement field, prestressing tendon stress state, and anchorage zone stress state of the slab under different tensioning construction steps. Step 3: The tensioning construction process of the inclined prestressed airport runway slab is constructed as a sequential decision problem, wherein the actions of each decision step include selecting the prestressed tendon group to be tensioned and selecting the tensioning target level or tensioning increment. Step 4: Construct a reinforcement learning environment, which includes a state space, a progressively stretched action space, a state transition method, a constraint verification method, and a reward function; Step 5: In each decision step of the reinforcement learning environment, the reinforcement learning agent selects a step-by-step tensioning action based on the current tensioning state. The step-by-step tensioning action includes selecting the next prestressed tendon group to be tensioned and selecting the tensioning target level or tensioning increment of the prestressed tendon group to be tensioned. Step 6: Accumulate the progressive tensioning actions selected up to the current decision step into a staged tensioning strategy. Convert the staged tensioning strategy into a finite element analysis step sequence. Each analysis step in the finite element analysis step sequence corresponds to an accumulated progressive tensioning action. Then, call the finite element solver through the script interface. The parameterized finite element model is regenerated according to the staged tensioning strategy, or the calculation continues based on the finite element calculation results of the previous decision step by restarting the analysis, inheriting the state file, or adding analysis steps. After the analysis job is completed, read the output database to obtain the structural response data after the current decision step. Step 7: Extract the structural response data after the current decision step from the calculation results of the finite element solver, perform constraint verification and reward calculation based on the structural response data, and update the next state of the reinforcement learning environment. When the structural response corresponding to any decision step does not meet the structural safety constraints, apply constraint penalties to the current step-by-step tensioning action and / or terminate the current training round. When all tensioning construction steps in the entire process meet the structural safety constraints, determine the corresponding candidate tensioning strategy as the tensioning strategy that meets the constraints. Step 8: Repeat steps 5 to 7 until the tension completion criterion or the training round termination criterion is met to obtain all candidate tensioning strategies; Step 9: Verify the full-process finite element structural response corresponding to the candidate tensioning strategy, and output the tensioning strategy that satisfies the structural safety constraints and construction constraints, as well as the corresponding structural response verification results. The tensioning strategy is the optimized inclined prestressed airport pavement tensioning strategy, including the tensioning sequence of prestressed tendon groups and the tensioning target level or tensioning increment corresponding to each prestressed tendon group.

[0009] Furthermore, the pavement panel parameters mentioned in step 1 include at least one of the following: pavement panel length, pavement panel width, pavement panel thickness, concrete elastic modulus, concrete Poisson's ratio, concrete tensile strength, concrete compressive strength, edge constraint form, and anchorage zone location; the base layer parameters include at least one of the following: base layer elastic modulus, base layer thickness, base layer support stiffness, bottom friction parameters, and bottom contact parameters; the prestressing tendon group parameters include at least one of the following: prestressing tendon oblique arrangement angle, prestressing tendon group number, number of prestressing tendon groups, prestressing tendon spacing, prestressing tendon diameter, prestressing tendon elastic modulus, prestressing tendon anchorage location, and coupling relationship between prestressing tendons and pavement panel; the tensioning level parameters include at least one of the following: graded tensioning force, tensioning control stress, single-step tensioning increment, number of tensioning levels, final tensioning level, and tensioning completion criterion; the construction constraints include at least one of the following: single-step maximum tensioning increment constraint, constraint prohibiting further tensioning of prestressing tendon groups that have reached the final tensioning level, tensioning sequence constraint of adjacent prestressing tendon groups, symmetrical tensioning constraint, maximum number of construction steps constraint, and tensioning equipment operation constraint.

[0010] Furthermore, the parametric finite element model of the inclined prestressed airport runway slab mentioned in step 2 includes a finite element solver that can be called through a script interface. The finite element solver includes Abaqus, and the script interface includes a Python script interface. The parametric finite element model automatically generates or updates the runway slab geometric model, base layer model, prestressed tendon group model, anchorage zone model, contact relationship, boundary conditions, tension analysis steps, and result output variables based on the engineering input parameters.

[0011] Furthermore, the state space mentioned in step 4 includes at least one of the following: the identifier of the tensioned prestressing tendon group, the current tensioning level of each prestressing tendon group, the current construction step number, the current maximum tensile stress, the current maximum compressive stress, the current slab displacement, the current slab edge warping displacement, the current anchorage zone stress index, the current stress uniformity index, and the current constraint verification identifier; the current constraint verification identifier includes at least one of the following: stress exceeding limit identifier, displacement exceeding limit identifier, anchorage zone stress exceeding limit identifier, and action legality identifier; the progressive tensioning action space consists of the prestressing tendon group selection action and the tensioning level selection action; wherein, the prestressing tendon group selection action corresponds to determining the next prestressing tendon group to be tensioned from multiple prestressing tendon groups, and the tensioning level selection action corresponds to determining the tensioning target level or tensioning increment of the next prestressing tendon group to be tensioned; the constraint verification method includes an action masking mechanism, which is used to mask candidate actions that do not meet the tensioning level constraint, tensioning sequence constraint, maximum construction step constraint, or action legality constraint according to the construction constraint conditions.

[0012] Furthermore, the reward function in step 4 is determined based on stress safety index, stress uniformity index, displacement control index, and process efficiency index. The stress safety index is determined based on the degree to which the maximum tensile stress, maximum compressive stress, and local stress in the anchorage zone exceed the corresponding allowable values. The stress uniformity index is determined based on the stress standard deviation, stress variation coefficient, or stress range of the plate surface, plate bottom, or a specified finite element output area. The displacement control index is determined based on the degree to which the vertical displacement, plate edge warping displacement, or adjacent area difference displacement exceeds the corresponding displacement limit. The process efficiency index is determined based on at least one of the total number of tensioning steps, the number of invalid actions, and the number of times the tensioning equipment is moved. The reward function includes a safety reward item, a uniformity reward item, a displacement control reward item, a process efficiency reward item, and a constraint penalty item. When the maximum tensile stress, maximum compressive stress, local stress in the anchorage zone, vertical displacement, plate edge warping displacement, or adjacent area difference displacement corresponding to any tensioning construction step exceeds the corresponding preset threshold, a constraint penalty is applied to the current action and / or the current training round is terminated; otherwise, a reward value is output.

[0013] Furthermore, the structural response data mentioned in step 7 includes at least one of the following: maximum principal tensile stress, maximum compressive stress, plate surface stress uniformity index, plate bottom stress uniformity index, maximum vertical displacement, plate edge warping displacement, adjacent area differential displacement, local stress in the anchorage zone, prestressed tendon stress, plate bottom contact pressure distribution, out-of-limit location coordinates, and constraint violation records.

[0014] Furthermore, after obtaining the candidate tensioning strategies in step 8, a sensitivity analysis is performed on the candidate tensioning strategies. The variables in the sensitivity analysis include at least one of the following: the oblique arrangement angle of the prestressing tendons, the tensioning control force, the graded tensioning method, the base stiffness, the boundary constraint stiffness, and the pavement material parameters. Based on the results of the sensitivity analysis, the applicable scope of the tensioning strategy, the strategy correction rules, and the recommended tensioning sequence or recommended tensioning level under different engineering parameters are output.

[0015] An optimization device for tensioning strategy of inclined prestressed airport pavement includes a parameter acquisition module for acquiring engineering input parameters of the inclined prestressed airport pavement panel, including pavement panel parameters, base layer parameters, prestressing tendon group parameters, tensioning level parameters, boundary conditions, and construction constraints; a finite element modeling module for establishing a parameterized finite element model of the inclined prestressed airport pavement panel based on the engineering input parameters; a sequential decision modeling module for constructing the tensioning construction process of the inclined prestressed airport pavement panel as a sequential decision problem, wherein each decision step includes selecting the prestressing tendon group to be tensioned and selecting the tensioning target level or tensioning increment; an online finite element invocation module for converting the phased tensioning strategy accumulated by the reinforcement learning agent at each decision step into a sequence of finite element analysis steps, and invoking a finite element solver through a script interface for finite element solution; and a response extraction module for extracting the structural response after the current decision step from the calculation results of the finite element solver. The system includes a reinforcement learning optimization module, comprising a state encoding unit, an action masking unit, a reward calculation unit, and a constraint processing unit. This module constructs a reinforcement learning environment including a state space, a progressive tensioning action space, state transition methods, constraint verification methods, and a reward function. It trains a reinforcement learning agent to progressively generate candidate tensioning strategies. The state encoding unit generates the current tensioning state vector, the action masking unit masks candidate actions that do not meet construction constraints, the reward calculation unit calculates reward values ​​based on stress safety indicators, stress uniformity indicators, displacement control indicators, and process efficiency indicators, and the constraint processing unit applies constraint penalties and / or terminates the current training round when the structural response exceeds limits. A full-process verification module verifies the full-process finite element structural response corresponding to the candidate tensioning strategies to obtain tensioning strategies that meet structural safety and construction constraints. A strategy output module outputs the optimized tensioning strategy and the corresponding structural response verification results.

[0016] An electronic device includes a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the computer program to implement the steps of the above-described method for optimizing the tensioning strategy of inclined prestressed airport pavement.

[0017] A computer-readable storage medium storing a computer program that, when executed by an obliquely prestressed airport pavement tensioning strategy optimization device, implements the steps of the obliquely prestressed airport pavement tensioning strategy optimization method described above.

[0018] Beneficial effects: 1) This invention abstracts the construction process of tensioning inclined prestressed airport pavement into a sequential decision problem, enabling the selection of prestressed tendon groups and tensioning levels to be uniformly described through state space, action space, reward function, and constraint verification, thereby improving the systematic nature of tensioning strategy optimization. It adopts a step-by-step tensioning action design, treating "which group of prestressed tendons to tension next" and "to which target level or increment" as interpretable actions, so that reinforcement learning output can be directly converted into construction tensioning steps and Abaqus analysis steps.

[0019] 2) In each decision step of reinforcement learning, Abaqus is called through the script interface to solve the current cumulative tensioning strategy. This can directly reflect the changes in the stress field, displacement field, prestressed tendon stress state and anchorage stress state of the plate under different tensioning construction steps, providing finite element mechanical basis for strategy optimization.

[0020] 3) By comprehensively considering stress safety, stress uniformity, displacement control and process efficiency in the reward function, and limiting unsafe tensioning paths through constraint penalties and / or training round termination mechanisms, it is beneficial to obtain a tensioning strategy that meets the multi-objective constraints of the entire construction process.

[0021] 4) Embedding finite element solution into the reinforcement learning state transition and reward calculation process enables the full-process structural response verification to be completed during the policy training phase, avoiding inconsistencies between the reinforcement learning training path and the actual force path of the finite element, and improving the engineering interpretability and reliability of the output strategy.

[0022] 5) By using sensitivity analysis, the applicable scope of the tensioning strategy, the strategy correction rules, and the recommended tensioning sequence or recommended tensioning level under different engineering parameters can be output, which is beneficial to adapting to different oblique angles, tensioning control forces, graded tensioning methods and base stiffness conditions. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating the overall method of the present invention; Figure 2 This is a schematic diagram of the parametric finite element model of the inclined prestressed airport runway panel in the embodiment. Figure 3 This is a schematic diagram of the sequential decision-making state and action transition during tensioning construction in the embodiment. Figure 4 This is a schematic diagram illustrating the interaction between reinforcement learning and Abaqus online invocation in the embodiment; Figure 5This is a logic diagram of the reward function and security constraint judgment in the embodiment; Figure 6 This is a flowchart of the full-process verification and sensitivity analysis in the embodiment. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit it. Terms such as "upper," "lower," "front," "rear," "left," "right," "bottom," "inner," and "outer" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the purpose of describing the invention. They do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention.

[0025] like Figure 1 The proposed optimization method for tensioning inclined prestressed airport pavement includes the following steps: Step 1: Obtain the engineering input parameters for the inclined prestressed airport pavement slab. These parameters include pavement slab parameters, base course parameters, prestressing tendon group parameters, tension level parameters, boundary conditions, and construction constraints. The pavement slab parameters include at least one of the following: pavement slab length, pavement slab width, pavement slab thickness, concrete elastic modulus, concrete Poisson's ratio, concrete tensile strength, concrete compressive strength, edge constraint type, and anchorage zone location. The base course parameters include at least one of the following: base course elastic modulus, base course thickness, base course support stiffness, slab bottom friction parameters, and slab bottom contact parameters. The prestressing tendon group parameters include the inclined arrangement of the prestressing tendons. The prestressing tendon group includes at least one of the following: angle, prestressing tendon group number, number of prestressing tendon groups, prestressing tendon spacing, prestressing tendon diameter, prestressing tendon elastic modulus, prestressing tendon anchorage position, and coupling relationship between prestressing tendon and paving slab; the tensioning level parameters include at least one of the following: graded tensioning force, tensioning control stress, single-step tensioning increment, number of tensioning levels, final tensioning level, and tensioning completion criterion; the construction constraints include at least one of the following: single-step maximum tensioning increment constraint, constraint prohibiting further tensioning of prestressing tendon groups that have reached the final tensioning level, tensioning sequence constraint of adjacent prestressing tendon groups, symmetrical tensioning constraint, maximum number of construction steps constraint, and tensioning equipment operation constraint; Step 2: Based on the engineering input parameters, establish a parametric finite element model of the inclined prestressed airport runway slab. The parametric finite element model is used to simulate the stress field, displacement field, prestressing tendon stress state, and anchorage zone stress state of the slab under different tensioning construction steps. The parametric finite element model of the inclined prestressed airport runway slab includes a finite element solver that can be called through a script interface. The finite element solver includes Abaqus, and the script interface includes a Python script interface. The parametric finite element model automatically generates or updates the slab geometric model, base layer model, prestressing tendon group model, anchorage zone model, contact relationship, boundary conditions, tensioning analysis steps, and result output variables according to the engineering input parameters. Step 3: The tensioning construction process of the inclined prestressed airport runway slab is constructed as a sequential decision problem, wherein the actions of each decision step include selecting the prestressed tendon group to be tensioned and selecting the tensioning target level or tensioning increment. Step 4: Construct a reinforcement learning environment, which includes a state space, a progressive tensioning action space, a state transition method, a constraint verification method, and a reward function. The state space includes at least one of the following: identifiers of tensioned prestressing tendon groups, current tensioning level of each prestressing tendon group, current construction step number, current maximum tensile stress, current maximum compressive stress, current slab displacement, current slab edge warping displacement, current anchorage zone stress index, current stress uniformity index, and current constraint verification identifiers. The current constraint verification identifiers include stress exceeding limits identifiers and displacement exceeding limits identifiers. The system includes at least one of the following: an anchorage zone stress exceeding limit indicator and an action legality indicator; the progressive tensioning action space consists of a prestressed tendon group selection action and a tensioning level selection action; wherein, the prestressed tendon group selection action corresponds to determining the next prestressed tendon group to be tensioned from multiple prestressed tendon groups, and the tensioning level selection action corresponds to determining the tensioning target level or tensioning increment of the next prestressed tendon group to be tensioned; the constraint verification method includes an action masking mechanism, which is used to shield tensioning that does not meet the tensioning level constraint according to the construction constraint conditions. Candidate actions are selected based on sequence constraints, maximum number of construction steps constraints, or action legality constraints. The reward function is determined based on stress safety indicators, stress uniformity indicators, displacement control indicators, and process efficiency indicators. The stress safety indicators are determined based on the degree to which the maximum tensile stress, maximum compressive stress, and local stress in the anchorage zone exceed the corresponding allowable values. The stress uniformity indicators are determined based on the stress standard deviation, stress variation coefficient, or stress range of the plate surface, plate bottom, or a specified finite element output area. The displacement control indicators are determined based on the degree to which the vertical displacement, plate edge warping displacement, or adjacent area difference displacement exceeds the corresponding displacement limit. The process efficiency indicators are determined based on at least one of the total number of tensioning steps, the number of invalid actions, and the number of times the tensioning equipment is transferred. The reward function includes a safety reward item, a uniformity reward item, a displacement control reward item, a process efficiency reward item, and a constraint penalty item. When the maximum tensile stress, maximum compressive stress, local stress in the anchorage zone, vertical displacement, plate edge warping displacement, or adjacent area difference displacement corresponding to any tensioning construction step exceeds the corresponding preset threshold, a constraint penalty is applied to the current action and / or the current training round is terminated; otherwise, a reward value is output. Step 5: In each decision step of the reinforcement learning environment, the reinforcement learning agent selects a step-by-step tensioning action based on the current tensioning state. The step-by-step tensioning action includes selecting the next prestressed tendon group to be tensioned and selecting the tensioning target level or tensioning increment of the prestressed tendon group to be tensioned. Step 6: Accumulate the progressive tensioning actions selected up to the current decision step into a staged tensioning strategy. Convert the staged tensioning strategy into a finite element analysis step sequence. Each analysis step in the finite element analysis step sequence corresponds to an accumulated progressive tensioning action. Then, call the finite element solver through the script interface. The parameterized finite element model is regenerated according to the staged tensioning strategy, or the calculation continues based on the finite element calculation results of the previous decision step by restarting the analysis, inheriting the state file, or adding analysis steps. After the analysis job is completed, read the output database to obtain the structural response data after the current decision step. Step 7: Extract the structural response data after the current decision step from the calculation results of the finite element solver. Perform constraint verification and reward calculation based on the structural response data, and update the next state of the reinforcement learning environment. When the structural response corresponding to any decision step does not meet the structural safety constraints, apply constraint penalties to the current step-by-step tensioning action and / or terminate the current training round. When all tensioning construction steps meet the structural safety constraints, determine the corresponding candidate tensioning strategy as the tensioning strategy that meets the constraints. The structural response data includes at least one of the following: maximum principal tensile stress, maximum compressive stress, plate surface stress uniformity index, plate bottom stress uniformity index, maximum vertical displacement, plate edge warping displacement, adjacent area differential displacement, local stress in the anchorage zone, prestressed tendon stress, plate bottom contact pressure distribution, out-of-limit position coordinates, and constraint violation records. Step 8: Repeat steps 5 to 7 until the tensioning completion criterion or the training round termination criterion is met to obtain all candidate tensioning strategies. After obtaining the candidate tensioning strategies, perform sensitivity analysis on the candidate tensioning strategies. The variables in the sensitivity analysis include at least one of the following: prestressed tendon oblique arrangement angle, tensioning control force, graded tensioning method, base layer stiffness, boundary constraint stiffness, and pavement slab material parameters. Based on the results of the sensitivity analysis, output the applicable scope of the tensioning strategy, strategy correction rules, and recommended tensioning sequence or recommended tensioning level under different engineering parameters. Step 9: Verify the full-process finite element structural response corresponding to the candidate tensioning strategy, and output the tensioning strategy that satisfies the structural safety constraints and construction constraints, as well as the corresponding structural response verification results. The tensioning strategy is the optimized inclined prestressed airport pavement tensioning strategy, including the tensioning sequence of prestressed tendon groups and the tensioning target level or tensioning increment corresponding to each prestressed tendon group.

[0026] An optimization device for tensioning strategy of inclined prestressed airport pavement includes a parameter acquisition module for acquiring engineering input parameters of the inclined prestressed airport pavement panel, including pavement panel parameters, base layer parameters, prestressing tendon group parameters, tensioning level parameters, boundary conditions, and construction constraints; a finite element modeling module for establishing a parameterized finite element model of the inclined prestressed airport pavement panel based on the engineering input parameters; a sequential decision modeling module for constructing the tensioning construction process of the inclined prestressed airport pavement panel as a sequential decision problem, wherein each decision step includes selecting the prestressing tendon group to be tensioned and selecting the tensioning target level or tensioning increment; an online finite element invocation module for converting the phased tensioning strategy accumulated by the reinforcement learning agent at each decision step into a sequence of finite element analysis steps, and invoking a finite element solver through a script interface for finite element solution; and a response extraction module for extracting the structural response after the current decision step from the calculation results of the finite element solver. The system includes a reinforcement learning optimization module, comprising a state encoding unit, an action masking unit, a reward calculation unit, and a constraint processing unit. This module constructs a reinforcement learning environment including a state space, a progressive tensioning action space, state transition methods, constraint verification methods, and a reward function. It trains a reinforcement learning agent to progressively generate candidate tensioning strategies. The state encoding unit generates the current tensioning state vector, the action masking unit masks candidate actions that do not meet construction constraints, the reward calculation unit calculates reward values ​​based on stress safety indicators, stress uniformity indicators, displacement control indicators, and process efficiency indicators, and the constraint processing unit applies constraint penalties and / or terminates the current training round when the structural response exceeds limits. A full-process verification module verifies the full-process finite element structural response corresponding to the candidate tensioning strategies to obtain tensioning strategies that meet structural safety and construction constraints. A strategy output module outputs the optimized tensioning strategy and the corresponding structural response verification results.

[0027] An electronic device includes a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the computer program to implement the steps of the above-described method for optimizing the tensioning strategy of inclined prestressed airport pavement.

[0028] A computer-readable storage medium storing a computer program that, when executed by an obliquely prestressed airport pavement tensioning strategy optimization device, implements the steps of the obliquely prestressed airport pavement tensioning strategy optimization method described above. The computer-readable storage medium can be a hard disk, solid-state drive, memory card, read-only memory, random access memory, optical disk, or any medium with program storage capabilities.

[0029] Example: Taking the inclined prestressed runway slab of a certain airport as the object, the multi-step tensioning selection in the tensioning construction process is transformed into a sequential decision problem, and a closed loop is formed through online interaction between the reinforcement learning environment and the parametric Abaqus finite element model.

[0030] First, obtain the engineering input parameters, which include pavement panel parameters, base layer parameters, prestressed tendon group parameters, tension level parameters, boundary conditions, and construction constraints. Pavement panel parameters describe the geometry and material properties of the pavement panel, base layer parameters describe the support conditions under the pavement panel, prestressed tendon group parameters describe the arrangement of the inclined prestressed tendon groups, tension level parameters describe the graded tensioning method, boundary conditions describe the edge, bottom, and anchorage constraints of the slab, and construction constraints limit the optional actions during the tensioning construction process.

[0031] Secondly, a parametric finite element model is established. The parametric finite element model can automatically update the geometry, material, mesh, prestressed tendon group, contact relationship, boundary conditions and analysis steps of the finite element model according to the engineering input parameters. For a given tensioning strategy, the parametric finite element model can perform step-by-step tensioning analysis and output the structural response data after each tensioning construction step.

[0032] Next, a reinforcement learning environment is constructed. The reinforcement learning environment includes a state space, a progressive tensioning action space, a state transition method, a constraint verification method, and a reward function. The progressive tensioning action includes selecting the next prestressed tendon group to be tensioned and selecting the tensioning target level or tensioning increment for that prestressed tendon group.

[0033] Next, the reinforcement learning agent is trained. At each decision step, the agent selects a progressive tensioning action based on the current state. The reinforcement learning environment accumulates this action with previously executed actions to form a phased tensioning strategy, and then uses a Python script interface to call Abaqus to regenerate or update the tensioning analysis steps. After Abaqus completes the solution, the reinforcement learning environment reads the stress, displacement, anchorage zone response, and contact response of the current phase from the output database, and receives a reward or penalty based on the reward function.

[0034] Finally, when all prestressed tendon groups reach the target tension level and all constraints are met throughout the process, the tensioning strategy and structural response verification results are output. When any tensioning construction step fails to meet the structural safety constraints, the current strategy path is judged as a failed path and is fed back to the reinforcement learning agent through constraint penalties, round termination, or reduction of action selection probability.

[0035] The parametric finite element model includes a pavement slab model, a base course model, a prestressed tendon group model, an anchorage zone model, a contact model, a boundary condition model, and a tension analysis step model. The pavement slab model can be simulated using solid elements, while the base course model can be simulated using solid elements, elastic foundation elements, or equivalent support boundaries. The prestressed tendon group model can be simulated using truss elements, beam elements, embedded steel reinforcement elements, or an equivalent prestressed loading region. The anchorage zone model can have locally refined meshes to extract the local stress response within the anchorage zone.

[0036] In a preferred implementation, Abaqus is used to automatically build or update the parametric finite element model via a Python script interface. The Python script can automatically generate the pavement geometry, base geometry, oblique arrangement path of prestressed tendons, prestressed tendon group number, anchorage zone location, material properties, contact relationships, boundary conditions, mesh generation, and analysis step settings based on the engineering input parameters.

[0037] Each tensioning action is converted into one or more finite element analysis steps. The tensioning load can be applied through the initial stress of the prestressing tendon, equivalent prestrain, temperature equivalent shrinkage, or prestressing loading methods supported by the finite element solver. After each analysis step is completed, the stress field of the plate, the displacement field of the plate, the stress state of the prestressing tendon, and the stress state of the anchorage zone are automatically extracted.

[0038] The output variables of the finite element method include the maximum principal tensile stress, the maximum compressive stress, the stress uniformity index of the plate surface, the stress uniformity index of the plate bottom, the maximum vertical displacement, the plate edge warping displacement, the differential displacement between adjacent areas, the local stress in the anchorage zone, the prestressing tendon stress, the contact pressure distribution at the bottom of the plate, the coordinates of the out-of-limit location, and the constraint violation record. The above output variables are used to update the reinforcement learning state, calculate the reward function, perform safety constraint judgments, and form the full-process verification results.

[0039] The reinforcement learning environment invokes Abaqus online to generate the cumulative tensioning strategy as follows: In the initial state, the current tensioning level of each prestressed tendon group can be set to level zero or the initial prestressing level. The reinforcement learning agent selects action At at the t-th decision step. Action At includes the number of the prestressed tendon group to be tensioned next and the target tensioning level or tensioning increment for that prestressed tendon group.

[0040] The reinforcement learning environment accumulates the actions from A1 to At in chronological order to form a phased tensioning strategy πt. The phased tensioning strategy πt includes the sequence of tensioned prestressed tendon groups, the target level or increment of each tensioned prestressed tendon group, the corresponding construction step number, and the method of applying the tensioning load.

[0041] In one implementation, each decision step is regenerated by a Python script based on the phased tensioning strategy πt, and the finite element calculation is completed in the order of analysis steps A1 to At. In another implementation, Abaqus continues the calculation based on the results of the previous decision step, using methods such as restarting the analysis, state inheritance, or appending analysis steps to reduce repetitive modeling operations. Both methods use the Abaqus solution results as the basis for state transitions in the reinforcement learning environment.

[0042] After each finite element method (FEM) calculation, the structural response following the current construction step is read from the Abaqus output database, forming a state St, action At, next state St+1, reward Rt, and constraint identifier Ct. This data can be stored in the experience replay pool for the reinforcement learning agent to update its strategy, but approximate prediction results are not used as the basis for the master state transition.

[0043] The method for constructing the progressively tensile motion space and constrained motion mask is as follows: The action space consists of the prestressed tendon group selection action and the tensioning level selection action. The prestressed tendon group selection action corresponds to determining the next prestressed tendon group to be tensioned from multiple prestressed tendon groups. The tensioning level selection action corresponds to determining the tensioning target level or tensioning increment of the prestressed tendon group. Each action can be represented as a combination of the prestressed tendon group number and the tensioning level number.

[0044] The constraint action mask is used to block non-executable actions before action selection. Non-executable actions include actions such as continuing to tension a prestressed tendon group that has reached the final tension level beyond the control level, single-step tension increment exceeding the allowable value, tension level lower than the current level, exceeding the maximum number of construction steps, violating symmetrical tension constraints, violating the construction sequence constraints of adjacent prestressed tendon groups, or violating the operating range constraints of tensioning equipment.

[0045] By using action masks, reinforcement learning agents can only choose their next action from a set of engineering-feasible actions. If the agent's tentative output is a masked action, the reinforcement learning environment can either replace that action with the highest-probability action from the set of legal actions or directly impose an illegal action penalty.

[0046] The progressive tensioning action space inherently imbues the reinforcement learning output strategy with the meaning of construction steps. The strategy output can be directly represented as a sequence of construction instructions: "Step 1: Tension the i-th group to the k-th level; Step 2: Tension the j-th group to the m-th level," and can be directly mapped to the Abaqus tensioning analysis step sequence.

[0047] The reinforcement learning environment is constructed as follows: the state space is used to describe the current tensioning construction state. The state space includes at least one of the following: the identifier of the tensioned prestressing tendon group, the current tensioning level of each prestressing tendon group, the current construction step number, the current maximum tensile stress, the current maximum compressive stress, the current slab displacement, the current slab edge warping displacement, the current anchorage zone stress index, the current stress uniformity index, and the current constraint check identifier.

[0048] The current constraint verification indicators include at least one of the following: stress exceeding limit indicator, displacement exceeding limit indicator, anchorage zone stress exceeding limit indicator, and action legality indicator. For example, the stress exceeding limit indicator is used to indicate whether the maximum tensile stress, maximum compressive stress, or local stress in the anchorage zone exceeds the corresponding allowable value; the displacement exceeding limit indicator is used to indicate whether the vertical displacement, plate edge warping displacement, or differential displacement between adjacent areas exceeds the corresponding limit value; and the action legality indicator is used to indicate whether the candidate action meets the construction constraint conditions.

[0049] The action space consists of prestressed tendon group selection action and tension level selection action. The prestressed tendon group selection action corresponds to determining the prestressed tendon group to be tensioned from multiple prestressed tendon groups, and the tension level selection action corresponds to determining the tension target level or tension increment of the prestressed tendon group. Each action can be represented as a combination of prestressed tendon group number and tension level number.

[0050] The constraint verification method includes an action masking mechanism. The action masking mechanism masks candidate actions that do not meet the tensioning level constraint, tensioning sequence constraint, maximum number of construction steps constraint, or action legality constraint based on the construction constraints. Through the action masking mechanism, the reinforcement learning agent can only select tensioning actions within the action space that can be implemented in the project.

[0051] State transitions are accomplished through online finite element analysis. After the reinforcement learning agent selects an action, the reinforcement learning environment calls the parameterized Abaqus finite element model according to the cumulative tension strategy, reads the structural response after the current decision step, and updates the next state based on this structural response. For scenarios with limited computational resources, parallel Abaqus jobs, restart analysis, mesh and boundary condition caching, or experience replay reuse can be used to improve training efficiency, but this does not change the main process of using finite element solution results as the basis for state transitions.

[0052] The reward function and safety constraint handling methods are as follows: The reward function includes a safety reward item, a uniformity reward item, a displacement control reward item, a process efficiency reward item, and a constraint penalty item. The safety reward item is determined based on the degree to which the maximum tensile stress, maximum compressive stress, and local stress in the anchorage zone exceed the corresponding allowable values. The uniformity reward item is determined based on the stress standard deviation, stress variation coefficient, or stress range of the plate surface, plate bottom, or a specified finite element output area. The displacement control reward item is determined based on the degree to which the vertical displacement, plate edge warping displacement, or differential displacement between adjacent areas exceeds the corresponding displacement limit. The process efficiency reward item is determined based on at least one of the following: total tensioning steps, number of invalid actions, and number of times the tensioning equipment is moved.

[0053] In one implementation, the reward function can be expressed as: R = w1Rs + w2Ru + w3Rd + w4Re – P, where R represents the reward value, Rs represents the safety reward, Ru represents the uniformity reward, Rd represents the displacement control reward, Re represents the process efficiency reward, P represents the constraint penalty, and w1, w2, w3, and w4 represent the corresponding weights. The above formula is used to illustrate the composition of the reward function and does not limit the reward function to this form of expression.

[0054] When the maximum tensile stress, maximum compressive stress, local stress in the anchorage zone, vertical displacement, edge warping displacement, or differential displacement between adjacent areas corresponding to any tensioning construction step exceeds the corresponding preset threshold, a constraint penalty can be applied to the current action. The constraint penalty is used to reduce the probability that the action that causes the over-limit response will be selected again.

[0055] In another implementation, when stress, displacement, or anchorage zone response exceeds the limit in any tensioning construction step, the current training round is terminated directly and marked as a failed round. Constraint penalties and training round termination can also be used simultaneously to balance training efficiency and engineering safety constraints.

[0056] The finite element method for verifying the tensioning strategy throughout the entire process is as follows: During the reinforcement learning training, each decision step converts the current cumulative tensioning strategy into a sequence of analysis steps for the parameterized finite element model. Each analysis step corresponds to a tensioning action, including the prestressed tendon group to be tensioned and the tensioning target level or tension increment.

[0057] Tension loads are applied step-by-step using finite element solvers such as Abaqus, and the stress field, displacement field, prestressing tendon stress state, and anchorage zone stress state are extracted after each step. The entire process is verified focusing on maximum tensile stress, maximum compressive stress, local stress in the anchorage zone, vertical displacement, edge warping displacement, differential displacement between adjacent areas, and stress uniformity indices.

[0058] When all tensioning construction steps meet the structural safety constraints, the current strategy is determined as the optimized tensioning strategy, and the construction step number, the tensioning sequence of the prestressed tendon group, the tensioning target level or tensioning increment, and the structural response verification results for each step are output. When any tensioning construction step does not meet the structural safety constraints, the current strategy path is determined as a failed path.

[0059] For failed paths, their corresponding tensioning state, actions, structural responses, and constraint violation types can be recorded. This record can then be used to update the experience replay pool, adjust reward and penalty weights, or add action mask constraints. Subsequently, the reinforcement learning agent continues to search for new tensioning strategies within the legal action space.

[0060] The sensitivity analysis and strategy correction rules are as follows: After obtaining the optimized tensioning strategy, change the oblique arrangement angle of the prestressing tendons, the tensioning control force, the graded tensioning method, the base stiffness, the boundary constraint stiffness, or the paving material parameters, and repeat the parametric finite element full-process verification. By comparing the changes in structural response under different variable conditions, the sensitivity of the tensioning strategy to changes in engineering parameters can be determined.

[0061] For example, when the angle of the prestressing tendons is changed, the stress transmission path on the slab surface and the local stress distribution in the anchorage zone will change accordingly; when the stiffness of the base layer changes, the vertical displacement of the slab, the warping displacement of the slab edge, and the contact pressure distribution at the bottom of the slab will change accordingly; when the staged tensioning method changes, the peak stress, the total number of tensioning steps, and the process efficiency will change accordingly.

[0062] Based on the sensitivity analysis results, the applicable scope of the tensioning strategy, strategy correction rules, recommended tensioning sequence or recommended tensioning level under different engineering parameters can be output. The strategy correction rules may include increasing the number of graded tensioning levels, reducing the single-step tensioning increment, adjusting the tensioning sequence of the edge prestressed tendon group, increasing the safety constraint weight of the anchorage zone, or changing the operation organization method of the tensioning equipment.

[0063] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the principles and spirit of the present invention should be included within the protection scope of the present invention.

Claims

1. An optimization method for tensioning strategies of inclined prestressed airport pavement, characterized in that: Includes the following steps: Step 1: Obtain the engineering input parameters for the inclined prestressed airport pavement slab. The engineering input parameters include pavement slab parameters, base course parameters, prestressed tendon group parameters, tensioning level parameters, boundary conditions, and construction constraints. Step 2: Based on the engineering input parameters, establish a parametric finite element model of the inclined prestressed airport runway slab. The parametric finite element model is used to simulate the stress field, displacement field, prestressing tendon stress state, and anchorage zone stress state of the slab under different tensioning construction steps. Step 3: The tensioning construction process of the inclined prestressed airport runway slab is constructed as a sequential decision problem, wherein the actions of each decision step include selecting the prestressed tendon group to be tensioned and selecting the tensioning target level or tensioning increment. Step 4: Construct a reinforcement learning environment, which includes a state space, a progressively stretched action space, a state transition method, a constraint verification method, and a reward function; Step 5: In each decision step of the reinforcement learning environment, the reinforcement learning agent selects a step-by-step tensioning action based on the current tensioning state. The step-by-step tensioning action includes selecting the next prestressed tendon group to be tensioned and selecting the tensioning target level or tensioning increment of the prestressed tendon group to be tensioned. Step 6: Accumulate the progressive tensioning actions selected up to the current decision step into a staged tensioning strategy. Convert the staged tensioning strategy into a finite element analysis step sequence. Each analysis step in the finite element analysis step sequence corresponds to an accumulated progressive tensioning action. Then, call the finite element solver through the script interface. The parameterized finite element model is regenerated according to the staged tensioning strategy, or the calculation continues based on the finite element calculation results of the previous decision step by restarting the analysis, inheriting the state file, or adding analysis steps. After the analysis job is completed, read the output database to obtain the structural response data after the current decision step. Step 7: Extract the structural response data after the current decision step from the calculation results of the finite element solver, perform constraint verification and reward calculation based on the structural response data, and update the next state of the reinforcement learning environment. When the structural response corresponding to any decision step does not meet the structural safety constraints, apply constraint penalties to the current step-by-step tensioning action and / or terminate the current training round. When all tensioning construction steps in the entire process meet the structural safety constraints, the corresponding candidate tensioning strategy will be determined as the tensioning strategy that meets the constraints. Step 8: Repeat steps 5 to 7 until the tension completion criterion or the training round termination criterion is met to obtain all candidate tensioning strategies; Step 9: Verify the full-process finite element structural response corresponding to the candidate tensioning strategy, and output the tensioning strategy that satisfies the structural safety constraints and construction constraints, as well as the corresponding structural response verification results. The tensioning strategy is the optimized inclined prestressed airport pavement tensioning strategy, including the tensioning sequence of prestressed tendon groups and the tensioning target level or tensioning increment corresponding to each prestressed tendon group.

2. The method for optimizing the tensioning strategy of inclined prestressed airport pavement as described in claim 1, characterized in that: The paving panel parameters mentioned in step 1 include at least one of the following: paving panel length, paving panel width, paving panel thickness, concrete elastic modulus, concrete Poisson's ratio, concrete tensile strength, concrete compressive strength, edge restraint type, and anchorage zone location; the base layer parameters include at least one of the following: base layer elastic modulus, base layer thickness, base layer support stiffness, slab bottom friction parameters, and slab bottom contact parameters; the prestressing tendon group parameters include at least one of the following: prestressing tendon oblique arrangement angle, prestressing tendon group number, number of prestressing tendon groups, prestressing tendon spacing, prestressing tendon diameter, prestressing tendon elastic modulus, prestressing tendon anchorage location, and coupling relationship between prestressing tendons and paving panel; the tensioning level parameters include at least one of the following: graded tensioning force, tensioning control stress, single-step tensioning increment, number of tensioning levels, final tensioning level, and tensioning completion criterion; the construction constraints include at least one of the following: single-step maximum tensioning increment constraint, constraint prohibiting further tensioning of prestressing tendon groups that have reached the final tensioning level, tensioning sequence constraint of adjacent prestressing tendon groups, symmetrical tensioning constraint, maximum number of construction steps constraint, and tensioning equipment operation constraint.

3. The method for optimizing the tensioning strategy of inclined prestressed airport pavement as described in claim 1, characterized in that: The parametric finite element model of the inclined prestressed airport runway slab mentioned in step 2 includes a finite element solver that can be called through a script interface. The finite element solver includes Abaqus, and the script interface includes a Python script interface. The parametric finite element model automatically generates or updates the runway slab geometry model, base layer model, prestressed tendon group model, anchorage zone model, contact relationship, boundary conditions, tension analysis steps, and result output variables based on the engineering input parameters.

4. The optimization method for inclined prestressed airport pavement tensioning strategy as described in claim 1, characterized in that: The state space described in step 4 includes at least one of the following: identifier of the tensioned prestressing tendon group, current tensioning level of each prestressing tendon group, current construction step number, current maximum tensile stress, current maximum compressive stress, current slab displacement, current slab edge warping displacement, current anchorage zone stress index, current stress uniformity index, and current constraint verification identifier. The current constraint verification identifier includes at least one of the following: stress exceeding limit identifier, displacement exceeding limit identifier, anchorage zone stress exceeding limit identifier, and action legality identifier. The progressive tensioning action space consists of prestressing tendon group selection action and tensioning level selection action. The prestressing tendon group selection action corresponds to determining the next prestressing tendon group to be tensioned from multiple prestressing tendon groups, and the tensioning level selection action corresponds to determining the tensioning target level or tensioning increment of the next prestressing tendon group to be tensioned. The constraint verification method includes an action masking mechanism, which is used to mask candidate actions that do not meet the tensioning level constraint, tensioning sequence constraint, maximum construction step constraint, or action legality constraint according to the construction constraint conditions.

5. The method for optimizing the tensioning strategy of inclined prestressed airport pavement as described in claim 1, characterized in that: The reward function described in step 4 is determined based on stress safety index, stress uniformity index, displacement control index, and process efficiency index. The stress safety index is determined based on the degree to which the maximum tensile stress, maximum compressive stress, and local stress in the anchorage zone exceed the corresponding allowable values. The stress uniformity index is determined based on the stress standard deviation, stress variation coefficient, or stress range of the plate surface, plate bottom, or a specified finite element output area. The displacement control index is determined based on the degree to which the vertical displacement, plate edge warping displacement, or adjacent area difference displacement exceeds the corresponding displacement limit. The process efficiency index is determined based on at least one of the total number of tensioning steps, the number of invalid actions, and the number of times the tensioning equipment is moved. The reward function includes a safety reward item, a uniformity reward item, a displacement control reward item, a process efficiency reward item, and a constraint penalty item. When the maximum tensile stress, maximum compressive stress, local stress in the anchorage zone, vertical displacement, plate edge warping displacement, or adjacent area difference displacement corresponding to any tensioning construction step exceeds the corresponding preset threshold, a constraint penalty is applied to the current action and / or the current training round is terminated; otherwise, a reward value is output.

6. The method for optimizing the tensioning strategy of inclined prestressed airport pavement as described in claim 1, characterized in that: The structural response data mentioned in step 7 includes at least one of the following: maximum principal tensile stress, maximum compressive stress, plate surface stress uniformity index, plate bottom stress uniformity index, maximum vertical displacement, plate edge warping displacement, differential displacement between adjacent areas, local stress in the anchorage zone, prestressed tendon stress, plate bottom contact pressure distribution, coordinates of out-of-limit locations, and constraint violation records.

7. The method for optimizing the tensioning strategy of inclined prestressed airport pavement as described in claim 1, characterized in that: After obtaining the candidate tensioning strategies as described in step 8, a sensitivity analysis is performed on the candidate tensioning strategies. The variables in the sensitivity analysis include at least one of the following: the oblique arrangement angle of the prestressing tendons, the tensioning control force, the graded tensioning method, the base stiffness, the boundary constraint stiffness, and the pavement material parameters. Based on the results of the sensitivity analysis, the applicable scope of the tensioning strategy, the strategy correction rules, and the recommended tensioning sequence or recommended tensioning level under different engineering parameters are output.

8. A device for optimizing the tensioning strategy of inclined prestressed airport pavement, characterized in that: The system includes a parameter acquisition module for acquiring engineering input parameters of the inclined prestressed airport pavement slab, including pavement slab parameters, base course parameters, prestressed tendon group parameters, tensioning level parameters, boundary conditions, and construction constraints; a finite element modeling module for establishing a parametric finite element model of the inclined prestressed airport pavement slab based on the engineering input parameters; and a sequential decision modeling module for constructing the tensioning construction process of the inclined prestressed airport pavement slab as a sequential decision problem, wherein each decision step includes selecting the prestressed tendon group to be tensioned and selecting the tensioning target level or tensioning increment. The online finite element invocation module is used to convert the phased tension strategy accumulated by the reinforcement learning agent at each decision step into a sequence of finite element analysis steps, and call the finite element solver through the script interface to perform finite element solution; The response extraction module is used to extract structural response data after the current decision step from the calculation results of the finite element solver. The reinforcement learning optimization module includes a state encoding unit, an action masking unit, a reward calculation unit, and a constraint processing unit. It is used to construct a reinforcement learning environment that includes a state space, a progressive tensioning action space, a state transition method, a constraint verification method, and a reward function. It also trains the reinforcement learning agent to progressively generate candidate tensioning strategies. The state encoding unit is used to generate the current tensioning state vector, the action masking unit is used to mask candidate actions that do not meet the construction constraints, the reward calculation unit is used to calculate the reward value based on the stress safety index, stress uniformity index, displacement control index, and process efficiency index, and the constraint processing unit is used to apply constraint penalties and / or terminate the current training round when the structural response exceeds the limit. The full-process verification module is used to verify the full-process finite element structural response corresponding to the candidate tensioning strategy, and obtain the tensioning strategy that meets the structural safety constraints and construction constraints. The strategy output module is used to output the optimized tensioning strategy and the corresponding structural response verification results.

9. An electronic device, characterized in that: The method includes a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the computer program to implement the steps of the method for optimizing the tensioning strategy of inclined prestressed airport pavement as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: It stores a computer program that, when executed by the inclined prestressed airport pavement tensioning strategy optimization device, implements the steps of the inclined prestressed airport pavement tensioning strategy optimization method as described in any one of claims 1 to 7.

Images