A road structure design method based on large model and reinforcement learning

Abstract

The application belongs to the intersection field of road engineering technology, artificial intelligence and engineering mechanics, and particularly relates to a road structure design method based on a large model and reinforcement learning. A technical closed loop of "natural language input-automatic parameter mapping-standard value acquisition-finite element verification-reinforcement learning optimization" is constructed: by constructing a load parameter mapping table and a material semantic encoder, the system can automatically convert natural language input into engineering parameters such as load, material and layer thickness, solving the problem of non-standard working condition adaptability. Further combined with the mechanical theory solution and the finite element simulation, a deviation feature space is constructed to drive the reinforcement learning agent to quickly optimize under the condition of meeting the theoretical constraints. Compared with the traditional trial method and the existing intelligent optimization scheme, the number of design iterations and the material cost are greatly reduced, the safety and compliance of the output scheme are improved, and the intelligent leap from experience driving to theory guiding is realized.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of road engineering, the intersection of artificial intelligence and engineering mechanics, and specifically relates to a road structure intelligent design method fusing semantic analysis of a large language model and reinforcement learning optimization, which is suitable for intelligent design and optimization of flexible pavement structures of highways, urban roads and the like. BACKGROUND

[0002] Traditional road structure design follows the elastic layered system theory and adopts a trial calculation method based on an empirical formula to determine the thickness of each structure layer. A typical process is as follows: a structure combination is initially selected according to the traffic load level, the thickness parameters are adjusted through mechanical response verification, and the iteration is repeated until the design standard is met. This method has two major problems: first, the degree of intelligence is low, the design parameter input depends on manual filling of standardized forms, natural language interaction cannot be supported, and the adaptability to non-standard working conditions is poor; second, the optimization efficiency is insufficient, the thickness adjustment depends on the experience of engineers, the trial calculation needs to be read repeatedly, the finite element verification has a high calculation cost, and multi-objective collaborative optimization is difficult.

[0003] In recent years, the emergence of artificial intelligence technology, especially deep learning algorithms, has made it possible to quickly design and optimize pavement structures. For example, a neural network-based asphalt pavement structure optimization method improves the efficiency of parameter sensitivity analysis through a data-driven strategy, but does not solve the problem of deep integration of natural language interaction and reinforcement learning strategies; an optimization scheme combining a genetic algorithm can globally optimize, but has defects such as slow convergence speed and easy falling into local optimum. Reinforcement learning technology, especially the proximal policy optimization algorithm (PPO), provides a new paradigm for pavement structure design and optimization due to its stable and efficient, multi-objective collaborative, and high sample reusability. SUMMARY

[0004] In view of the problems existing in the prior art, the application proposes a road structure design method based on a large model and reinforcement learning, which realizes the intelligent upgrading of road structure design by constructing a technical closed loop of "natural language input-automatic parameter mapping-standard value acquisition-finite element verification-reinforcement learning optimization", and significantly improves the design efficiency.

[0005] TECHNICAL SCHEME

[0006] A road structure design method based on a large model and reinforcement learning, comprising the following steps:

[0007] Step 101, a user inputs a natural language design requirement, converts the natural language design requirement into a prompt word according to a preset prompt word template, and sends it to a pre-trained large language model;

[0008] Step 102, the pre-trained large language model performs semantic analysis on the input prompt word text, identifies key engineering parameters and the correlation between parameters;

[0009] Step 103, based on the engineering parameter mapping rule library, map the keywords to standard engineering parameters as parameter vectors that can be used for subsequent standard tolerance value extraction and finite element modeling;

[0010] Step 104, generate a finite element modeling script according to the parameter vectors parsed in steps 102 and 103, drive the finite element software to perform four-layer road structure (surface layer, base layer, subbase layer, soil base) finite element modeling, boundary condition application and response solving, and extract the finite element analysis result vector ;

[0011] Step 105, query the standard value, according to the key design parameters (especially traffic load grade, pavement structure combination type) obtained in steps 102 and 103, based on JTG D50-2017 "Highway Asphalt Pavement Design Specification", calculate the design traffic volume N e according to the input pavement structure parameters, and according to the design traffic volume N e and the corresponding standard tolerance value vector of the design parameters ;

[0012] Step 106, take the deviation between the finite element analysis result and the standard tolerance value as the input of the reinforcement learning agent, construct the state space of the reinforcement learning agent, and use the reinforcement learning algorithm (PPO) of the proximal policy optimization to iteratively adjust the thickness of the structure layer, realizing the optimization of the pavement structure.

[0013] Step 107, set the reward function for pavement structure design optimization and the training stability guarantee mechanism in the reinforcement learning agent, which is used to train the agent action suggestion.

[0014] Step 108, the agent outputs reinforcement learning action suggestion , combined with the action suggestion given by the large language model , hybrid output pavement structure layer adjustment scheme, forming a new pavement structure layer.

[0015] Step 109, according to the action suggestion in step 108, modify and construct a new pavement structure parameter vector, re-perform finite element modeling and input into the reinforcement learning agent for training until the convergence requirement is met.

[0016] Wherein, the convergence requirement is that all key mechanical responses (σ , , ) of the current design scheme meet the standard tolerance value (i.e. ≤ ), and in the continuous N times (N=5) iterations, the increase of the comprehensive reward r is less than the threshold value delta (delta=0.01) or the structural layer parameter change is small.

[0017] Step 110, output the optimized road structure design scheme meeting the safety, economy and engineering requirements.

[0018] Beneficial effects:

[0019] The application breaks through the limitation of traditional methods relying on manual form filling by constructing a semantic analysis engine through pre-training of a large language model, and automatically converting user natural language input into standardized loads, materials and structure parameters. Combined with an engineering parameter mapping rule library, intelligent analysis of unstructured text into engineering parameters is realized, significantly improving non-standard working condition adaptability and input efficiency; The application innovatively integrates the specification allowable standard value and the finite element numerical simulation result to construct a state space with the deviation between the two as the core, driving the reinforcement learning optimization process. Compared with a pure data-driven scheme, it can ensure that the optimization direction meets the specification requirements and engineering actual constraints at the same time;

[0020] The application adopts a multi-objective reward function driven engineering decision mechanism, constructs a reward function integrating safety, economy, dynamic penalty and smoothness, and ensures that the structure reliability, economy and meeting of engineering specification requirements can be stably converged at the same time;

[0021] The intelligent optimization process of the application adopts a closed-loop verification and has an interpretable decision support, embeds a parameter verification module in the modeling stage to ensure the compliance of design assumptions. At the same time, the optimization layer thickness parameters, material configuration and multi-dimensional verification report are output, providing an automatic and interpretable decision that cannot be realized by traditional methods. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 is a road structure design method flowchart of the application based on a large model and reinforcement learning;

[0023] Figure 2 is a natural language input and parameter completion logic schematic diagram of an embodiment of the application;

[0024] Figure 3 is a reinforcement learning agent state space construction schematic diagram of an embodiment of the application;

[0025] Figure 4 is a pavement structure layer parameter adjustment and training stability guarantee mechanism schematic diagram of an embodiment of the application. DETAILED DESCRIPTION

[0026] The technical solutions provided by the application will be further described below with reference to specific embodiments and the accompanying drawings. The advantages and features of the application will be clearer in combination with the following description.

[0027] In related technologies, traditional road structure design methods rely on manual experience and trial-and-error algorithms, requiring repeated adjustments to layer thickness parameters and verification through finite element analysis. This results in low levels of intelligence and long optimization cycles. While existing intelligent optimization schemes (such as neural networks and genetic algorithms) can improve efficiency, they face two major bottlenecks: first, parameter input relies on standardized tables, making it unable to parse non-standard working conditions described in natural language; second, the optimization process lacks standardized constraints, easily leading to engineering infeasibility solutions such as unbalanced layer thickness ratios and excessive material modulus, requiring secondary manual verification.

[0028] To address the aforementioned issues, this embodiment proposes a road structure design method integrating a large language model and reinforcement learning. By constructing a load parameter mapping table and a material semantic encoder, the system can automatically convert natural language input into engineering parameters such as load, material, and layer thickness, solving the adaptability problem for non-standard working conditions. Furthermore, by combining mechanical theory solutions with finite element simulations to construct a deviation feature space, the reinforcement learning agent is driven to rapidly find the optimal solution while satisfying theoretical constraints. Compared to traditional trial-and-error methods and existing intelligent optimization schemes, this embodiment significantly reduces the number of design iterations and material costs, improves the safety and compliance of the output solution, and achieves an intelligent leap from experience-driven to theory-guided approaches.

[0029] like Figure 1 As shown, a road structure design method based on large models and reinforcement learning includes the following steps:

[0030] Step 101: The user inputs natural language design requirements. Based on a preset prompt word template, the natural language design requirements are converted into prompt words and sent to a pre-trained large language model, such as... Figure 2 As shown.

[0031] The prompt word template is:

[0032] systemPrompt = ['You are a road engineering expert. Parse road structure parameters based on user input and return them strictly in JSON format:' ...

[0033] '{"traffic_level":"light load / medium load / heavy load / extremely heavy load","road_type":"highway / urban road / rural road / industrial road",' ...

[0034] '"vehicle_speed_kmh": value, "subgrade_conditions": "soft soil / general soil / hard soil / rock",' ...

[0035] '"subgrade_treatment":"natural soil foundation / improved soil foundation / replacement soil foundation / composite foundation",' ...

[0036] '{"layer_type": "base course", "thickness_cm": 35, "material": "cement stabilized macadam", "modulus_mpa": 400, "poisson": 0.35}, '...

[0037] '{"layer_type": "base course", "thickness_cm": 35, "material": "cement stabilized macadam", "modulus_mpa": 400, "poisson": 0.35}, '...

[0038] '{"layer_type": "base course", "thickness_cm": 35, "material": "cement stabilized macadam", "modulus_mpa": 400, "poisson": 0.35}, '...

[0039] '{"layer_type": "base course", "thickness_cm": 35, "material": "cement stabilized macadam", "modulus_mpa": 400, "poisson": 0.35}, '...

[0040] Natural language design requirement example: "Design heavy load highway base course, bottom base course uses graded macadam";

[0041] Step 102, the pre-trained large language model (such as DeepSeek) receives the prompt word text described above, and performs semantic analysis on the input content through the natural language understanding capability built in the large language model to identify key engineering parameters and the relationship between parameters.

[0042] The large language model (LLM) receives the natural language prompt input by the user, and the prompt contains the road structure design requirement; through the pre-trained large language model (such as DeepSeek), the prompt is semantically segmented, and key engineering parameters and the relationship between parameters are extracted or inferred;

[0043] The key engineering parameters include: the thickness of each pavement structure layer and material parameters; material parameters include: elastic modulus and Poisson's ratio , wherein represents the structure layer number;

[0044] The specific processing process of the large language model includes:

[0045] (1) Entity recognition: Large language model identifies key entities related to road engineering in the text, including but not limited to traffic load grade (such as "heavy load"), road type (such as "expressway"), structural layer material (such as "graded crushed stone"), etc.

[0046] (b) Relationship extraction: Large language model identifies semantic and engineering relationships between entities, such as "bottom base layer uses graded crushed stone" is parsed as the material membership relationship between "bottom base layer" and "graded crushed stone";

[0047] (c) Keyword extraction: Based on the above analysis, the large language model extracts a set of keywords that can be used for subsequent parameter mapping, such as "heavy load", "expressway", "graded crushed stone", etc. These keywords will be used as input for engineering parameter mapping.

[0048] Through the above processing, the large language model completes the deep semantic analysis of the user's natural language demand, identifies the key design parameters and their relationships, and provides a structured information foundation for the subsequent standard mapping of parameters.

[0049] Step 103, based on the engineering parameter mapping rule library, map the keywords to standard engineering parameters, and return to the system in JSON format as a parameter vector that can be used for subsequent standard allowable value extraction and finite element modeling.

[0050] The engineering parameter mapping rule library includes load parameter mapping library, material semantic encoder, and structure parameter reasoning module, which is used to map the key engineering parameters identified by the large language model to the engineering parameter database, and generate the vector of standard load P, each pavement structure layer thickness , material parameters (elastic modulus , Poisson's ratio ) that can be used for subsequent standard allowable value query and finite element modeling.

[0051] For example, the load parameter mapping table matches "heavy load expressway" as axle load P=0.7MPa, action radius r=15.3cm;

[0052] The material semantic encoder parses "graded crushed stone" as modulus =180MPa, Poisson's ratio =0.3;

[0053] The structure parameter reasoning module completes the missing bottom base layer thickness h=35cm (according to the soil modulus Es=50MPa).

[0054] Step 104: Generate a finite element modeling script based on the parameter vectors analyzed in Steps 102 and 103. Drive the finite element software to perform finite element modeling, boundary condition application, and response solving of the four-layer road structure (surface layer, base layer, subbase layer, and subgrade), and extract the finite element analysis result vector. :

[0055]

[0056] Including the maximum tensile stress at the bottom of the surface layer Maximum allowable tensile strain at the bottom of the base layer Maximum deflection of the top surface of the roadbed .

[0057] The finite element software is used for finite element modeling. Key engineering parameters are input into the finite element software (such as ABAQUS, MATLAB's built-in PDEToolbox, etc.), and an initial finite element model of the road structure is constructed through an automatic script generation module. This model then drives the finite element software to solve the problem, and the finite element solution results are extracted through post-processing files. .

[0058] Specifically, it includes:

[0059] Material property definition: modulus Poisson's ratio Associated with the cross-sections of each structural layer;

[0060] Load application: According to the BZZ-100 standard, a uniformly distributed load P=0.7MPa is applied to the wheel track area with a radius of action r=15.3cm;

[0061] Mesh generation: Generate a refined model by dividing the current road surface structure into meshes based on its dimensions and boundary conditions;

[0062] Solution and Result Extraction: The finite element simulation software was used to solve the problem and extract the maximum tensile stress at the bottom of the structural layers, including the surface layer. Maximum allowable tensile strain at the bottom of the base layer Maximum deflection of the top surface of the roadbed The response value.

[0063] Step 105: Query the standard values. Based on the key design parameters (especially traffic load level and pavement structure combination type) obtained from Steps 102 and 103, calculate the design traffic volume N based on JTG D50-2017 "Specification for Design of Highway Asphalt Pavement" and the input pavement structure parameters. e According to the design traffic volume N e Obtain the corresponding allowable standard value vector from the design parameters. (The calculation method is based on existing technology):

[0064]

[0065] wherein,

[0066] represents the allowable tensile strain at the bottom of the surface layer (unit: MPa) for controlling fatigue cracking;

[0067] represents the allowable tensile stress at the bottom of the base layer (unit: ) for controlling structural damage;

[0068] represents the allowable deflection at the top surface of the subgrade (unit: mm) ， for controlling overall deformation.

[0069] Specifically, a specification standard value database is used to query the core design standard data corresponding to the input pavement structure: in which the “Highway Asphalt Pavement Design Specification” (JTG D50-2017) is pre-installed, and the key design parameters parsed from the large language model (LLM) and the engineering parameter mapping rule library are received, including the traffic load level (such as extremely heavy, heavy, medium, and light traffic), and the pavement structure combination type (such as typical structure of asphalt pavement). Based on the above parameters, the “Highway Asphalt Pavement Design Specification” (JTG D50-2017) is queried and the allowable standard value vector corresponding to the design condition specified in the specification is output , specifically including: the allowable tensile strain at the bottom of the surface layer (for controlling fatigue cracking), the allowable tensile stress at the bottom of the base layer (mainly for controlling structural damage), and the allowable deflection at the top surface of the subgrade (mainly for controlling overall deformation).

[0070] Step 106, the deviation between the finite element analysis result and the specification allowable standard value is taken as the input of the reinforcement learning agent, and the state space s of the reinforcement learning agent is constructed, as shown in Figure 3 . The reinforcement learning algorithm (PPO) of the proximal policy optimization is used to iteratively adjust the thickness of the structure layer, realizing the optimization of the pavement structure.

[0071] The deviation between the finite element analysis result and the standard allowable value is used to construct the state space s of the reinforcement learning agent as follows:

[0072]

[0073] wherein,

[0074] 、 、 respectively are the stress value, strain value and displacement value obtained by finite element modeling respectively;

[0075] 、 respectively are the allowable standard stress value, allowable standard strain value and allowable standard displacement value obtained by querying the standard value database;

[0076] The deviation of the finite element analysis result from the standard allowable value obtained by querying the standard is used to construct a reinforcement learning state space as the input of the agent;

[0077] Step 107, a reward function for road structure design optimization is set in the reinforcement learning agent, and a training stability guarantee mechanism is set, which is used to train the generated agent action suggestion.

[0078] Specifically:

[0079] Considering the safety, economy, dynamic penalty and process stability in the design of road structure, by adopting a weight distribution mechanism of safety first, economy second, penalty and smoothing auxiliary, the reward function of reinforcement learning is constructed as follows:

[0080]

[0081] Among them 、 、 、 respectively are the safety item, economic item, dynamic penalty item and smoothing item.

[0082] The safety item is designed to quantify the deviation of the finite element analysis result from the standard allowable value to evaluate the safety of the structure. In the form of , when the standard ( ≤ ) is fully met, the item gets full score weight, and when the standard is exceeded ( > ), it is deducted according to the proportion. The weight distribution reflects the priority: the stress control weight is the highest ( ), the strain is the second ( ), and the displacement / settlement is the lowest ( ), and the formula is as follows:

[0083]

[0084] The economic item The design measures the ratio of actual material cost to theoretical cost by an exponential function. The lower the actual cost, the higher the reward value. The design method uses nonlinear incentives to strengthen the reward difference of low-cost solutions in an exponential form, accelerating convergence. It considers the product relationship of unit price , thickness , modulus , accurately reflects the driving action of cost, and can optimize material consumption and combination in design, and reduce engineering cost, as follows:

[0085]

[0086] wherein, , is the measured material cost (yuan / m2) of the current design scheme, is the unit price of the i-th layer material (yuan / MPa·m3), , are the thickness and modulus of the i-th layer, respectively;

[0087] Dynamic penalty term Design: When any finite element extraction exceeds 1.15 times of its specification allowable value, an additional linear penalty is applied. The penalty coefficient is set to be the most severe for stress term ( ), followed by strain ( ), and displacement / settlement is the lowest ( ), i.e.:

[0088]

[0089] Smooth term The design is used to evaluate the fluctuation degree of thickness change in the last 3 iterations. The smaller the fluctuation, the higher the reward. This design scheme suppresses parameter mutation, thereby avoiding oscillatory convergence, effectively ensuring the stability of reinforcement learning. At the same time, this design can ensure gradual optimization of the design scheme, avoiding increased construction difficulty or unstable structure performance due to dramatic changes in layer thickness. For example, if the base layer thickness fluctuates more than 5% in adjacent iterations, the system will reduce the learning rate to smooth the adjustment.

[0090]

[0091] To ensure the rationality of action recommendations, a reinforcement learning agent training stability guarantee mechanism is set, including rollback mechanism, adaptive learning rate, and policy entropy constraint.

[0092] Wherein, the rollback mechanism design detects plastic strain εp≥2% or layer thickness ratio >2.0 by finite element analysis, terminates the current training period and restores the optimal state of the last 5 feasible schemes from the history cache; The mechanism design can timely terminate the parameter adjustment that may lead to structural failure (such as plastic deformation caused by too thin base layer), so as to realize risk avoidance; At the same time, it can avoid wasting computing resources on invalid paths, improve training efficiency and realize resource optimization. In actual road structure design, if the algorithm excessively reduces the thickness of the base layer (for example, from 30cm to 25cm) in pursuit of economy, it may lead to imbalance of layer thickness ratio, and this mechanism will automatically return to the historical feasible state to ensure that the design meets the interlayer compatibility requirements;

[0093] Adaptive learning rate design: In the PPO algorithm, the learning rate is designed as follows:

[0094]

[0095] Wherein is the preset maximum training step number, t is the current training step number (t ).

[0096] This design method can realize fast exploration of parameter space at the beginning of design and fine adjustment in the later stage, and can also avoid the oscillation or premature convergence caused by fixed learning rate;

[0097] Strategy entropy constraint: set the target KL divergence threshold to 0.02, that is, when the actual KL divergence , automatically reduce the learning rate as follows:

[0098]

[0099] This design method can limit the amplitude of policy update and prevent mutation, while automatically suppressing the learning rate when the KL divergence exceeds the standard, maintaining the stability of training and realizing adaptive adjustment.

[0100] The objective function of PPO algorithm is:

[0101]

[0102] Wherein,

[0103] Is the objective function of PPO, which is the optimization objective of policy parameter Therefore, we hope to maximize the value in operation;

[0104] Is the importance sampling ratio, and In the formula, represents the probability of selecting action a t under state s t ​ represents the old policy in state s t selects action a t at state s

[0105] is the expectation for time step t, usually the average value in a batch of sampling data;

[0106] is the advantage function estimate, representing the good or bad of action a t in state s t relative to the average performance;

[0107] is the clipping function, which limits to the interval [ This operation is the core of the PPO policy, which is used to prevent the policy update from being too aggressive, is the clipping parameter.

[0108] Step 108, the agent outputs reinforcement learning action suggestion , combined with the action suggestion given by the large language model, the road structure layer adjustment scheme is output, and a new road structure layer is formed. Among them:

[0109] The reinforcement learning agent action suggestion is as follows:

[0110] According to the constructed state space and environmental interaction, the reinforcement learning agent outputs the action for each layer structure, and the original adjustment value is as follows:

[0111]

[0112] Among them, ;

[0113] are the original adjustment values of the surface layer, base layer and sub-base layer thickness respectively; are the original adjustment values of the surface layer, base layer and sub-base layer modulus respectively;

[0114] It should be noted that in the process of road structure design, the soil foundation is often regarded as a semi-infinite body and only used for checking calculation without participating in actual design, so the reinforcement learning agent only iteratively optimizes the surface layer, base layer and sub-base layer in the road structure.

[0115] The layer thickness adjustment value calculation formula is:

[0116]

[0117] Among them, represent the surface layer, base layer and sub-base layer respectively;

[0118] The modulus adjustment calculation formula is:

[0119]

[0120] wherein, , respectively represent the surface layer, the base layer, and the subbase layer;

[0121] The reinforcement learning agent action suggestion is composed of the calculated layer thickness adjustment value and the modulus adjustment value:

[0122]

[0123] Large language model action suggestion As follows:

[0124] Convert the reinforcement learning state space s into a natural language description, construct a prompt word, and send it to the large language model:

[0125] Wherein, the large language model action suggestion prompt word template is as follows:

[0126] "You are a senior road structure engineer. Focus on structural performance, load distribution, and JTG D50-2017 specification compliance.

[0127] The current structure state is:

[0128] Stress utilization rate: {σ_FEA / σ_std × 100:.1f}%

[0129] Strain utilization rate: {ε_FEA / ε_std × 100:.1f}%

[0130] Deflection utilization rate: {D_FEA / D_std × 100:.1f}%,,

[0131] Current:

[0132] Surface layer: {surface layer thickness:.1f} cm, {surface layer modulus:d} MPa

[0133] Base layer: {base layer thickness:.1f} cm, {base layer modulus:d} MPa

[0134] Subbase layer: {subbase layer thickness:.1f} cm, {subbase layer modulus:d} MPa,

[0135] Current cost: {current total cost:.0f} yuan / m², please provide structure optimization suggestions focusing on JTG D50-2017 specification compliance. Please return in JSON format:

[0136] {"thickness_adjustments": [t1,t2,t3],"modulus_adjustments": [m1,m2,m3],"engineering_analysis": {"design_recommendation": "Specific recommendations", "priority_layer": "Priority layer", "structural_concern": "Main problem", "compliance_status": "Compliance status"}}. Adjustment values range between -0.15 and 0.15.

[0137] The large language model will return a JSON response, for example:

[0138] { "thickness_adjustments": [-0.05, 0.08, 0.03], "modulus_adjustments": [-0.10, 0.05, 0.00], "engineering_analysis": { "design_recommendation": "Recommend reducing surface layer thickness and increasing base layer thickness to improve stress distribution", "priority_layer": "Base layer", "structural_concern": "Surface layer stress utilization is too high", "compliance_status": "Exceeds JTG D50-2017 stress limit"}}.

[0139] After system analysis, the large language model action vector is formed:

[0140]

[0141] Reinforcement learning and large language model mixed action recommendations:

[0142] According to the mixed strategy defined in the invention content, the reinforcement learning action recommendation and the large language model action recommendation are mixed as follows:

[0143]

[0144] wherein,

[0145] is the mixed recommendation given by the agent to the pavement structure layer adjustment;

[0146] β=0.3 is the preset influence weight of the large language model recommendation;

[0147] The hybrid action suggestion is decomposed into a layer thickness adjustment component and a modulus adjustment component :

[0148]

[0149] wherein,

[0150]

[0151]

[0152] The structure layer parameter updating rule is:

[0153] Based on the hybrid action suggestion , each structure layer parameter is updated according to the following formula:

[0154] Layer thickness updating:

[0155]

[0156] wherein, and are the layer thicknesses before and after updating respectively;

[0157] Modulus updating:

[0158]

[0159] wherein, and are the moduli before and after updating respectively;

[0160] In the reinforcement learning agent, a design progressive action constraint method is set, when the adjacent layer thickness ratio exceeds a preset buffer threshold, that is, when the ratio of the base layer thickness to the bottom base layer thickness satisfies the following formula , the action of increasing the base layer thickness or reducing the bottom base layer thickness is prohibited, wherein, denotes the base layer thickness, denotes the bottom base layer thickness, and the constraint process is shown in Figure 4 ;

[0161] Step 109, according to the action suggestion of step 108, modify and construct a new pavement structure parameter vector, re-perform finite element modeling and input into the reinforcement learning agent for training until the convergence requirement is met. Among them, the convergence requirement is that all key mechanical responses (σ FEA, ε FEA , D FEA ) of the current design scheme after finite element modeling meet the specification allowable standard value (i.e. ≤ ), and in the last N (N=5) iterations, the increase of the comprehensive reward r is less than the threshold value delta (delta=0.01) or the structural layer parameter change is small.

[0162] Step 110, output the optimized road structure design scheme meeting the safety, economy and engineering requirements, and the output content includes:

[0163] The optimized thickness of each structural layer (such as "the base layer is adjusted from 32 cm to 28 cm"), material parameters and safety factors;

[0164] Visual comparison: comparison of theoretical solution and finite element stress nephogram;

[0165] Economic analysis: generate a material usage comparison table, such as "save 15.2 tons of cement / km, cost reduction 14.7%".

[0166] The application constructs a semantic analysis engine by pre-training a large language model, automatically converts user natural language input into standardized loads, materials and structure parameters, and breaks through the limitations of traditional methods relying on manual form filling. The large language model has high accuracy in parameter analysis. As shown in Table 1. The introduction of the large language model can greatly improve efficiency.

[0167] Table 1 Comparison table of parameter analysis accuracy of different large models

[0168]

[0169] The above description is only a description of the preferred embodiments of the application, and is not any limitation on the scope of the application. Any modification or modification made by any person skilled in the art according to the above disclosed technical content should be regarded as an equivalent effective embodiment, and belongs to the protection scope of the technical scheme of the application.

Claims

1. A road structure design method based on a large model and reinforcement learning, characterized by, Includes the following steps: Step 101: The user inputs natural language design requirements, and the natural language design requirements are converted into prompt words according to the preset prompt word template and sent to the pre-trained large language model; Step 102: The pre-trained large language model performs semantic parsing on the input prompt text to identify key engineering parameters and the relationships between parameters; Step 103: Based on the engineering parameter mapping rule library, map keywords to standard engineering parameters, which will serve as parameter vectors for subsequent standard allowable value extraction and finite element modeling. Step 104, generating a finite element modeling script according to the parameter vector parsed in step 102 and step 103, driving the finite element software to perform finite element modeling of the four-layer road structure, boundary condition application and response solving, and extracting a finite element analysis result vector , the four-layer road structure comprising a surface layer, a base layer, a bottom base layer, and a soil base; Step 105, query the specification standard value, according to the key design parameters obtained in steps 102 and 103, based on JTGD50-2017 "Highway Asphalt Pavement Design Specification", calculate the design traffic volume N according to the input pavement structure parameters e , according to the design traffic volume N e , the corresponding specification allowable standard value vector of the design parameter ; Step 106, finite element analysis results deviation from the standard allowable standard value The deviation from the standard allowable standard value is used as the input of the reinforcement learning agent, the state space of the reinforcement learning agent is constructed, the reinforcement learning algorithm PPO of the proximal policy optimization is used to iteratively adjust the thickness of the structure layer, and the pavement structure optimization is realized. Step 107: Set up a reward function and a training stability guarantee mechanism in the reinforcement learning agent to optimize the road structure design, and use them to train and generate action suggestions for the agent. Step 108, the agent outputs a reinforcement learning action suggestion , combined with the action suggestion given by the large language model , the mixed output road structure layer adjustment scheme forms a new road structure layer; Step 109: Based on the action suggestions in Step 108, modify and construct a new road structure parameter vector, re-model the finite element model, and input it into the reinforcement learning agent for training until the convergence requirement is met. Step 110: Output an optimized road structure design scheme that meets safety, economy, and engineering requirements; In step 106, the finite element analysis result deviation from the standard allowable value The reinforcement learning agent state space s is constructed as follows: in, , , respectively are the stress values, strain values and displacement values obtained by finite element modeling of the finite element software; , are the allowable standard stress value, the allowable standard strain value and the allowable standard displacement value respectively obtained by querying the norm standard value database; with the finite element analysis results deviation from the standard tolerance value obtained by querying the specification The reinforcement learning state space is constructed as the agent input. 2.The road structure design method based on a large model and reinforcement learning according to claim 1, wherein, In step 101, the prompt word template is: systemPrompt = ['You are a road engineering expert.'] Based on user input, road structure parameters are parsed and returned strictly in JSON format: '...' '{"traffic_level":"light load / medium load / heavy load / extremely heavy load","road_type":"highway / urban road / rural road / industrial road",' ... '"vehicle_speed_kmh": numerical value, "subgrade_conditions": "soft soil / general soil / hard soil / rock",'... '"subgrade_treatment":"natural soil foundation / improved soil foundation / replacement soil foundation / composite foundation",' ... '"layers":[{"layer_type":"surface layer","thickness_cm":18,"material":"asphalt concrete","modulus_mpa":1200,"poisson":0.25},' ... '{"layer_type":"base layer","thickness_cm":35,"material":"cement-stabilized crushed stone","modulus_mpa":400,"poisson":0.35},' ... '{"layer_type":"substrate","thickness_cm":25,"material":"graded crushed stone","modulus_mpa":180,"poisson":0.40},' ... '{"layer_type":"soil base","thickness_cm":150,"material":"modulus_mpa":50,"poisson":0.45}]}'). 3.The road structure design method based on large model and reinforcement learning according to claim 1, characterized in that, In step 102, the large language model receives a user input natural language prompt word containing road structure design requirements; the pre-trained large language model performs semantic segmentation on the prompt word, extracts or infers key engineering parameters and the relationship between parameters; The key engineering parameters include: thickness of each pavement structure layer and material parameters, the material parameters include: elastic modulus and Poisson's ratio , wherein denotes the structure layer number; In step 103, the engineering parameter mapping rule library includes a load parameter mapping library, a material semantic encoder, and a structure parameter reasoning module, which is used to map the key engineering parameters identified by the large language model to the engineering parameter database, to generate standard loads P for subsequent specification standard allowable value query, finite element modeling, and thicknesses of each pavement structure layer , material parameters of the vector; In step 104, the finite element analysis result vector is: Maximum tensile stress at the bottom of the surface layer Maximum tensile strain at the bottom of the base layer Maximum deflection at the top of the embankment . 4.The road structure design method based on large model and reinforcement learning according to claim 3, characterized in that, In step 105, the specification allows the standard value vector to be: Wherein, represents the allowable tensile strain at the bottom of the surface layer for controlling fatigue cracking, in MPa; Representative of the base of the bottom of the allowable tensile stress, for the control of structural damage, units ; representative subgrade surface deflection ， for controlling the overall deformation, in mm; The specific process is as follows: Using the specification standard value database for querying the core design standard data corresponding to the input pavement structure: In which the "Highway Asphalt Pavement Design Specification" JTG D50-2017 is preset, receiving the key design parameters parsed from the large language model and the engineering parameter mapping rule library, including the traffic load grade, the pavement structure combination type; Based on the above parameters, query the "Highway Asphalt Pavement Design Specification" JTG D50-2017 and output the allowable standard value vector specified in the specification corresponding to the design condition , Specifically including: the allowable tensile strain at the bottom of the surface layer , the allowable tensile stress at the bottom of the base layer , the allowable deflection of the subgrade top surface . 5.The road structure design method based on large model and reinforcement learning according to claim 4, characterized in that, In step 107, the reward function of reinforcement learning is as follows: wherein , , , are a safety term, an economic term, a dynamic penalty term and a smoothing term, respectively; Safety items The formula is as follows: Economic item As follows: Wherein, , is the measured material cost of the current design scheme, unit: yuan / m2; is the unit price of the i-th layer material, unit: yuan / MPa·m3, , are the thickness and modulus of the i-th layer, respectively; Dynamic penalty term The formula is as follows: smooth term Designed to evaluate the degree of fluctuation in thickness change in the last 3 iterations, the smaller the fluctuation, the higher the reward: ​ In order to ensure the rationality of the action suggestion, a reinforcement learning agent training stability guarantee mechanism is set, including: rollback mechanism, adaptive learning rate and policy entropy constraint; Wherein, the rollback mechanism design detects plastic strain εp≥2% or layer thickness ratio >2.0 through finite element analysis, terminates the current training period and restores the optimal state of the last 5 feasible schemes from the history cache; Adaptive learning rate design: in the PPO algorithm, the learning rate is designed as follows: wherein is a preset maximum training step number, t is a current training step number, ; Policy entropy constraint: Set the target KL divergence threshold to 0.02, i.e., when the actual KL divergence is automatically reduced by the learning rate as follows: ； The objective function of PPO algorithm is: where is the importance sampling ratio, where is the probability of the new policy to select action a t in state s t , is the probability of the old policy to select action a t in state s t ; is the advantage function estimate, is the clipping parameter; is the clipping function, which limits to the interval . 6.The road structure design method based on large model and reinforcement learning according to claim 4, characterized in that, In step 108, the reinforcement learning agent action suggestion As follows: According to the interaction between the constructed state space and the environment, the reinforcement learning agent outputs the action for each layer structure, and the original adjustment value is as follows: wherein ; original adjustment values of the thickness of the surface layer, base layer, and sub-base layer, respectively; original adjustment values of the modulus of the surface layer, base layer, and sub-base layer, respectively; The layer thickness adjustment value calculation formula is: wherein , respectively, represent the surface layer, the base layer, the subbase layer; The modulus adjustment calculation formula is: wherein , respectively, represent the surface layer, the base layer, the subbase layer; To reinforce the learning agent action suggestion, the calculated layer thickness adjustment value and the modulus adjustment value are composed: ； Large language model action suggestions As follows: Convert the reinforcement learning state space s into natural language description, construct the prompt word and send it to the large language model; The large language model will return a JSON response, which is parsed by the system to form a large language model action vector: 。 7.The road structure design method based on large model and reinforcement learning according to claim 6, characterized in that, Reinforcement learning action suggestions Large language model action suggestions Mixing by the following formula: Wherein, a hybrid suggestion for the agent to adjust to the road structure layer; β=0.3 is the preset influence weight of large language model suggestion; wherein the mixing action recommendation is decomposed into a layer thickness adjustment component and a modulus adjustment component : Wherein, Structure layer parameter update rule: Based on hybrid action suggestion each structure layer parameter is updated as follows: Layer thickness update: wherein and respectively the layer thickness before and after the update; Modulus update: wherein and Epre and Epost are the modulus before and after updating, respectively; A method of setting a design progressive action constraint in a reinforcement learning agent is provided, wherein when the thickness ratio of adjacent layers exceeds a pre-set buffer threshold, i.e. when the action of increasing the thickness of the base layer or decreasing the thickness of the sub-base layer is prohibited from being performed. 8.The road structure design method based on large model and reinforcement learning according to claim 6, wherein, In step 108, the template of the prompt word is as follows: "You are a senior road structure engineer. Focus on structural performance, load distribution and JTG D50-2017 specification compliance. The current structure state is: Stress utilization rate: {σ_FEA / σ_std × 100:.1f}% Strain utilization rate: {ε_FEA / ε_std × 100:.1f}% Deflection utilization rate: {D_FEA / D_std × 100:.1f}%, current: Surface layer: {surface layer thickness:.1f}cm, {surface layer modulus:d}MPa Base layer: {base layer thickness:.1f}cm, {base layer modulus:d}MPa Subbase layer: {subbase layer thickness:.1f}cm, {subbase layer modulus:d}MPa, Current cost: {current total cost:.0f} yuan / m², please provide structure optimization suggestions focusing on JTG D50-2017 specification compliance. Please return in JSON format: ​ {"thickness_adjustments": [t1, t2, t3], "modulus_adjustments": [m1, m2, m3], "engineering_analysis": {"design_recommendation": "specific recommendations", "priority_layer": "priority layer", "structural_concern": "main issues", "compliance_status": "compliance status"}}. Adjustment values range between -0.15 and 0.

15. 9.The road structure design method based on large model and reinforcement learning according to claim 1, characterized in that, In step 109, the convergence requirement is that all key mechanical responses of the current design scheme, after finite element modeling, meet the allowable standard values ​​specified in the code. Furthermore, in five consecutive iterations, the increase in the overall reward r is less than the threshold δ or the change in the structural layer parameters is small.

Images