Low-carbon construction decision-making method and system based on digital twinning and multi-objective optimization

Abstract

This application discloses a low-carbon construction decision-making method and system based on digital twins and multi-objective optimization. The method includes: acquiring basic data of the target road; constructing a low-carbon construction multi-objective optimization model based on the basic data; solving the low-carbon construction multi-objective optimization model to obtain a Pareto optimal construction strategy set; constructing a road digital twin corresponding to the target road; inputting multiple representative construction strategies from the Pareto optimal construction strategy set into the road digital twin and performing dynamic simulations during the analysis period to obtain simulation results corresponding to each representative construction strategy; evaluating and ranking the multiple representative construction strategies based on the simulation results corresponding to each representative construction strategy, and obtaining and outputting a recommended construction strategy sequence. This application can take into account the carbon emissions throughout the entire life cycle, the economic cost throughout the entire life cycle, and the long-term pavement performance during the analysis period, and verify the construction strategies in conjunction with the long-term service process of the target road.

Description

Technical Field

[0001] This application relates to the interdisciplinary field of road engineering management and digital twin technology, and in particular to a low-carbon construction decision-making method and system based on digital twin and multi-objective optimization. Background Technology

[0002] With the continuous advancement of "dual carbon" goals and the increasing demand for refined management and maintenance of highway infrastructure, the highway construction sector is placing higher demands on the scientific, economical, and low-carbon aspects of construction strategies. The formulation of highway construction plans typically involves multiple factors, including construction timing, construction technology, and material selection. These factors not only affect construction costs and organizational efficiency but also directly impact the long-term performance of the pavement and its life-cycle carbon emission levels. Therefore, how to achieve synergistic optimization of low-carbon, economical, and long-term effective highway construction strategies while meeting road service performance requirements has become an important research direction in the fields of road engineering management and intelligent decision-making.

[0003] In related technologies, highway construction decision-making systems often employ single-objective optimization methods, such as prioritizing minimum cost or using simple weighted methods to comprehensively evaluate multiple indicators like cost and performance. However, these methods typically fail to accurately reflect the complex trade-offs between carbon emissions, economic costs, and long-term pavement performance, easily leading to strategies that excel only in one aspect while neglecting overall benefits. Although some research has attempted to introduce multi-objective optimization methods into the construction decision-making process, the optimization results often remain at the theoretical optimal level, failing to fully consider the long-term dynamic factors and uncertainties such as traffic load growth, climate change, and material price fluctuations, thus making it difficult to accurately assess the actual effectiveness of construction strategies over their service life. On the other hand, while digital twin technology can dynamically simulate and analyze road operating conditions, it is still in its early stages in the field of highway construction. It generally suffers from complex pavement performance degradation mechanisms, difficulty in quantifying the effects of construction interventions, and high long-term simulation costs, making it difficult for existing technologies to form highly reliable construction decision support capabilities for long-term service.

[0004] Therefore, in formulating low-carbon construction strategies for highways, the complex trade-offs between multiple objectives, the difficulty in effectively characterizing long-term dynamic influencing factors, and the lack of reliable verification capabilities for construction strategies in the long-term service process have become urgent problems to be solved. Summary of the Invention

[0005] This application provides a low-carbon construction decision-making method and system based on digital twins and multi-objective optimization, aiming to solve the problems in the formulation of low-carbon highway construction strategies, such as the complex trade-offs between multiple objectives, the difficulty in effectively representing long-term dynamic influencing factors, and the lack of reliable verification capabilities for construction strategies in the long-term service process.

[0006] Firstly, a low-carbon construction decision-making method based on digital twins and multi-objective optimization, the method comprising: Obtain basic data for the target road; A multi-objective optimization model for low-carbon construction is constructed based on the aforementioned basic data. Solving the multi-objective optimization model for low-carbon construction yields a set of Pareto-optimal construction strategies. Construct a digital twin of the target road; Select several representative construction strategies from the Pareto optimal construction strategy set; The representative construction strategies are input into the road digital twin and dynamic simulations are performed during the analysis period to obtain the simulation results corresponding to each representative construction strategy. Based on the simulation results corresponding to each of the representative construction strategies, the multiple representative construction strategies are evaluated and ranked to obtain and output a recommended construction strategy sequence.

[0007] Optionally, in the above scheme, obtaining the basic data of the target road includes: Acquire road structure data, pavement condition data, traffic load data, climate environment data, and construction history data of the target road; Based on the road segment division results of the target road, extract at least one of the following for each road segment: road age, existing road surface condition, traffic volume, climate zone, structural type, and existing construction records. The data corresponding to each road segment are correlated to form the basic dataset used to construct the low-carbon construction multi-objective optimization model.

[0008] Optionally, in the above scheme, the step of constructing a low-carbon construction multi-objective optimization model based on the basic data includes: Use at least one of the following as a construction strategy decision variable: construction timing, construction technology, and material type. The optimization targets are: carbon emissions throughout the entire life cycle, economic costs throughout the entire life cycle, and long-term pavement performance during the analysis period. Introduce at least one of the following constraints: budget constraints, pavement performance constraints, construction period constraints, and material supply constraints; The low-carbon construction multi-objective optimization model is constructed based on the construction strategy decision variables, the optimization objective, and the constraints.

[0009] Optionally, in the above scheme, the multi-objective optimization model for low-carbon construction is solved to obtain a set of Pareto-optimal construction strategies, including: The construction strategy decision variables are coded according to the strategy. Based on the aforementioned strategy encoding method and historical construction cases, the initial population is initialized to obtain the initialized initial population. Based on the initial population after initialization, the improved NSGA-III algorithm is used to iteratively solve the low-carbon construction multi-objective optimization model to obtain the Pareto optimal construction strategy set.

[0010] Optionally, in the above scheme, the initialization of the initial population based on the strategy encoding method and historical construction cases to obtain the initialized initial population includes: Successful construction cases were selected from the historical construction case database; Extract the road features and construction strategy features corresponding to the successful construction cases; Cluster analysis is performed on the road features and construction strategy features corresponding to the successful construction cases to generate typical strategy templates; The initial population is initialized based on the typical strategy template and the strategy encoding method to obtain the initialized initial population; Based on the initialized initial population, the improved NSGA-III algorithm is used to iteratively solve the low-carbon construction multi-objective optimization model to obtain the Pareto optimal construction strategy set, including: Based on the initial population after initialization, crossover and mutation operations are performed, and the candidate construction strategies formed after the crossover and mutation operations are constrained, verified or repaired to update the population and continue to iteratively solve the low-carbon construction multi-objective optimization model to obtain the Pareto optimal construction strategy set.

[0011] Optionally, in the above scheme, constructing the digital twin of the target road includes: A dynamically updatable digital twin of the target road is constructed based on its BIM model, historical inspection data, and real-time monitoring data. The road digital twin integrates a pavement performance degradation prediction model, a traffic load impact model, and a climate and environmental impact model.

[0012] Optionally, in the above scheme, multiple representative construction strategies can be selected from the Pareto optimal construction strategy set, including: Based on the distribution of objective function values, Pareto front distribution, or reference point coverage of each construction strategy in the Pareto optimal construction strategy set, select several representative construction strategies from the Pareto optimal construction strategy set. The process involves inputting the multiple representative construction strategies into the road digital twin and performing dynamic simulations during the analysis period to obtain simulation results corresponding to each representative construction strategy, including: The multiple representative construction strategies are loaded into the road digital twin according to the preset analysis period; Based on the pavement performance degradation prediction model, the traffic load impact model, and the climate environment effect model, dynamic simulations are performed on the multiple representative construction strategies respectively. Output the pavement performance evolution trajectory, cumulative carbon emission results, and cost cash flow results corresponding to each representative construction strategy.

[0013] Optionally, in the above scheme, a pavement performance degradation prediction model, a traffic load impact model, and a climate and environmental impact model can be integrated into the road digital twin, including: The pavement performance degradation prediction model is constructed based on historical test data, and the correlation between pavement performance status and standard axle load, pavement age, and construction intervention is established. The traffic load impact model is constructed based on traffic load data, and the change in traffic load is used as an external input to the road surface performance evolution process. The climate and environmental action model is constructed based on climate and environmental data, and climate and environmental changes are used as external inputs into the pavement performance evolution process. After acquiring new detection data, the parameters of the pavement performance degradation prediction model are calibrated and dynamically corrected based on historical detection data and the new detection data.

[0014] Optionally, in the above scheme, the step of evaluating and ranking the multiple representative construction strategies based on the simulation results corresponding to each representative construction strategy, and obtaining and outputting a recommended construction strategy sequence, includes: For at least one uncertainty factor among traffic volume growth, material price fluctuations, and climate anomalies, uncertainty analysis is performed on the multiple representative construction strategies to obtain the probability distribution of the objective function value corresponding to each representative construction strategy. Based on the probability distribution of the objective function value, the conditional risk value corresponding to each of the representative construction strategies is determined, and a robust ranking result is formed; Based on the simulation results and robust ranking results corresponding to each of the representative construction strategies, the multiple representative construction strategies are evaluated and ranked to obtain the ranking results; The recommended construction strategy sequence is output based on the sorting results.

[0015] In a second aspect, a low-carbon construction decision-making system based on digital twins and multi-objective optimization is provided. The system includes a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, the processor performs the method described in the first aspect.

[0016] Compared with the prior art, this application has at least the following beneficial effects: Based on further analysis and research of existing technologies, this application recognizes that existing technologies suffer from complex trade-offs between multiple objectives, difficulty in effectively representing long-term dynamic influencing factors, and a lack of reliable verification capabilities for long-term service processes in the formulation of low-carbon construction strategies for highways. By acquiring basic data of the target road and constructing a multi-objective optimization model for low-carbon construction based on this data, construction decisions are no longer limited to a single cost objective or simple weighted evaluation. Instead, a unified multi-objective decision-making relationship is established between life-cycle carbon emissions, life-cycle economic costs, and long-term pavement performance during the analysis period. This addresses the problem of complex trade-offs between multiple objectives and difficulty in considering overall benefits in existing technologies at the strategy generation stage. Furthermore, this application obtains a Pareto-optimal set of construction strategies by solving the multi-objective optimization model for low-carbon construction, and then selects multiple representative construction strategies from this set, avoiding the problem of existing technologies only outputting a single static solution. The limitations of decision-making allow candidate strategies to cover the feasible solution space under different objective trade-offs. Based on this, this application constructs a digital twin of the target road and inputs multiple representative construction strategies into the road digital twin for dynamic simulation during the analysis period. This yields simulation results for each representative construction strategy, enabling the evaluation of construction strategies to move beyond theoretical optimization results and instead combine the dynamic evolution of the target road during the analysis period to perform long-term verification of each strategy. This addresses the problems in existing technologies where long-term dynamic influencing factors are difficult to effectively characterize and construction strategies lack reliable verification capabilities for long-term service. Finally, based on the simulation results corresponding to each representative construction strategy, this application evaluates and ranks the multiple representative construction strategies, obtaining and outputting a recommended construction strategy sequence. The final output is based on both multi-objective optimization generation and long-term dynamic simulation verification, thus enabling low-carbon construction decisions for the target road. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a low-carbon construction decision-making method based on digital twins and multi-objective optimization, provided as an embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0019] In one embodiment, such as Figure 1As shown, a low-carbon construction decision-making method based on digital twins and multi-objective optimization is provided, including the following steps: Obtain basic data for the target road; A multi-objective optimization model for low-carbon construction is constructed based on the aforementioned basic data. Solving the multi-objective optimization model for low-carbon construction yields a set of Pareto-optimal construction strategies. Construct a digital twin of the target road; Select several representative construction strategies from the Pareto optimal construction strategy set; The representative construction strategies are input into the road digital twin and dynamic simulations are performed during the analysis period to obtain the simulation results corresponding to each representative construction strategy. Based on the simulation results corresponding to each of the representative construction strategies, the multiple representative construction strategies are evaluated and ranked to obtain and output a recommended construction strategy sequence.

[0020] This embodiment provides a low-carbon construction decision-making method based on digital twins and multi-objective optimization. This method, oriented towards low-carbon construction decision-making scenarios for target roads, combines multi-objective optimization solutions, dynamic simulation of road digital twins, and evaluation and ranking based on simulation results to form a complete technical chain from candidate construction strategy generation to recommended construction strategy output.

[0021] In this embodiment, the basic data of the target road is first acquired. This basic data can be derived from road design data, operation and maintenance data, detection systems, monitoring systems, and historical construction records. The basic data is used to characterize the current structural state, service status, traffic load level, climatic conditions, and existing construction status of the target road, thus providing an input basis for the subsequent construction of a multi-objective optimization model for low-carbon construction.

[0022] After obtaining the basic data, a multi-objective optimization model for low-carbon construction is constructed based on this data. This model uses construction decision-making content as the optimization object, carbon emissions, economic costs, and long-term pavement performance as optimization objectives, and incorporates practical engineering constraints such as budget, construction period, material supply, and minimum performance requirements to form a multi-objective optimization problem. This model can describe the differences between different construction strategies across multiple objective dimensions.

[0023] Subsequently, the multi-objective optimization model for low-carbon construction is solved to obtain a Pareto-optimal set of construction strategies. The solution process searches for non-dominated solutions that satisfy the constraints within a multi-dimensional construction strategy space, thus forming several candidate construction strategies. Each construction strategy in the Pareto-optimal set establishes different balances between objectives such as carbon emissions, cost, and performance, providing a candidate set for further analysis.

[0024] After obtaining the Pareto optimal construction strategy set, a digital twin of the target road is constructed. This digital twin reflects the structural, state, and external influence information of the target road, supporting subsequent construction strategy loading and long-term dynamic simulation. The digital twin can integrate pavement performance degradation prediction models, traffic load impact models, and climate environment impact models to achieve dynamic evolution representation of the road's service status.

[0025] After the digital twin of the road is constructed, several representative construction strategies are selected from the Pareto optimal construction strategy set. The representative construction strategies can be selected based on the distribution of the Pareto front, the distribution of the objective function value, or the coverage of reference points, so as to maintain coverage of the distribution characteristics of the Pareto optimal construction strategy set while controlling the amount of simulation computation.

[0026] Subsequently, the multiple representative construction strategies are input into the road digital twin for dynamic simulation during the analysis period. The analysis period can be set to several years, such as 10, 15, or 20 years, depending on the construction decision-making requirements. During the dynamic simulation, construction strategies can be loaded into the road digital twin at preset time steps to simulate the evolution of pavement conditions, the accumulation of carbon emissions, and the cost changes under the action of each representative construction strategy, and the simulation results corresponding to each representative construction strategy are obtained.

[0027] Finally, based on the simulation results corresponding to each representative construction strategy, the strategies are evaluated and ranked to obtain and output a recommended construction strategy sequence. The evaluation and ranking can be combined with performance evolution, cumulative carbon emissions, cost cash flow, and subsequent uncertainty analysis results from the simulation results, ultimately forming a recommended construction strategy sequence suitable for the target road.

[0028] This embodiment establishes a complete decision-making process of "multi-objective optimization to generate candidate strategies, digital twin verification of representative strategies, and output of recommended sequences based on simulation results". This enables low-carbon construction decisions to consider carbon emissions, economic costs and long-term performance at the same time, and to be verified in conjunction with long-term service processes, thereby improving the systematicness and completeness of construction decisions.

[0029] In this embodiment, obtaining the basic data of the target road includes: Acquire road structure data, pavement condition data, traffic load data, climate environment data, and construction history data of the target road; Based on the road segment division results of the target road, extract at least one of the following for each road segment: road age, existing road surface condition, traffic volume, climate zone, structural type, and existing construction records. The data corresponding to each road segment are correlated to form the basic dataset used to construct the low-carbon construction multi-objective optimization model.

[0030] This implementation method further explains the process of acquiring basic data for the target road. To ensure that the subsequent multi-objective optimization model for low-carbon construction and the digital twin of the road have sufficient data support, it is necessary to collect and construct a basic dataset from multiple dimensions.

[0031] During the data acquisition phase, road structure data, pavement condition data, traffic load data, climate environment data, and construction history data of the target road can be collected. Road structure data may include road grade, structural layer composition, thickness information, material information, BIM model information, etc.; pavement condition data may include pavement condition index, smoothness, damage rate, rutting, cracks, skid resistance index, etc.; traffic load data may include traffic volume, vehicle type composition, standard axle load conversion results, overloading situation, and annual growth trend, etc.; climate environment data may include temperature, rainfall, freeze-thaw cycle, humidity, solar radiation, etc.; construction history data may include existing construction time, construction technology type, material type, construction scope, and post-construction effects, etc.

[0032] Furthermore, based on the road segmentation results of the target road, at least one of the following can be extracted for each road segment: road age, existing pavement condition, traffic volume, climate zoning, structural type, and existing construction records. Road segmentation can be based on road mileage division, defect distribution characteristics, structural type boundaries, or management unit boundaries. Through road segmentation, it is beneficial to form more refined construction decision-making units.

[0033] After completing the collection of various basic data, the data corresponding to each road segment can be correlated and organized to form a basic dataset for building a multi-objective optimization model for low-carbon construction. This basic dataset can be stored in the form of tables, database records, or structured files, and data indexing and mapping can be completed according to road segment identification. For missing data, preprocessing methods such as interpolation, statistical completion, or historical pattern correction can be used.

[0034] This embodiment provides multi-dimensional data support for the construction of multi-objective optimization models for low-carbon construction and digital twin modeling of roads by building a basic dataset containing structural, state, load, environmental, and historical construction information. This helps to improve the accuracy and applicability of subsequent construction strategy generation and verification processes.

[0035] In this embodiment, the construction of a low-carbon construction multi-objective optimization model based on the basic data includes: Use at least one of the following as a construction strategy decision variable: construction timing, construction technology, and material type. The optimization targets are: carbon emissions throughout the entire life cycle, economic costs throughout the entire life cycle, and long-term pavement performance during the analysis period. Introduce at least one of the following constraints: budget constraints, pavement performance constraints, construction period constraints, and material supply constraints; The low-carbon construction multi-objective optimization model is constructed based on the construction strategy decision variables, the optimization objective, and the constraints.

[0036] This embodiment further explains the construction process of the multi-objective optimization model for low-carbon construction. This model is used to abstract the low-carbon construction problem of the target road into a multi-objective optimization problem with multiple decision variables, multiple optimization objectives, and multiple constraints.

[0037] In terms of setting decision variables, at least one of the following can be used as decision variables for construction strategy: construction timing, construction technology, and material type. Construction timing can be represented as the target year, time window, or construction cycle for a certain road segment; construction technology can include preventive construction technology, corrective construction technology, structural construction technology, and recyclable construction technology; material type can include materials with different gradations, recycled materials, low-carbon materials, and modified materials. For different road segments, a joint decision variable vector can be constructed to describe the combination of construction strategies for the entire target road or target road network.

[0038] In terms of setting optimization targets, the life-cycle carbon emissions, life-cycle economic costs, and long-term pavement performance during the analysis period can be used as optimization targets. Life-cycle carbon emissions can comprehensively consider carbon emissions from material production, transportation, construction, construction operations, and subsequent repeated construction processes; life-cycle economic costs can comprehensively consider material costs, construction costs, equipment costs, and subsequent expenditures after discounting; long-term pavement performance during the analysis period can be characterized by the average value, weighted value, or comprehensive performance index of the pavement condition index during the analysis period.

[0039] In terms of setting constraints, at least one of the following can be introduced: budget constraints, pavement performance constraints, construction period constraints, and material supply constraints. Budget constraints can be used to limit cumulative construction expenditures within an year or analysis period; pavement performance constraints can be used to ensure that the road meets minimum performance requirements within the analysis period; construction period constraints can be used to limit the duration of a single construction project or the annual construction window; and material supply constraints can be used to limit the availability or usage ratio of specific materials.

[0040] Based on the aforementioned construction strategy decision variables, optimization objectives, and constraints, a multi-objective optimization model for low-carbon construction can be constructed. In practical implementation, this model can be represented using a vectorized approach and solved by subsequent optimization algorithms.

[0041] This embodiment unifies the expression of construction decision variables, carbon emission targets, economic cost targets, long-term performance targets, and actual engineering constraints in the same model, enabling low-carbon construction problems to have a computable and optimizable modeling foundation, which is conducive to generating feasible candidate construction strategies under multi-objective conflict conditions.

[0042] In this embodiment, the multi-objective optimization model for low-carbon construction is solved to obtain a set of Pareto optimal construction strategies, including: The construction strategy decision variables are coded according to the strategy. Based on the aforementioned strategy encoding method and historical construction cases, the initial population is initialized to obtain the initialized initial population. Based on the initial population after initialization, the improved NSGA-III algorithm is used to iteratively solve the low-carbon construction multi-objective optimization model to obtain the Pareto optimal construction strategy set.

[0043] This implementation further explains the solution process of the multi-objective optimization model for low-carbon construction. To obtain the Pareto optimal set of construction strategies, it is necessary to first encode the decision variables of the construction strategies, then initialize the population using historical construction cases, and finally use the improved NSGA-III algorithm for iterative solution.

[0044] First, the construction strategy decision variables are coded using strategies. The coding method can be set according to the composition of the construction decision variables; for example, road segment numbers, construction timing, construction technology, and material type can be coded as fixed-length chromosomes, or segmented coding, integer coding, or combined coding can be used. Through strategy coding, the original construction decision variables can be transformed into a computational representation suitable for genetic evolutionary solutions.

[0045] Secondly, based on the strategy encoding method and historical construction cases, the initial population is initialized to obtain an initialized initial population. This process can utilize existing strategy information from successful historical construction cases, avoiding the use of only random methods to generate initial solutions. The initialized initial population can ensure solution space coverage while making the initial candidate solutions closer to the strategy patterns that have been verified to be feasible in actual engineering.

[0046] After forming an initial population, an improved NSGA-III algorithm is used to iteratively solve the multi-objective optimization model for low-carbon construction. During the iterative solution process, fitness calculations, non-dominated sorting, reference point association, crossover operations, mutation operations, and population updates are performed on individuals within the population to continuously search for optimal solutions that satisfy the constraints. After multiple rounds of iterative evolution, a Pareto-optimal set of construction strategies is obtained.

[0047] This embodiment achieves an effective solution to the multi-objective optimization model for low-carbon construction through a continuous process of "strategy encoding, initial population initialization, and improved NSGA-III iterative solution," which is beneficial for obtaining a Pareto optimal construction strategy set that covers the balance relationship of multiple optimization objectives.

[0048] In this embodiment, the initialization of the initial population based on the strategy encoding method and historical construction cases to obtain the initialized initial population includes: Successful construction cases were selected from the historical construction case database; Extract the road features and construction strategy features corresponding to the successful construction cases; Cluster analysis is performed on the road features and construction strategy features corresponding to the successful construction cases to generate typical strategy templates; The initial population is initialized based on the typical strategy template and the strategy encoding method to obtain the initialized initial population; Based on the initialized initial population, the improved NSGA-III algorithm is used to iteratively solve the low-carbon construction multi-objective optimization model to obtain the Pareto optimal construction strategy set, including: Based on the initial population after initialization, crossover and mutation operations are performed, and the candidate construction strategies formed after the crossover and mutation operations are constrained, verified or repaired to update the population and continue to iteratively solve the low-carbon construction multi-objective optimization model to obtain the Pareto optimal construction strategy set.

[0049] This implementation further explains the initial population initialization and the iterative solution process of the improved NSGA-III algorithm. To make the initial population more consistent with actual construction patterns, typical strategy templates can be extracted from historical successful construction cases and combined with strategy encoding to complete population initialization; at the same time, in subsequent iterations, the population is continuously updated through crossover, mutation, and constraint verification or repair.

[0050] During the initialization phase, successful construction cases can be selected from a historical construction case library. These successful cases can be historical projects that achieved the intended construction results, met budget requirements, and demonstrated good service performance. The case library can consist of a regional construction database, historical construction archives, or a compilation of cross-project cases.

[0051] Subsequently, road characteristics and construction strategy characteristics corresponding to successful construction cases are extracted. Road characteristics may include road age, traffic volume, current performance indicators, climate zoning, and structural layer type; construction strategy characteristics may include the timing of construction, type of construction technology, type of materials, and post-construction effect data. By extracting these characteristics, a feature set can be formed for case classification and strategy summarization.

[0052] Then, cluster analysis is performed on the road features and construction strategy features to generate typical strategy templates. Clustering methods can include K-means clustering, hierarchical clustering, or other clustering methods. Typical strategy templates can be understood as several common and successful combinations of construction strategies, with each template corresponding to a set of typical parameter configurations.

[0053] After generating a typical policy template, the initial population is initialized based on the template and the policy encoding method, resulting in an initialized initial population. This method preserves a certain degree of diversity while incorporating engineering experience from successful historical cases, which is beneficial for improving subsequent solution efficiency.

[0054] During the iterative solution phase, crossover and mutation operations are performed on the initial population after initialization. Constraint verification or repair processing is then applied to the candidate construction strategies generated after crossover and mutation. Crossover operations can exchange construction timing, construction techniques, or material types for certain road segments among candidate strategies; mutation operations can locally perturb the construction parameters of a specific road segment. Constraint verification checks whether candidate construction strategies meet conditions such as budget, schedule, performance, and material supply; repair processing adjusts individuals that do not meet the constraints to generate feasible strategies. After repair, the population is updated, and the multi-objective optimization model for low-carbon construction continues iteratively until the termination condition is met.

[0055] This embodiment combines historical successful construction cases, typical strategy templates, and population evolution mechanisms to improve the quality of the initial population and maintain the feasibility and diversity of solutions during the iterative solution process, thereby providing a more stable solution process for obtaining the Pareto optimal set of construction strategies.

[0056] In this embodiment, constructing the digital twin of the target road includes: A dynamically updatable digital twin of the target road is constructed based on its BIM model, historical inspection data, and real-time monitoring data. The road digital twin integrates a pavement performance degradation prediction model, a traffic load impact model, and a climate and environmental impact model.

[0057] This embodiment further explains the construction process of the road digital twin. The road digital twin is used to carry out the time-series loading and dynamic simulation during the analysis period of representative construction strategies. Therefore, its construction should simultaneously consider the road structure, historical status, real-time status, and external factors.

[0058] When constructing a digital twin of a road, modeling can be based on the BIM model of the target road, historical inspection data, and real-time monitoring data. The BIM model provides static information such as road geometry, structural layer composition, component relationships, and engineering attributes; historical inspection data provides information on the road's performance status over the years, distribution of defects, and deterioration patterns; and real-time monitoring data provides dynamic information such as traffic flow, environmental conditions, and sensor data. By integrating the above data, a dynamically updatable digital twin of the road can be constructed.

[0059] Furthermore, a pavement performance degradation prediction model, a traffic load impact model, and a climate environment effect model are integrated into the road digital twin. The pavement performance degradation prediction model describes the changes in pavement performance over time and with construction interventions; the traffic load impact model describes the influence of traffic load changes on pavement condition evolution; and the climate environment effect model describes the impact of environmental factors such as temperature, humidity, rainfall, and freeze-thaw cycles on pavement deterioration. The three models work together to enable the road digital twin to simulate the service performance of the target road under different construction strategies during the analysis period.

[0060] This embodiment integrates BIM models, historical inspection data, and real-time monitoring data, and integrates pavement performance degradation prediction models, traffic load impact models, and climate environment effect models into the road digital twin, providing a unified digital carrier for long-term dynamic simulation of representative construction strategies.

[0061] In this embodiment, several representative construction strategies are selected from the Pareto optimal construction strategy set, including: Based on the distribution of objective function values, Pareto front distribution, or reference point coverage of each construction strategy in the Pareto optimal construction strategy set, select several representative construction strategies from the Pareto optimal construction strategy set. The process involves inputting the multiple representative construction strategies into the road digital twin and performing dynamic simulations during the analysis period to obtain simulation results corresponding to each representative construction strategy, including: The multiple representative construction strategies are loaded into the road digital twin according to the preset analysis period; Based on the pavement performance degradation prediction model, the traffic load impact model, and the climate environment effect model, dynamic simulations are performed on the multiple representative construction strategies respectively. Output the pavement performance evolution trajectory, cumulative carbon emission results, and cost cash flow results corresponding to each representative construction strategy.

[0062] This embodiment further explains the selection of representative construction strategies and the dynamic simulation process. In order to balance computational efficiency and candidate strategy coverage, multiple representative construction strategies can be selected from the Pareto optimal construction strategy set and loaded into the road digital twin for separate simulation.

[0063] During the selection of representative construction strategies, multiple representative construction strategies can be selected based on the distribution of objective function values, Pareto front distribution, or reference point coverage of each construction strategy in the Pareto optimal construction strategy set. If there are many candidate strategies on the Pareto front, strategies that are representative of different objective trade-off regions can be prioritized to ensure that subsequent simulations can cover different solution domains such as high-cost low-carbon regions, low-cost high-performance regions, and compromise regions.

[0064] During the dynamic simulation phase, multiple representative construction strategies are loaded into the road digital twin according to a preset analysis period. The loading process can be performed according to annual, quarterly, or other preset time steps, inputting the corresponding construction timing, construction technology, and material configuration for each representative construction strategy into the road digital twin. Different strategies should be simulated independently to obtain comparable results between strategies.

[0065] Based on pavement performance degradation prediction models, traffic load impact models, and climate environment effect models, dynamic simulations are performed on several representative construction strategies. During the simulation, pavement performance status, cumulative carbon emission status, and cost status are continuously updated until the end of the analysis period. The final outputs are the pavement performance evolution trajectory, cumulative carbon emission results, and cost cash flow results for each representative construction strategy.

[0066] This embodiment selects representative construction strategies from the Pareto optimal construction strategy set and performs long-term dynamic simulations on each strategy in a road digital twin. This reduces the simulation scale while maintaining the representativeness of the strategies and provides simulation results that correspond one-to-one with each representative construction strategy for subsequent evaluation and ranking.

[0067] In this embodiment, the road digital twin integrates a pavement performance degradation prediction model, a traffic load impact model, and a climate and environmental impact model, including: The pavement performance degradation prediction model is constructed based on historical test data, and the correlation between pavement performance status and standard axle load, pavement age, and construction intervention is established. The traffic load impact model is constructed based on traffic load data, and the change in traffic load is used as an external input to the road surface performance evolution process. The climate and environmental action model is constructed based on climate and environmental data, and climate and environmental changes are used as external inputs into the pavement performance evolution process. After acquiring new detection data, the parameters of the pavement performance degradation prediction model are calibrated and dynamically corrected based on historical detection data and the new detection data.

[0068] This implementation further explains the construction and parameter correction methods of the three core models in the road digital twin. By constructing a pavement performance degradation prediction model, a traffic load impact model, and a climate environment effect model respectively, and updating the model parameters after new detection data arrives, the adaptability of the road digital twin in long-term simulation can be improved.

[0069] In constructing a pavement performance degradation prediction model, the correlation between pavement performance status and standard axle load, pavement age, and construction intervention can be established based on historical monitoring data. In practice, mechanistic models, empirical models, or hybrid models combining mechanistic and data-driven approaches can be used to describe this relationship. The model can incorporate current performance indicators, cumulative standard axle load exposure times, pavement age, and construction intervention effect parameters to reflect the performance changes of the pavement under the combined effects of external loads and maintenance activities.

[0070] In constructing a traffic load impact model, the influence relationships of parameters such as traffic flow, vehicle type structure, and standard axle load conversion results on pavement performance evolution can be established based on traffic load data. Traffic load changes can then be used as an external input to the pavement performance evolution process. In practical implementation, factors such as annual traffic volume growth, changes in the proportion of heavy loads, and periodic traffic fluctuations can be considered.

[0071] In constructing a climate and environmental impact model, the influence relationships of environmental factors such as temperature, rainfall, humidity, and freeze-thaw cycles on pavement performance evolution can be established based on climate and environmental data. Climate and environmental changes can be used as external inputs into the pavement performance evolution process. Different climate zones can be assigned different environmental impact parameters to accommodate regional differences.

[0072] After acquiring new detection data, the pavement performance degradation prediction model can be calibrated and dynamically corrected based on historical and new detection data. Parameter calibration can employ least squares, Bayesian updates, or other parameter identification methods to ensure that the model's prediction results continuously closely approximate the actual road condition. Dynamic correction can improve the reliability of the road digital twin in long-term simulations.

[0073] This embodiment improves the ability of the road digital twin to characterize the evolution of road service status by constructing and coupling the pavement performance degradation prediction model, traffic load influence model and climate environment effect model respectively, and continuously correcting key model parameters by combining new detection data.

[0074] In this embodiment, the step of evaluating and ranking the multiple representative construction strategies based on the simulation results corresponding to each representative construction strategy, and obtaining and outputting a recommended construction strategy sequence, includes: For at least one uncertainty factor among traffic volume growth, material price fluctuations, and climate anomalies, uncertainty analysis is performed on the multiple representative construction strategies to obtain the probability distribution of the objective function value corresponding to each representative construction strategy. Based on the probability distribution of the objective function value, the conditional risk value corresponding to each of the representative construction strategies is determined, and a robust ranking result is formed; Based on the simulation results and robust ranking results corresponding to each of the representative construction strategies, the multiple representative construction strategies are evaluated and ranked to obtain the ranking results; The recommended construction strategy sequence is output based on the sorting results.

[0075] This implementation further explains the evaluation and ranking process of representative construction strategies. Based on the simulation results output by the road digital twin, this process further incorporates uncertainty analysis, conditional value of risk, and robustness ranking results to form a more robust sequence of recommended construction strategies.

[0076] First, uncertainty analysis is performed on multiple representative construction strategies, considering at least one uncertainty factor among traffic volume growth, material price fluctuations, and climate anomalies. In implementation, probability distributions or scenario distributions can be set for each uncertainty factor, such as setting the range of traffic volume growth rate, material price fluctuation range, and the probability of extreme weather occurrence. Subsequently, under these uncertainties, the objective function values ​​of the representative construction strategies can be repeatedly calculated or simulated to obtain the probability distribution of the objective function values ​​corresponding to each representative construction strategy.

[0077] After obtaining the probability distribution of the objective function values, the conditional risk values ​​corresponding to each representative construction strategy can be further determined, resulting in a robustness ranking. Conditional risk values ​​can be used to characterize the tail risk of strategy performance under adverse scenarios, while the robustness ranking can be used to characterize the stability of different construction strategies in uncertain environments. This process allows the assessment to focus not only on average performance but also on performance under extreme scenarios.

[0078] Subsequently, based on the simulation results and robustness ranking results corresponding to each representative construction strategy, several representative construction strategies were evaluated and ranked, resulting in a ranking. The evaluation comprehensively considers the pavement performance evolution trajectory obtained from the simulation, cumulative carbon emission results, cost cash flow results, and robustness information obtained based on uncertainty analysis. The ranking results can reflect the comprehensive priority of different strategies in terms of long-term performance, low-carbon goals, economic goals, and risk levels.

[0079] Finally, a recommended construction strategy sequence is output based on the ranking results. This sequence may include the optimal recommended strategy and several alternative strategies to meet the decision-making needs of different management entities with varying risk appetites. In some implementations, the TOPSIS multi-criteria decision-making method, weight allocation method, or other ranking methods may be used to complete the ranking, but this is not limited to these methods.

[0080] This embodiment introduces uncertainty analysis, conditional value of risk, and robustness ranking results based on simulation results, enabling the evaluation and ranking of representative construction strategies to take into account both long-term performance and stability under uncertain environments, thereby obtaining a more valuable sequence of recommended construction strategies.

[0081] In one embodiment, a method for generating and validating low-carbon construction strategies based on digital twins and multi-objective optimization is provided, including the following steps: Multi-objective optimization modeling steps: Construct a multi-objective optimization model with total carbon emissions over the entire life cycle, economic cost over the entire life cycle, and long-term pavement performance index as optimization objectives. Constraints include annual budget limits, minimum smoothness requirements, maximum construction period, and material supply restrictions. Pareto optimal solution set search steps: An improved NSGA-III algorithm is used to perform an efficient search in the policy solution space. The improvements include introducing knowledge from the construction engineering domain to initialize the population, designing crossover and mutation operators for construction strategies, and outputting a set of non-dominated Pareto optimal construction strategies. High-fidelity digital twin construction steps: Based on the BIM model of the target road, historical inspection data and real-time monitoring data, construct a dynamically updatable digital twin. This twin integrates a pavement performance degradation prediction model, a traffic load impact model and a climate and environmental impact model. Long-term simulation verification steps: Input representative strategies from the Pareto optimal strategy set into a digital twin and conduct dynamic simulations over a service life of up to 20 years to predict the evolution trajectory of road performance, cumulative carbon emissions, and cost cash flow under each strategy. Multi-criteria decision support steps: Based on the simulation results, the TOPSIS multi-criteria decision method is used to reorder the original Pareto solution set, taking into account the decision-maker's risk preferences (risk aversion, risk neutrality, risk preference), and outputting a risk-adjusted recommended strategy sequence and a long-term impact assessment report.

[0082] In the improved NSGA-III algorithm, population initialization adopts a case-based reasoning method, extracting strategy features from historical successful construction cases as initial solutions to accelerate the convergence process.

[0083] The objective function vector F(X) of the multi-objective optimization model is defined as: minF(X)=[fcost(X),fcarbon(X), fperformance(X)]T Where X is the vector of construction strategy decision variables; fcost(X) = The present value of the total life-cycle economic cost is calculated by discounting to the present, where Ct is the cost in year t and r is the discount rate. fcarbon(X) = ×EFm represents the total carbon emissions over the entire life cycle, Am,t represents the activity level of type m in year t, and EFm represents the corresponding emission factor. fperformance(X) = This represents the average value of the road surface condition index during the analysis period.

[0084] The pavement performance degradation prediction model adopts a hybrid model based on the fusion of physical mechanisms and data-driven approaches, and its core degradation equation can be expressed as: PCIt+1=PCIt α (ESALt)β eγ Aget+δ Iintervention+ t Where PCIt is the pavement condition index in year t; ESALt is the cumulative number of standard axle loads in year t; Aget is the pavement age; Iintervention is the construction intervention effect factor; α, β, γ, δ are model parameters; t represents the random error term. The model parameters are calibrated and dynamically adjusted using historical data and a Bayesian update method.

[0085] This embodiment also includes an uncertainty analysis step: Monte Carlo simulation is performed on key uncertainty parameters such as traffic volume growth, material prices, and climate anomalies to calculate the probability distribution of the objective function values ​​of each strategy, and the risk of the strategy is measured by conditional value of risk to output the robustness ranking of the strategies.

[0086] This embodiment proposes a three-layer framework of "optimization-simulation-decision-making," and its core innovation lies in: Domain-knowledge-enhanced multi-objective optimization: Integrating construction engineering expertise into optimization algorithms improves search efficiency and solution quality; High-fidelity dynamic digital twin: Constructing a road digital twin that integrates multiple physics fields and multiple scales, which can realistically reflect the complex interactions of "climate-transportation-materials-structure"; Decision support based on long-term risk perception: It not only provides optimization results, but also reveals the long-term risks and uncertainties of each strategy, supporting decision-making based on risk preferences.

[0087] The technical solution process is as follows: A three-objective optimization model is established: min (total carbon emissions), min (life cycle cost), and max (average performance index over 20 years); The Pareto front is searched using an improved NSGA-III algorithm; Fifty representative strategies that are evenly distributed on the frontier were selected; Parallel simulations are performed in the digital twin, with 100 simulations per policy (considering uncertainty). Based on the simulation results, the expected value, variance, and value at risk of each strategy are calculated. Based on the decision-maker's risk preferences, a personalized recommendation ranking is generated.

[0088] Example: Optimize target settings: total carbon emissions (tons of CO2), life cycle cost (RMB 10,000), and average International Roughness Index (IRI); Decision variables: construction timing, construction technology (thin overlay, micro-surfacing, in-situ thermal recycling, etc.), and material type for each road section in the next 5 years; Constraints: Total budget ≤ 50 million RMB / year, IRI ≤ 4.5 m / km, single construction period ≤ 30 days; Running the improved NSGA-III algorithm: population size 200, 500 generations, 128 Pareto optimal policies were obtained; Digital twin construction: integrating the road network BIM model, nearly 10 years of inspection data, traffic volume monitoring data, and meteorological data; Fifty representative strategies were selected for simulation: each strategy was simulated for 20 years, taking into account factors such as annual traffic volume growth of 5%-8%, material price fluctuation of ±15%, and increased probability of extreme weather events. Simulation results show that Strategy #45 (early preventive construction on key road sections + corrective construction on general road sections) performs well in terms of carbon emissions (-12%), cost (-8%), and long-term performance (+15%), and has the least fluctuation under different scenarios, so it is recommended as a robust strategy. The system generates detailed reports, including carbon emission cost curves, risk probability distributions, and sensitivity factor analyses for each strategy.

[0089] The improved NSGA-III algorithm uses a domain knowledge initialization method: 100 successful cases are selected from the historical database, features (road age, traffic volume, current status, climate zone) are extracted, and cluster analysis is used to generate 10 typical strategy templates as the initial population, which improves the algorithm's convergence speed by 40%.

[0090] Uncertainty modeling of the digital twin: A Bayesian update-based method is used to dynamically adjust the decay model parameters as new detection data is input, reducing the prediction error from 25% in the traditional method to 12%.

[0091] This embodiment shifts from "experience-based decision-making" to "data-driven + simulation verification" scientific decision-making; it clearly demonstrates the long-term risks and uncertainties of each strategy, avoiding the decision-making trap of "short-term gains and long-term losses"; it finds the true optimal balance point among hundreds of thousands of possible combinations, and is expected to reduce carbon emissions over the entire life cycle by 10%-20%, while saving costs by 5%-15%; it effectively balances the triple goals of "carbon reduction, cost reduction, and performance preservation", supporting the dual carbon goals of the highway industry.

[0092] In one embodiment, a low-carbon construction decision-making system based on digital twins and multi-objective optimization is provided, including a memory and a processor. The memory stores a computer program, which, when executed by the processor, causes the processor to perform the method described in the above embodiment.

[0093] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A low-carbon construction decision-making method based on digital twinning and multi-objective optimization, characterized in that, The method includes: Obtain basic data for the target road; A multi-objective optimization model for low-carbon construction is constructed based on the aforementioned basic data. Solving the multi-objective optimization model for low-carbon construction yields a set of Pareto-optimal construction strategies. Construct a digital twin of the target road; Select several representative construction strategies from the Pareto optimal construction strategy set; The representative construction strategies are input into the road digital twin and dynamic simulations are performed during the analysis period to obtain the simulation results corresponding to each representative construction strategy. Based on the simulation results corresponding to each of the representative construction strategies, the multiple representative construction strategies are evaluated and ranked to obtain and output a recommended construction strategy sequence.

2. The method of claim 1, wherein, The basic data for obtaining the target road includes: Acquire road structure data, pavement condition data, traffic load data, climate environment data, and construction history data of the target road; Based on the road segment division results of the target road, extract at least one of the following for each road segment: road age, existing road surface condition, traffic volume, climate zone, structural type, and existing construction records. The data corresponding to each road segment are correlated to form the basic dataset used to construct the low-carbon construction multi-objective optimization model.

3. The method of claim 1, wherein, The construction of a low-carbon construction multi-objective optimization model based on the aforementioned basic data includes: Use at least one of the following as a construction strategy decision variable: construction timing, construction technology, and material type. The optimization targets are: carbon emissions throughout the entire life cycle, economic costs throughout the entire life cycle, and long-term pavement performance during the analysis period. Introduce at least one of the following constraints: budget constraints, pavement performance constraints, construction period constraints, and material supply constraints; The low-carbon construction multi-objective optimization model is constructed based on the construction strategy decision variables, the optimization objective, and the constraints.

4. The method of claim 3, wherein, Solving the multi-objective optimization model for low-carbon construction yields a set of Pareto-optimal construction strategies, including: The construction strategy decision variables are coded according to the strategy. Based on the aforementioned strategy encoding method and historical construction cases, the initial population is initialized to obtain the initialized initial population. Based on the initial population after initialization, the improved NSGA-III algorithm is used to iteratively solve the low-carbon construction multi-objective optimization model to obtain the Pareto optimal construction strategy set.

5. The method of claim 4, wherein, The initial population is initialized based on the strategy encoding method and historical construction cases to obtain the initialized initial population, including: Successful construction cases were selected from the historical construction case database; Extract the road features and construction strategy features corresponding to the successful construction cases; Cluster analysis is performed on the road features and construction strategy features corresponding to the successful construction cases to generate typical strategy templates; The initial population is initialized based on the typical strategy template and the strategy encoding method to obtain the initialized initial population; Based on the initialized initial population, the improved NSGA-III algorithm is used to iteratively solve the low-carbon construction multi-objective optimization model to obtain the Pareto optimal construction strategy set, including: Based on the initial population after initialization, crossover and mutation operations are performed, and the candidate construction strategies formed after the crossover and mutation operations are constrained, verified or repaired to update the population and continue to iteratively solve the low-carbon construction multi-objective optimization model to obtain the Pareto optimal construction strategy set.

6. The method of claim 1, wherein, The construction of the digital twin of the target road includes: A dynamically updatable digital twin of the target road is constructed based on its BIM model, historical inspection data, and real-time monitoring data. The road digital twin integrates a pavement performance degradation prediction model, a traffic load impact model, and a climate and environmental impact model.

7. The method of claim 1, wherein, Several representative construction strategies are selected from the Pareto optimal construction strategy set, including: Based on the distribution of objective function values, Pareto front distribution, or reference point coverage of each construction strategy in the Pareto optimal construction strategy set, select several representative construction strategies from the Pareto optimal construction strategy set. The process involves inputting the multiple representative construction strategies into the road digital twin and performing dynamic simulations during the analysis period to obtain simulation results corresponding to each representative construction strategy, including: The multiple representative construction strategies are loaded into the road digital twin according to the preset analysis period; Based on the pavement performance degradation prediction model, the traffic load impact model, and the climate environment effect model, dynamic simulations are performed on the multiple representative construction strategies respectively. Output the pavement performance evolution trajectory, cumulative carbon emission results, and cost cash flow results corresponding to each representative construction strategy.

8. The method of claim 6, wherein, The road digital twin integrates a pavement performance degradation prediction model, a traffic load impact model, and a climate and environmental impact model, including: The pavement performance degradation prediction model is constructed based on historical test data, and the correlation between pavement performance status and standard axle load, pavement age, and construction intervention is established. The traffic load impact model is constructed based on traffic load data, and the change in traffic load is used as an external input to the road surface performance evolution process. The climate and environmental action model is constructed based on climate and environmental data, and climate and environmental changes are used as external inputs into the pavement performance evolution process. After acquiring new detection data, the parameters of the pavement performance degradation prediction model are calibrated and dynamically corrected based on historical detection data and the new detection data.

9. The method of claim 1, wherein, Based on the simulation results corresponding to each of the representative construction strategies, the multiple representative construction strategies are evaluated and ranked to obtain and output a recommended construction strategy sequence, including: For at least one uncertainty factor among traffic volume growth, material price fluctuations, and climate anomalies, uncertainty analysis is performed on the multiple representative construction strategies to obtain the probability distribution of the objective function value corresponding to each representative construction strategy. Based on the probability distribution of the objective function value, the conditional risk value corresponding to each of the representative construction strategies is determined, and a robust ranking result is formed; Based on the simulation results and robust ranking results corresponding to each of the representative construction strategies, the multiple representative construction strategies are evaluated and ranked to obtain the ranking results; The recommended construction strategy sequence is output based on the sorting results.

10. A low-carbon construction decision-making system based on digital twins and multi-objective optimization, characterized in that, The system includes a memory and a processor, wherein the memory stores a computer program that, when executed by the processor, causes the processor to perform the method as described in any one of claims 1 to 9.

Images