Steel structure engineering intelligent energy consumption analysis method for low carbon target

Abstract

The present application relates to energy consumption management and control analysis technical field, especially to a kind of steel structure engineering intelligent energy consumption analysis method for low carbon target.The method includes the following steps: collecting the multidimensional data corresponding to the whole life cycle of steel structure engineering is decoupled for low carbon feature, generate engineering space-time feature sequence under the constraint of low carbon target, and carry out low carbon process mode mining and space-time energy consumption field construction, generate space-time coupling energy consumption field distribution atlas;Based on the space-time coupling energy consumption field distribution atlas, energy consumption hotspot and carbon flow path tracing analysis are carried out, and low carbon process matching is carried out, while dynamic carbon resistance analysis and carbon flow bottleneck identification are carried out and are fused to generate low carbon engineering optimization configuration parameters for the whole life cycle;Low carbon construction deduction model is driven based on the low carbon engineering optimization configuration parameters, and the low carbon engineering optimization configuration parameters are dynamically adjusted, to realize closed-loop intelligent energy consumption management and control for low carbon target.The present application can greatly improve the landing effect of engineering low carbon optimization.

Description

Technical Field

[0001] This invention relates to the field of energy consumption management and analysis technology, and in particular to an intelligent energy consumption analysis method for steel structure engineering aimed at achieving low-carbon goals. Background Technology

[0002] As my country's "dual-carbon" strategy continues to deepen, the construction industry, as a core sector of the national economy characterized by high energy consumption and high carbon emissions, has become a key industry for energy conservation, emission reduction, and green and low-carbon transformation. Steel structure engineering, with its advantages of light weight, high strength, high construction efficiency, and recyclability, is widely used in various infrastructure and construction scenarios such as super high-rise buildings, large stadiums, industrial plants, and bridge projects, and is the mainstream structural form in modern construction engineering. However, steel structure engineering encompasses the entire life cycle, including component processing, factory prefabrication, on-site hoisting, welding and painting, and operation and maintenance dismantling. It is characterized by complex energy sources, dispersed carbon emission nodes, strong spatiotemporal energy coupling, and concealed carbon flow transmission paths, resulting in persistently high overall carbon emissions. Precise control of its energy consumption and carbon emissions has become a key bottleneck restricting the low-carbon development of the construction industry.

[0003] Currently, mainstream energy consumption analysis methods in the industry are mostly static and single-dimensional energy consumption statistics, simply calculating the water, electricity, and mechanical energy consumption of a single stage during construction. This ignores the interconnected energy consumption impacts throughout the entire lifecycle of steel structure design, component production, transportation, operation and maintenance, dismantling, and recycling. The data collection dimensions are limited, making it impossible to achieve integrated analysis of multi-source heterogeneous data, resulting in biased and inaccurate energy consumption analysis results. Furthermore, existing technologies lack the ability to decouple the low-carbon characteristics of steel structure engineering, failing to construct spatiotemporal characteristic sequences that fit the actual project, making it difficult to characterize the differences in energy consumption across different construction periods and spatial areas. This hinders the spatiotemporal linkage analysis of energy consumption and carbon emissions, thus reducing the efficiency of energy consumption management in steel structure engineering. Summary of the Invention

[0004] Therefore, it is necessary for the present invention to provide an intelligent energy consumption analysis method for steel structure engineering with low carbon objectives, in order to solve at least one of the above-mentioned technical problems.

[0005] To achieve the above objectives, a smart energy consumption analysis method for steel structure engineering oriented towards low-carbon goals includes the following steps: Step S1: Collect multi-dimensional data corresponding to the entire life cycle of steel structure engineering, decouple the multi-dimensional data for low-carbon features, and generate a spatiotemporal feature sequence of engineering under the constraint of low-carbon goals; obtain historical low-carbon benchmark engineering data and mine low-carbon process modes to generate a feature library of low-carbon process modes for steel structures. Step S2: Construct a spatiotemporal energy consumption field for the spatiotemporal feature sequence of the project to generate a spatiotemporal coupled energy consumption field distribution map; perform energy consumption hotspot and carbon flow path tracing analysis based on the spatiotemporal coupled energy consumption field distribution map to generate a set of high energy consumption spatiotemporal units and a network of key carbon flow paths; Step S3: Match the high-energy-consuming spatiotemporal unit set with the low-carbon process mode feature library of steel structure to generate a cluster of low-carbon process optimization schemes; perform dynamic carbon resistance analysis and carbon flow bottleneck identification on the key carbon flow path network to generate a set of carbon flow optimization and control strategies. Step S4: Integrate the process low-carbon optimization scheme cluster with the carbon flow optimization and control strategy set to generate low-carbon engineering optimization configuration parameters for the entire life cycle; drive the low-carbon construction simulation model based on the low-carbon engineering optimization configuration parameters to output the low-carbon target attainability assessment results and dynamic optimization control instructions; dynamically adjust the low-carbon engineering optimization configuration parameters based on the dynamic optimization control instructions to achieve closed-loop intelligent energy consumption management for low-carbon targets.

[0006] The beneficial effects of this invention are: The intelligent energy consumption analysis method for steel structure engineering oriented towards low-carbon goals proposed in this invention has the following advantages compared with existing technologies: It collects multi-dimensional data throughout the entire lifecycle of steel structure engineering, generates a spatiotemporal feature sequence of engineering under the constraint of low-carbon goals through low-carbon feature decoupling, and simultaneously mines low-carbon process modes based on historical low-carbon benchmark engineering data to construct a feature library of low-carbon process modes for steel structures. This step achieves full coverage of energy consumption, materials, processes, and environmental data at each stage through multi-dimensional data collection throughout the entire lifecycle. By decoupling low-carbon features, it identifies core features related to low-carbon management, forming a standardized spatiotemporal feature sequence of engineering, fully reflecting the low-carbon evolution law of the entire engineering lifecycle. By mining historical benchmark engineering process modes to build a feature library, it accumulates mature low-carbon construction logic and technical paradigms, providing a sufficient data foundation and process support for subsequent energy consumption field modeling, process optimization, and carbon flow control, thus solving the industry shortcomings of traditional methods such as single data dimensions, insufficient feature mining, and the inability to reuse low-carbon experience. Secondly, based on the spatiotemporal characteristic sequence of the project, a spatiotemporal energy consumption field is constructed, generating a spatiotemporally coupled energy consumption field distribution map. Based on this map, energy consumption hotspots and carbon flow path tracing analysis are conducted, extracting high-energy-consuming spatiotemporal unit sets and key carbon flow path networks. This step builds a spatiotemporally coupled energy consumption field model based on the full-cycle spatiotemporal characteristic sequence, combining temporal evolution patterns with spatial distribution characteristics to fully present the dynamic distribution of energy consumption across the entire steel structure project. By tracing energy consumption hotspots, spatiotemporal units with continuously high energy consumption and high emissions are identified. Through carbon flow path analysis, the transmission links and core nodes of carbon emissions are identified, forming a complete carbon flow path network. This clearly exposes redundant energy consumption points and carbon emission transmission shortcomings throughout the entire life cycle of the project, providing a clear and targeted basis for subsequent process optimization and carbon flow control. This completely solves the shortcomings of traditional energy consumption analysis, such as being one-sided, lacking spatiotemporal linkage, and being unable to locate problem points. Then, by matching the high-energy-consuming spatiotemporal unit set with the low-carbon process mode feature library, a cluster of low-carbon optimization schemes for the process is generated. Simultaneously, carbon resistance analysis and bottleneck identification are performed on the key carbon flow path network to generate a set of carbon flow optimization and control strategies. This step, for the identified high-energy-consuming spatiotemporal units, combines a benchmark process feature library for multi-dimensional process adaptation screening, matching low-carbon optimization schemes suitable for different construction scenarios and energy consumption problems, forming a cluster of schemes covering multiple scenarios, achieving precise adaptation between problem scenarios and optimized processes. Carbon resistance analysis quantifies the factors hindering carbon flow transmission, identifies the core bottlenecks of carbon flow accumulation and emission exceeding standards, and generates a targeted set of carbon flow control strategies. This addresses the shortcomings of low-carbon management from two dimensions: process modification and carbon flow diversion, solving the problems of traditional optimization methods being singular, lacking adaptability, and unable to eradicate carbon emission bottlenecks, significantly improving the implementation effect of low-carbon optimization in engineering.Finally, by integrating process optimization schemes and carbon flow control strategies, optimized configuration parameters for low-carbon engineering throughout its entire lifecycle are generated. Based on a low-carbon construction simulation model, achievability assessment results and dynamic control commands are output. Closed-loop intelligent energy consumption management is achieved through dynamic parameter adjustment. This step integrates the dual optimization logic of process optimization and carbon flow control, generating standardized low-carbon configuration parameters covering the entire lifecycle of the project, enabling collaborative governance of process modification and carbon flow mitigation. The low-carbon construction simulation model simulates the implementation effect of optimization schemes, outputting target achievability assessment results and proactively mitigating the risk of optimization failure. Based on dynamic control commands, low-carbon configuration parameters are adjusted in real time, and optimization strategies are continuously iterated according to changes in on-site conditions. This constructs a complete closed-loop management system encompassing parameter configuration, strategy execution, effect evaluation, and dynamic updates, solving the problems of static, rigid, poorly adaptable, and lacking iterative optimization capabilities in traditional low-carbon management. This comprehensively improves the intelligent and refined management level of energy consumption and carbon emissions throughout the entire lifecycle of steel structure engineering. Attached Figure Description

[0007] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a schematic diagram of the steps in the intelligent energy consumption analysis method for steel structure engineering aimed at low carbon goals of the present invention. Figure 2 for Figure 1 A detailed flowchart of step S1; Figure 3 for Figure 1 A detailed flowchart of step S2. Detailed Implementation

[0008] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0009] To achieve the above objectives, please refer to Figure 1-2 This invention provides a smart energy consumption analysis method for steel structure engineering with low-carbon goals. In this example, the smart energy consumption analysis method for steel structure engineering with low-carbon goals includes the following steps: Step S1: Collect multi-dimensional data corresponding to the entire life cycle of steel structure engineering, decouple the multi-dimensional data for low-carbon features, and generate a spatiotemporal feature sequence of engineering under the constraint of low-carbon goals; obtain historical low-carbon benchmark engineering data and mine low-carbon process modes to generate a feature library of low-carbon process modes for steel structures. In this embodiment of the invention, multidimensional heterogeneous data covering the entire lifecycle of steel structure engineering—including design, materials, manufacturing, transportation, construction, operation and maintenance, and demolition—is collected. The data categories include component design parameters, bill of materials parameters, processing energy consumption time-series logs, three-dimensional spatiotemporal transportation trajectories, construction equipment operating parameters, and construction site environmental streaming monitoring data, forming a raw multidimensional dataset of the entire engineering process. A four-dimensional spatiotemporal grid matching model is employed. Spatiotemporal alignment and semantic fusion of multi-source heterogeneous data were completed to construct a unified engineering holographic data cube. Based on the low-carbon evaluation index system for steel structures, a linear dimensional mapping model was used. Accurate extraction of carbon source and carbon sink potential characteristics across all stages, combined with a coupled correlation analysis model. Carbon feature decoupling is achieved, separating independent carbon source factors, coupled carbon flow factors, and potential carbon sink factors. This is accomplished through a model reconstructed from spatiotemporal feature sequences. Generate standardized spatiotemporal feature sequences for engineering projects under low-carbon target constraints. Retrieve full data of low-carbon benchmark projects from the historical steel structure engineering database, employ density peak clustering algorithm to mine low-carbon process modes across multiple scenarios, and combine this with high-dimensional feature encoding to standardize and aggregate process features. Construct a steel structure low-carbon process mode feature library with unified dimensions and full scenario coverage, providing a standard feature base for subsequent process matching and energy consumption optimization.

[0010] Step S2: Construct a spatiotemporal energy consumption field for the spatiotemporal feature sequence of the project to generate a spatiotemporal coupled energy consumption field distribution map; perform energy consumption hotspot and carbon flow path tracing analysis based on the spatiotemporal coupled energy consumption field distribution map to generate a set of high energy consumption spatiotemporal units and a network of key carbon flow paths; In this embodiment of the invention, a four-dimensional spatiotemporal voxel grid is uniformly partitioned onto the spatiotemporal feature sequence of the engineering project generated in step S1, and a spatiotemporal grid discretization model is used. Continuous spatiotemporal data is discretized into standardized three-dimensional spatiotemporal units. This is based on a linear weighted aggregation model. Net carbon emission equivalents are calculated unit by unit, and a spatiotemporal energy transfer weight matrix is ​​constructed using a spatiotemporal coupling model to quantify the three-dimensional spatial diffusion and temporal cumulative effects of carbon emissions. A global four-dimensional spatiotemporal interpolation field strength model is then used. Generate a continuous, fault-free spatiotemporal coupled energy consumption field distribution map. Based on a four-dimensional spatiotemporal gradient model. Field strength gradient analysis and contour line threshold extraction were conducted, and a spatiotemporal curvature feature discrimination mechanism was used to filter out local fluctuation pseudo-convergence regions, identify stable carbon emission core convergence areas and spatiotemporal topological critical edges, and aggregate them into a standardized set of high-energy-consuming spatiotemporal units. Based on the spatiotemporal propagation mechanism of carbon particles and streamline tracing model, path tracing, convergence point analysis, and topological reconstruction of the global carbon flow vector field were completed. Redundant secondary paths were eliminated, and core channels with dominant transport functions were retained, ultimately generating a key carbon flow path network that accurately represents the spatiotemporal carbon emission migration law of engineering projects.

[0011] Step S3: Match the high-energy-consuming spatiotemporal unit set with the low-carbon process mode feature library of steel structure to generate a cluster of low-carbon process optimization schemes; perform dynamic carbon resistance analysis and carbon flow bottleneck identification on the key carbon flow path network to generate a set of carbon flow optimization and control strategies. In this embodiment of the invention, by uniformly extracting four core attribute features—engineering stage, process type, equipment type, and material type—from the high-energy-consuming spatiotemporal unit set output in step S2, a structured high-energy-consuming unit feature vector is constructed. Based on feature similarity matching model In the steel structure low-carbon process mode feature library, rigid matching of stages and process dimensions is completed, and a complete set of low-carbon technology packages for the adapted processes is analyzed to generate a multi-dimensional candidate low-carbon process mode set. Through a multi-objective collaborative optimization analysis model, the unit carbon reduction potential and carbon flow improvement benefits of each process mode are calculated, and the process combination and implementation strategy with the best comprehensive benefits are selected and integrated to form a complete cluster of low-carbon process optimization schemes. For key carbon flow path networks, global structural vulnerability analysis and dynamic carbon resistance calculation are carried out. The comprehensive carbon resistance value of each node and edge is solved by fusing the basic carbon resistance model and the dynamic impulse perturbation model. Based on attribute condition statistical benchmarks and adaptive noise correction mechanisms, the abnormal deviation index is purified to accurately identify high-blockage and high-dependency carbon flow bottleneck units. For various bottlenecks, reverse matching process iteration, material replacement, and energy scheduling optimization methods are used. Combined with network cascade effects, secondary bottleneck schemes are verified and eliminated, ultimately forming a set of carbon flow optimization and control strategies that take into account both local resistance reduction and global steady state.

[0012] Step S4: Integrate the process low-carbon optimization scheme cluster with the carbon flow optimization and control strategy set to generate low-carbon engineering optimization configuration parameters for the entire life cycle; drive the low-carbon construction simulation model based on the low-carbon engineering optimization configuration parameters to output the low-carbon target attainability assessment results and dynamic optimization control instructions; dynamically adjust the low-carbon engineering optimization configuration parameters based on the dynamic optimization control instructions to achieve closed-loop intelligent energy consumption management for low-carbon targets.

[0013] In this embodiment of the invention, a process-strategy coupling impact model is constructed. This study quantifies the synergistic and antagonistic relationships between low-carbon optimization schemes and carbon flow optimization control strategies, generating a standardized scheme-strategy interaction matrix. With minimizing total life-cycle carbon emissions as the core objective, and relying on the objective function... By coupling three hard constraints—cost, schedule, and technology—a multi-objective optimization configuration model for low-carbon engineering is constructed. All process schemes and control strategies are treated as multi-dimensional decision variables, and a constrained multi-objective intelligent iterative algorithm is employed. The global Pareto optimal solution set is obtained by solving the problem. A globally optimal configuration scheme is selected through a two-layer assessment model of feasibility and technology maturity, and the optimized configuration parameters for low-carbon engineering across all dimensions are decoded. The low-carbon construction simulation model is initialized with the optimal parameters, and a dynamic prediction model is then used. The simulation method tracks the evolution of subsequent construction, resource allocation, and carbon emissions, compares the quantitative differences with preset low-carbon targets, and generates an assessment of the attainability of low-carbon targets. This is based on a closed-loop correction model. Dynamically update and optimize configuration parameters to achieve closed-loop intelligent management and control of low-carbon energy consumption in steel structure engineering, including dynamic perception, evaluation, and regulation.

[0014] Furthermore, step S1 includes the following steps: Step S11: Collect raw heterogeneous data from the design, material, manufacturing, transportation, construction, operation and maintenance and dismantling stages of steel structure engineering. The raw heterogeneous data shall include at least design drawing parameters, material list, processing energy consumption log, transportation trajectory, construction machinery operation data and environmental monitoring data. In this embodiment of the invention, addressing the need for low-carbon energy consumption analysis throughout the entire lifecycle of steel structure engineering, the invention completes the fixed-point and continuous collection of original heterogeneous data across six core stages, strictly matching the data characteristics of the entire business scenario of steel structure engineering design, materials, manufacturing, transportation, construction, operation and maintenance, and demolition. During the design stage, five types of core structured and unstructured data are collected: parameters of steel structure BIM detailed drawings, component cross-sectional dimensions, node connection process parameters, design load parameters, and structural layout parameters, covering all core design indicators of the main steel structure, including steel columns, steel beams, and supporting components. During the materials stage, parameters of steel material composition, profile specifications, auxiliary material usage lists, material procurement batch parameters, and basic carbon emission parameters of raw materials are collected to form a complete inventory of all engineering materials. During the manufacturing stage, energy consumption time-series logs of all processes involved in cutting, welding, grinding, straightening, and painting of steel structure components are collected, along with synchronous data on workshop equipment start-up and shutdown status, process processing time, and equipment load operation. During the transportation stage, GPS spatiotemporal trajectory data of steel structure components from the factory to the construction site, mileage data of transport vehicles, ambient temperature data during transportation, and fixed parameters of vehicle load are collected. During the construction phase, real-time operating speed, operation duration, start-stop frequency, and load power data of hoisting machinery, welding equipment, and compaction equipment are collected. Simultaneously, real-time flow environmental monitoring data on wind speed, air humidity, ambient temperature, and dust concentration at the construction site are also collected. During the operation and maintenance and demolition phases, data on the energy consumption of daily maintenance equipment, structural loss monitoring, demolition equipment operation, and demolition sequence data are collected. All collected data retains its original format, original time sequence identifier, and stage-specific spatiotemporal labels without preprocessing, forming a multi-type, multi-dimensional, full-lifecycle original heterogeneous dataset for steel structure engineering. This provides complete original data support for subsequent data fusion and feature extraction.

[0015] Step S12: Perform spatiotemporal alignment and semantic fusion on the original heterogeneous data to generate a spatiotemporally aligned engineering holographic data cube; perform feature dimension mapping on the engineering holographic data cube based on the low-carbon evaluation index system, and extract the carbon source features and carbon sink potential features corresponding to each stage; In this embodiment of the invention, a spatiotemporal joint alignment algorithm and a semantic fusion model are used to regularize and integrate the collected multi-type, temporally heterogeneous raw engineering data, constructing a standardized engineering holographic data cube. The spatiotemporal alignment employs a spatiotemporal grid matching model, with the following model structure: ,in For spatiotemporal grid datasets, , , These represent the three-dimensional spatial coordinate dimensions of the project, corresponding to the horizontal, vertical, and longitudinal construction spaces, respectively. In terms of time dimension, For spatiotemporal matching impulse functions, This model maps multi-source data with different acquisition frequencies, spatial locations, and time series nodes to a unified three-dimensional spatiotemporal grid system for engineering projects, eliminating spatiotemporal misalignment issues. Semantic fusion employs a multi-level semantic association mapping model. The model input is spatiotemporally aligned multi-source data, the intermediate layer is an engineering business semantic tagging system, and the output is holographic data with unified semantic encoding. Through four core semantic rules—component technology, construction procedures, energy consumption type, and environmental impact—the business logic of heterogeneous data across different dimensions is associated, generating a spatiotemporally aligned and semantically unified holographic data cube for steel structure engineering. Based on a pre-set low-carbon evaluation index system for steel structure engineering, which includes four primary indicators (energy consumption, carbon emissions, carbon accumulation, and environmental degradation) and 28 secondary sub-indicators, the feature dimensions of the holographic data cube are accurately mapped. A linear dimension mapping model is used. ,in For the first The characteristic values ​​corresponding to each low-carbon indicator For the first The first item under the indicator Fixed weighting coefficients for each data dimension The model uses the holographic data cube to accurately extract the carbon emission source characteristics, energy consumption characteristics, material loss carbon source characteristics, and carbon sink potential characteristics corresponding to green plant carbon sequestration, material recycling, and energy-saving process applications at each stage of design, materials, manufacturing, transportation, construction, operation and maintenance, and demolition. This enables the accurate extraction and dimensional classification of carbon characteristics at all stages.

[0016] Step S13: Construct a graph of the relationship between carbon transfer between stages based on the carbon source characteristics and carbon sink potential characteristics, and analyze the topological structure of carbon element flow and transformation between stages; perform coupling correlation analysis on the carbon source characteristics and carbon sink potential characteristics based on the topological structure, and decouple independent carbon source factors, coupled carbon flow factors and potential carbon sink factors. In this embodiment of the invention, a topological diagram of carbon transfer influence relationships throughout the entire life cycle of a steel structure project is constructed based on the extracted carbon source and carbon sink potential characteristics at each stage. Six major engineering stages are identified as topological nodes, and material flow, energy consumption, process connection, and waste transfer between stages are identified as topological edges. The carbon transfer weights of each node and edge are quantified, and the flow path, transformation forms, and transmission topological structure of carbon elements from design and planning, material input, processing and manufacturing, on-site construction, post-construction operation and maintenance to demolition and recycling are fully analyzed. Based on the constructed carbon transfer topological structure, a coupled correlation analysis model is used to decouple and decompose carbon characteristics. The structure of the coupled correlation analysis model used is as follows: ,in For the first Carbon-like characteristics and the first The coupling coefficient of carbon-like characteristics, Let covariance be the variance of the two types of features. , The variance represents the variance of the two types of features. The model training data consists of a time-series dataset of carbon source and carbon sink features across the entire project phase. The algorithm logic is as follows: First, the correlation and coupling degree of any two types of carbon features is calculated, with a coupling degree threshold of 0.3. When the coupling degree coefficient is below the threshold, it is determined that the features have no correlation effect and are decoupled into independent carbon source factors, including carbon emission factors of stationary materials, energy consumption factors of basic equipment, and inherent process loss factors. When the coupling degree coefficient is above the threshold, it is determined that multiple types of features have mutual correlation effects and are integrated into coupled carbon flow factors, including carbon flow factors of material transfer between phases, energy consumption carbon flow factors of process connection, and carbon flow factors of environmental and construction coupling. At the same time, the algorithm mines the implicit features corresponding to underutilized carbon sequestration, material recycling, and energy-saving renovation from the topology structure, classifies and decouples them into potential carbon sink factors, including carbon sink factors of scrap steel recycling, carbon sink factors of operation and maintenance energy saving, and carbon sink factors of construction environment optimization. Finally, it completes the accurate decoupling and classification of the three core low-carbon impact factors.

[0017] Step S14: Integrate the independent carbon source factor, coupled carbon flow factor and potential carbon sink factor, and combine them with the spatiotemporal labels of each stage to generate the engineering spatiotemporal feature sequence under the low-carbon target constraint; In this embodiment of the invention, the independent carbon source factor, coupled carbon flow factor, and potential carbon sink factor obtained through decoupling are integrated and bound to three types of spatiotemporal labels: spatial location coordinates, construction sequence nodes, and process stage attributes, for each stage of the steel structure project's entire lifecycle. This constructs a spatiotemporal feature sequence for the project under low-carbon objective constraints. A spatiotemporal feature sequence reconstruction model is used to generate the sequence; the model structure is as follows: ,in for Time-space coordinates The spatiotemporal characteristic sequence values ​​of the project at that location, For an independent carbon source factor set, For the coupled carbon flow factor set, For a set of potential carbon sink factors, , , The fixed weighting coefficients are determined by the low-carbon emission standards and energy consumption control standards of the steel structure engineering industry. The model training data consists of a spatiotemporal label dataset for the entire engineering phase and a dataset of three types of decoupled factors. The algorithm logic involves superimposing the feature contribution values ​​of various low-carbon factors at each time step and each three-dimensional spatial unit according to the engineering construction sequence and the three-dimensional spatial distribution range. It unifies the sequence sampling interval and the scale of the three-dimensional spatial unit, eliminates invalid feature units with temporal disorder and spatial mismatch, and retains the core spatiotemporal feature information that fits the low-carbon control target. This forms a continuous, regular, dimensionally unified, and four-dimensional spatiotemporal feature sequence of low-carbon engineering in steel structure engineering, accurately mapping the three-dimensional spatial distribution and temporal evolution of energy consumption and carbon emissions throughout the entire life cycle of the project.

[0018] Step S15: Retrieve historical low-carbon benchmark project data with excellent low-carbon performance from the historical engineering database and mine low-carbon process patterns. Construct the feature library of the steel structure low-carbon process patterns through pattern clustering and feature encoding.

[0019] In this embodiment of the invention, a pre-set historical engineering database of steel structure projects is retrieved. This database stores raw lifecycle data, energy consumption data, carbon emission accounting data, and process parameter data for various industrial and civil steel structure projects completed in the past five years. From this database, a dataset of historical low-carbon benchmark projects with unit carbon emissions lower than the industry average by 20% and comprehensive energy consumption lower than the industry average by 15% is selected. A density peak clustering algorithm is used to mine low-carbon process patterns. The core model structure of the algorithm is as follows: , ,in For the first Local density of a process sample For the sample With sample The feature distance, To cut off the distance, For sample density distance, The algorithm uses a judgment function. The training data consists of comprehensive feature data on construction technology, processing flow, energy consumption control, and material utilization of benchmark projects. The algorithm logic calculates the local density and density distance of each process feature sample, automatically clustering it into four core process mode clusters: low-carbon processing technology, low-carbon transportation technology, low-carbon construction technology, and low-carbon operation and maintenance technology. High-dimensional feature encoding is then applied to each clustered low-carbon process mode. A coding model combining one-hot encoding and feature dimensionality reduction is used to transform process flow, energy consumption parameters, carbon emission parameters, and resource utilization parameters into standardized high-dimensional feature vectors. All clustered low-carbon process mode features are integrated, and low-carbon feature data from different scenarios, stages, and process types are categorized and collected. This constructs a low-carbon process mode feature library covering the entire lifecycle of steel structure engineering, adapting to multiple scenarios, and with standardized feature dimensions. This provides standard feature support for subsequent intelligent energy consumption analysis, low-carbon process selection, and carbon emission optimization in engineering projects.

[0020] Furthermore, step S2 includes the following steps: Step S21: Perform spatiotemporal grid subdivision on the spatiotemporal feature sequence of the project to discretize the project time progress and three-dimensional spatial location into spatiotemporal voxel units; In this embodiment of the invention, by using the low-carbon spatiotemporal characteristic sequence of steel structure engineering as the basic data source, a full-dimensional regularized spatiotemporal grid partitioning is implemented to achieve discretized voxel unit division of the project's time progress and three-dimensional spatial location. The spatial dimension relies on the three-dimensional coordinate system of the steel structure engineering construction site, including the horizontal X-axis, vertical Y-axis, and vertical Z-axis, defining a fixed and uniform three-dimensional spatial grid size, and uniformly partitioning the three-dimensional spatial area covered by the project construction to form standardized three-dimensional spatial grid units. The time dimension matches the construction sequence of the entire life cycle of the steel structure project, dividing the complete project period according to fixed time intervals to generate continuous and uninterrupted time slice units. The spatiotemporal voxel partitioning adopts a three-dimensional spatiotemporal grid discretization model, the model structure of which is... ,in It is a single standard four-dimensional spatiotemporal voxel unit. , , They are respectively three-dimensional spatial intervals. The model's training data consists of a full-domain 3D spatial coordinate dataset and a full-cycle time-series node dataset. The algorithm's logic involves orthogonally coupling the 3D spatial grid with time slices, decomposing the continuous spatiotemporal feature sequence of the project into non-overlapping, dimensionally unified, and temporally continuous four-dimensional spatiotemporal voxel units. Each voxel unit is bound to a specific 3D spatial coordinate range and time interval attribute, fully carrying the low-carbon spatiotemporal characteristics of the steel structure project for the corresponding time period and corresponding 3D spatial region. This achieves refined 3D discrete partitioning of the entire spatiotemporal domain of the project, adapting to the 3D engineering characteristics of layered construction of steel structures and vertical component operations, laying a refined spatiotemporal foundation for subsequent unit carbon emission accounting.

[0021] Step S22: Within each spatiotemporal voxel unit, aggregate the intensity of carbon source factor and carbon sink factor in the spatiotemporal characteristic sequence of its contained engineering elements, and calculate the unit's net carbon emission equivalent; In this embodiment of the invention, the intensity aggregation calculation of carbon source factors and carbon sink factors within a unit is completed by using a standardized four-dimensional spatiotemporal voxel unit as the smallest calculation unit, accurately solving the net carbon emission equivalent of each unit. Based on the three-dimensional spatial and temporal feature sequence data bound to the voxel unit, the intensity feature values ​​of independent carbon source factors, coupled carbon flow factors, and potential carbon sink factors within the four-dimensional spatiotemporal range of the unit are screened to distinguish between positive carbon emission factors and negative carbon sequestration factors. The unit intensity aggregation adopts a linear weighted aggregation model, the model structure of which is as follows: ,in Net carbon emissions equivalent per voxel unit. For the intensity of independent carbon source factors, For coupling carbon flow factor intensity, As the potential carbon sink factor intensity, , , The model uses fixed weighting coefficients. The training data consists of a dataset of full carbon source and carbon sink characteristic intensities embedded within each four-dimensional spatiotemporal voxel unit. The algorithm logic is to positively accumulate the emission contribution intensity of all carbon source factors and negatively subtract the carbon sequestration contribution intensity of all carbon sink factors, thus thoroughly completing the fusion calculation of multiple types of low-carbon factors within the unit. This yields a unique net carbon emission equivalent value for each three-dimensional spatiotemporal voxel unit, quantifying the carbon emission and carbon sequestration difference characteristics of steel structure engineering in each refined four-dimensional spatiotemporal unit, and accurately covering the differences in energy consumption and carbon emissions of steel structures at different construction floor heights and in different vertical working areas.

[0022] Step S23: Based on the net carbon emission equivalent of all spatiotemporal voxel units and their spatial adjacency and temporal order, construct a spatiotemporal energy transfer weight matrix to characterize the spatiotemporal diffusion and accumulation effects of carbon emissions. In this embodiment of the invention, by using net carbon emission equivalent data based on global four-dimensional spatiotemporal voxel units, combined with the three-dimensional spatial adjacency relationships and temporal progression relationships of the units, a spatiotemporal energy transfer weight matrix adapted to steel structure engineering is constructed to quantify the spatiotemporal diffusion and accumulation patterns of carbon emissions. At the spatial level, three-dimensional six-neighbor spatial adjacency rules are defined; at the temporal level, sequential temporal progression association rules are established; and a weight matrix is ​​constructed based on a spatiotemporal correlation coupling model. The model structure is as follows: ,in For unit With unit Spatiotemporal transfer weights, This refers to the adjacency weight coefficient in three-dimensional space. For time-series progressive weighting coefficients, The model training data consists of a 3D spatial topological relationship dataset, a temporal ranking dataset, and a net carbon emission equivalent dataset for all voxel units. The algorithm logic is that the closer the adjacent units are in the 3D spatial space, the larger the spatial weight coefficient; the shorter the temporal interval between adjacent units, the larger the temporal weight coefficient. Through weighted coupling, the transfer weight between units in the entire domain is obtained, forming a dimensionally regular and precisely correlated spatiotemporal energy transfer weight matrix. This matrix fully depicts the dynamic effects of carbon emission diffusion and accumulation across three-dimensional space and across time in steel structure engineering, and is adapted to the carbon emission transfer characteristics of vertical layer-by-layer construction and three-dimensional cross-operation of steel structures.

[0023] Step S24: Based on the spatiotemporal energy transfer weight matrix and the net carbon emission equivalent of each unit, generate the spatiotemporal coupled energy consumption field distribution map describing the continuous distribution of carbon emission intensity in spatiotemporal space through the field strength calculation model; In this embodiment of the invention, a continuous spatiotemporal coupled energy consumption field distribution map is generated by relying on the constructed spatiotemporal energy transfer weight matrix and the net carbon emission equivalent of each voxel unit through a spatiotemporal coupled field strength calculation model. A global interpolation field strength calculation model is adopted, and the model structure is as follows: ,in Let the carbon emission field strength be the value at any four-dimensional spacetime coordinate point. The weight values ​​corresponding to the spatiotemporal energy transfer weight matrix are: Net carbon emissions equivalent for discrete four-dimensional voxel units. The total number of voxel units in the entire domain. The model training data consists of a spatiotemporal energy transfer weight matrix dataset and a net carbon emission equivalent dataset of all voxel units. The algorithm logic is based on the carbon emission data of discrete three-dimensional spatiotemporal voxel units, combined with the spatiotemporal transfer weights between units to complete the full-domain four-dimensional spatiotemporal interpolation calculation, fill in the blank spatiotemporal region field strength values ​​between discrete three-dimensional units, eliminate the discrete boundary effect of units, realize the transformation of carbon emission intensity from discrete three-dimensional unit values ​​to a continuous four-dimensional spatiotemporal field, fully characterize the carbon emission differences of steel structures with different vertical floor heights, different planar positions, and different construction periods, and finally generate a spatiotemporal coupled energy consumption field distribution map that can intuitively characterize the distribution, diffusion and accumulation characteristics of carbon emission intensity in the entire spatiotemporal domain of steel structure engineering.

[0024] Step S25: Perform gradient analysis and contour line extraction on the spatiotemporal coupled energy consumption field distribution map to identify the peak carbon emission intensity region and its spatiotemporal boundary, and generate the high-energy-consuming spatiotemporal unit set; at the same time, perform streamline tracing and convergence point analysis on the carbon flow vector field in the spatiotemporal coupled energy consumption field distribution map to generate the key carbon flow path network describing the main migration paths of carbon emissions between stages and spatial units.

[0025] In this embodiment of the invention, a high-energy-consuming spatiotemporal unit set and a key carbon flow path network are extracted from the generated four-dimensional spatiotemporal coupled energy consumption field distribution map using a two-layer data analysis mechanism. Gradient analysis is performed using a four-dimensional spatiotemporal field strong gradient calculation model, the model structure of which is as follows: This method is used to solve for the rate and direction of change in carbon emission intensity within a four-dimensional spatiotemporal field. Combining contour line-based threshold extraction, it identifies peak regions of field intensity, delineates precise three-dimensional spatial and temporal boundaries corresponding to these peaks, and aggregates all four-dimensional spatiotemporal voxel units covered by these peaks to form a high-energy-consuming spatiotemporal unit set. A streamline tracing model is constructed based on the four-dimensional carbon flow vector field. The model structure is as follows: ,in The direction of the four-dimensional carbon emission field strength vector. The model training data consists of a four-dimensional spatiotemporally coupled energy consumption field global vector dataset. The algorithm logic involves continuously tracking the carbon flow migration trajectory along the four-dimensional field strength vector direction, statistically converging core nodes from the trajectory, sorting out the main migration channels between nodes in the three-dimensional space, and aggregating all effective migration channels to form a key carbon flow path network. This accurately characterizes the core migration and convergence patterns of carbon emissions from steel structure engineering at each construction stage and in each three-dimensional space unit, achieving accurate identification of carbon flow paths at the three-dimensional spatial scale.

[0026] Furthermore, the gradient analysis and contour line extraction of the spatiotemporal coupled energy consumption field distribution map described in step S25 includes the following steps: In the spatiotemporal coupled energy consumption field distribution map, a joint scalar field of carbon emission intensity in the spatial and temporal dimensions is defined, and the spatiotemporal joint gradient vector of each point in the joint scalar field is calculated to characterize the rate of change of carbon emission intensity in the three-dimensional spatial and temporal axis directions. In this embodiment of the invention, a spatiotemporal coupled energy consumption field distribution map of a steel structure project is used to construct a spatiotemporal dimensional carbon emission joint scalar field, thereby completing the quantitative calculation of the spatiotemporal joint gradient vector at all points across the entire region. The joint scalar field uses three-dimensional spatial coordinates and one-dimensional time coordinates as its basic dimensions, binds the carbon emission field strength values ​​across the entire map, and constructs a standard four-dimensional joint scalar field model. The model structure is as follows: ,in For a spacetime four-dimensional joint scalar field, This represents the carbon emission intensity field strength at the corresponding spatiotemporal points in the spatiotemporal coupled energy consumption field distribution map. A spatiotemporal joint gradient calculation model is built based on this scalar field, and the model structure is as follows: The algorithm outputs a four-dimensional spatiotemporal joint gradient vector. The model training data is a spatiotemporally coupled energy consumption field global continuous field strength dataset. The algorithm logic is to simultaneously solve the carbon emission intensity change rate of the three spatial axes and the time axis. The four-dimensional vector components correspond to the change characteristics of carbon emission in the horizontal, vertical, and temporal dimensions of steel structure engineering, respectively. It fully quantifies the evolution speed and basic change trend of carbon emission intensity in the entire spatiotemporal dimension, and provides accurate gradient base data for subsequent vector streamline tracking.

[0027] Based on the magnitude and direction of the spatiotemporal joint gradient vector, spatiotemporal vector field streamline tracing is performed on the spatiotemporal coupled energy consumption field distribution map to generate potential diffusion main paths and convergence main paths of carbon emission intensity in spatiotemporal space, and regions where the streamline endpoints are scattered points are identified as initial carbon emission convergence areas. In this embodiment of the invention, by using the solved global spatiotemporal joint gradient vector as the core basis, vector field streamline tracing calculations of the spatiotemporal coupled energy consumption field are performed to accurately extract the spatiotemporal diffusion and accumulation paths of carbon emissions. A spatiotemporal streamline iterative tracing model is constructed, with the following structure: ,in The equations for the position parameters of spacetime streamlines, The step size for streamline iteration. The model uses a spatiotemporal joint gradient vector. The training data consists of a global spatiotemporal gradient vector dataset and a scalar field distribution dataset. The algorithm iterates forward along the gradient vector to generate the main carbon emission diffusion path and iterates backward along the gradient vector to generate the main carbon emission accumulation path, covering the entire spatiotemporal region of the steel structure project. After completing the global streamline traversal and drawing, the distribution of all streamline termination points is statistically analyzed. The spatiotemporal voxel regions corresponding to the endpoints of streamlines without subsequent streamline connections and with independent discrete distributions are uniformly designated as initial carbon emission accumulation areas, thus locking in all basic spatiotemporal regional units with carbon emission accumulation characteristics.

[0028] Based on the initial carbon emission accumulation zone, the spatiotemporal region through which its streamline passes is traced in reverse, and the spatiotemporal second derivative features of the joint scalar field in the spatiotemporal region are extracted, i.e., spatiotemporal curvature features. According to the sign and magnitude of the spatiotemporal curvature features, pseudo-accumulation points formed by local small fluctuations are filtered out from the initial accumulation zone, and stable carbon emission core accumulation zones with positive and significant spatiotemporal curvature are retained. In this embodiment of the invention, by solving for the curvature characteristics of the identified initial carbon emission accumulation zone using the second-order spatiotemporal derivative, pseudo-accumulation point filtering and stable accumulation zone selection are completed. A spatiotemporal curvature calculation model is constructed, and the model structure is as follows: ,in For the spacetime curvature eigenvalue, The second-order partial derivative matrix of the spatiotemporal joint scalar field represents the curvature characteristics of the spatiotemporal variation of carbon emission intensity. The model training data consists of the initial global gradient vector dataset of the convergence zone and the numerical dataset of the scalar field. The algorithm logic is that negative spatiotemporal curvature corresponds to the local fluctuation and decay characteristics of carbon emissions, and is identified as pseudo-convergence points caused by small local fluctuations due to steel structure construction procedures and environmental disturbances, and is directly eliminated; regions with positive spatiotemporal curvature and values ​​higher than a fixed threshold correspond to the evolution characteristics of continuous accumulation and stable carbon emissions, and such spatiotemporal regions are retained, ultimately selecting undisturbed and highly stable core carbon emission convergence zones.

[0029] Critical edge analysis is performed on the core accumulation area of ​​stable carbon emissions and in conjunction with the joint scalar field to generate spatiotemporal topological critical edges; In this embodiment of the invention, by using the core accumulation zone of stable carbon emissions as the core analysis carrier and combining it with a global spatiotemporal joint scalar field to conduct critical edge quantification analysis, the spatiotemporal topological critical edge is accurately extracted. A spatiotemporal criticality discrimination model is constructed, and the model structure is as follows: ,in It is a set of spatiotemporal topological critical edges. This is the preset critical field strength threshold for carbon emissions. The model ensures that edge points exhibit characteristics of carbon emission evolution trends. The training data consists of a stable core convergence zone scalar field dataset and a global field strength threshold dataset. The algorithm logic involves traversing all spatiotemporal points around the core convergence zone, selecting points whose field strength values ​​are equal to the critical threshold and exhibit gradient changes, and connecting all points to form a continuous and closed spatiotemporal boundary. This boundary is the spatiotemporal topological critical edge separating high-carbon emission areas from conventional carbon emission areas, accurately defining the spatiotemporal boundary range of the core convergence zone.

[0030] The set of continuous spatiotemporal voxels defined by the core accumulation area of ​​stable carbon emissions and its corresponding spatiotemporal topological critical edge is defined as the set of high-energy-consuming spatiotemporal units.

[0031] In this embodiment of the invention, the final definition and aggregation of high-energy-consuming spatiotemporal units are achieved by integrating the stable carbon emission core accumulation zone with the spatiotemporal topological critical edge. A spatiotemporal region delineation model is constructed, and the model structure is as follows: ,in As a high-energy-consuming spatiotemporal unit set, To stabilize the spatiotemporal region of the core convergence area, It is a critical edge of spatiotemporal topology. To standardize spatiotemporal voxel units, the model training data consists of a core convergence zone dataset, a topological critical edge boundary dataset, and a global voxel unit affiliation dataset. The algorithm logic involves identifying all continuous spatiotemporal voxel units completely located within the core convergence zone and those bearing topological critical edges, and aggregating all eligible voxel units to form a complete and continuous set of spatiotemporal units. This set is uniformly defined as the high-energy-consuming spatiotemporal unit set for steel structure engineering, accurately identifying the core spatiotemporal regions with concentrated carbon emissions and high energy consumption throughout the entire life cycle of the project, providing precise targeting basis for subsequent low-carbon optimization and energy consumption management.

[0032] Furthermore, the critical edge analysis for the stable carbon emission core accumulation area, combined with the joint scalar field, includes the following steps: Within the core accumulation zone of stable carbon emissions, the spatial-time geodesic path from each point on the edge of the accumulation zone to its geometric center is calculated based on the joint scalar field. In this embodiment of the invention, by using the core accumulation zone of stable carbon emissions as the computational domain, and relying on a global four-dimensional spatiotemporal joint scalar field to perform spatiotemporal geodesic path solving calculations, a set of standard paths from the edge of the accumulation zone to the geometric center is accurately constructed. A four-dimensional spatiotemporal geodesic calculation model is constructed, and the model structure is as follows: ,in For the spatiotemporal geodesic path, For the four-dimensional spatiotemporal points at the edge of the convergence zone, The four-dimensional geometric center of the convergence region, For the balance of spatiotemporal dimensions, fixed coefficients, The step size for path integration is defined as follows. The model training data consists of a four-dimensional spatiotemporal coordinate dataset of the core accumulation area of ​​stable carbon emissions and a global point dataset of the joint scalar field. The algorithm logic is to fix the unique geometric center point of the accumulation area, traverse the sampling points of the entire edge of the accumulation area, and solve for the optimal spatiotemporal path from all edge points to the center point through the minimum spatiotemporal distance integration criterion. The obtained path simultaneously constrains the three-dimensional spatial distance and temporal span, which fits the spatiotemporal coupling evolution characteristics of carbon emissions in steel structure engineering, forming a set of spatiotemporal geodesic paths that cover the entire accumulation area, are non-redundant, and are standardized, providing a standard path carrier for subsequent intensity attenuation feature extraction.

[0033] Carbon emission intensity values ​​are sampled along each of the aforementioned space-time geodesic paths to construct a path carbon emission intensity decay sequence. The path carbon emission intensity decay sequence is then normalized according to the path length to generate a standard decay curve. In this embodiment of the invention, carbon emission intensity sampling and standardized decay curve construction are completed by using all spatiotemporal geodesic paths as sampling carriers, thereby quantifying the spatiotemporal decay law of carbon emission intensity within the accumulation zone. A path intensity sampling model is constructed, and the model structure is as follows: ,in The sampled value of carbon emission intensity corresponding to the path step size. For the numerical values ​​of the four-dimensional spacetime joint scalar field, The step size is used for path integration. The model training data consists of a global spatiotemporal geodesic path coordinate dataset and a joint scalar field strength dataset. The algorithm logic involves uniformly distributing sampling nodes along each geodesic path, collecting the corresponding carbon emission intensity values ​​for each node, and forming the original path carbon emission intensity decay sequence. A path length normalization model is employed. Perform a uniform scale mapping on all sequences, where For standard normalization step size, To determine the total spatiotemporal length of a single geodesic path, the original sequences with different spatiotemporal spans are uniformly mapped to a fixed interval, eliminating dimensional interference caused by differences in path length. This generates a standard carbon emission intensity decay curve with uniform dimensions that can be compared laterally, accurately characterizing the decay pattern of carbon emission intensity from the center to the edge within the core convergence area of ​​the steel structure.

[0034] Calculate the first derivative of each standard decay curve to obtain the decay rate curve, and further calculate the curvature of the decay rate curve on the space-time geodesic path to generate a decay acceleration sequence. In this embodiment of the invention, the dynamic characteristics of carbon emission intensity decay are quantified by solving the decay rate and decay acceleration characteristics layer by layer based on the standard decay curve. A first-order derivative rate solution model is constructed, and the model structure is as follows: ,in The rate of carbon emission intensity reduction under the standardized pathway, The values ​​represent the standard decay curves. This model yields the decay rate curve for each path, reflecting the real-time rate of carbon emission intensity decay. A path curvature calculation model is further constructed to solve for the decay acceleration rate; the model structure is as follows. ,in The numerical values ​​represent the decay acceleration sequence. , These are the first and second derivatives of the decay rate curve, respectively. The model training data consists of all standard decay curve datasets. The algorithm logic continuously solves for rate and curvature along each path, generating a complete decay acceleration sequence. This accurately captures the rapid and steady-state changes in the carbon emission decay process of steel structures, providing time-domain foundational data for frequency domain feature analysis.

[0035] The attenuation acceleration sequence is subjected to Fourier transform, and the harmonic components of the attenuation process are analyzed from the frequency domain perspective. The ratio of high-frequency harmonic energy representing nonlinear abrupt change characteristics to low-frequency harmonic energy representing smooth transition characteristics is extracted as the abrupt change characteristic parameter corresponding to the space-time geodesic path. In this embodiment of the invention, frequency domain transformation analysis is performed on the obtained decay acceleration sequence to extract nonlinear abrupt change characteristic parameters of path carbon emission decay. A discrete Fourier transform model is constructed, with the following model structure: ,in For frequency domain harmonic components, It is a discrete decay rate acceleration sequence. This represents the total number of sampling points in the sequence. This is a frequency domain index. The model training data is a time-series dataset of global path decay acceleration. The algorithm logic is to transform the time-domain decay characteristics into frequency-domain harmonic components through Fourier transform, divide the data into fixed low-frequency and high-frequency intervals, and solve for the total energy of low-frequency and high-frequency harmonics by integration. A model for calculating abrupt change characteristic parameters is constructed. ,in These are path mutation characteristic parameters. High-frequency harmonic energy, The higher the proportion of high-frequency energy, the more significant the nonlinear abrupt change in carbon emission attenuation, thus quantifying the difference in carbon emission abrupt change for each spatiotemporal path.

[0036] The abrupt change characteristic parameters of each space-time geodesic path are mapped back to their origin position at the edge of the convergence zone, and cluster analysis is performed on the edge to connect the positions where the abrupt change characteristic parameters are significantly higher than the neighborhood mean, forming the spatiotemporal topological critical edge.

[0037] In this embodiment of the invention, by binding the solved path mutation feature parameters to the corresponding edge origin positions, cluster analysis is used to accurately extract critical edges and define the spatiotemporal topological boundary of stable convergence regions. A feature mapping model is constructed, and the model structure is as follows: This method achieves a one-to-one binding between single-path mutation parameters and edge origin points, forming a feature parameter distribution dataset for all points along the convergence zone edge. Density clustering is used to partition edge point features; the core model structure is as follows. ,in For edge point feature density, For the neighborhood interval of the point, The mean of the parameters across the entire edge region is used. The model training data consists of edge point coordinate datasets and global mutation feature parameter datasets. The algorithm logic is to statistically analyze the differences in neighborhood features of each point, filter out mutation-type edge points with parameter values ​​higher than the neighborhood mean, and continuously connect all filtered points to form a closed, continuous, and carbon emission mutation-fitting four-dimensional spatiotemporal boundary curve. Finally, it generates a spatiotemporal topological critical edge that accurately divides the high-carbon emission core region and the normal region.

[0038] Furthermore, the streamline tracing and convergence point analysis of the carbon flow vector field in the spatiotemporal coupled energy consumption field distribution map described in step S25 includes the following steps: In the spatiotemporal coupled energy consumption field distribution map, a dynamic propagation network of carbon particles in spatiotemporal space is constructed based on the carbon flow vector field, and carbon potential energy reflecting the dynamic balance between carbon emission absorption and release is introduced at the nodes of the dynamic propagation network. In this embodiment of the invention, a spatiotemporal dynamic propagation network for carbon particles is constructed based on the existing four-dimensional spatiotemporal coupled energy consumption field distribution map and global carbon flow vector field of the steel structure engineering, thereby realizing a networked and concrete representation of the carbon emission propagation process. Using global four-dimensional spatiotemporal voxel units as the basic nodes of the network, and the spatiotemporal energy transfer relationships between voxel units as the directed edges, the basic topology of the dynamic propagation network is completed. All nodes are bound to corresponding three-dimensional spatial coordinates, time-series labels, and net carbon emission equivalent attributes, and all directed edges are bound to spatiotemporal transfer weight attributes. A carbon potential energy model is constructed to characterize the node equilibrium features. The model structure is as follows: ,in Carbon potential energy at four-dimensional spacetime nodes. Net carbon emission equivalent per unit. The magnitude of the spatiotemporal joint gradient vector. , These are fixed coefficients for the carbon emission baseline weight and the spatiotemporal diffusion weight, respectively. The model training data consists of a global voxel unit net carbon emission equivalent dataset and a carbon flow vector field gradient dataset. The algorithm logic integrates the inherent carbon emission stock of nodes with spatiotemporal diffusion trend characteristics, quantifies the dynamic balance state of carbon emission release and absorption at each node, assigns high carbon potential energy nodes to the dominant state of carbon emission release, and low carbon potential energy nodes to the dominant state of carbon emission absorption, and fully constructs a spatiotemporal dynamic propagation network of carbon particles with potential energy balance properties, providing a standardized topological carrier for carbon flow migration simulation.

[0039] Based on the gradient direction of the carbon potential energy, the adaptive and optimal migration trajectory of carbon particles in the spatiotemporal network is simulated to generate multiple probabilistic migration paths from high carbon potential energy nodes to low carbon potential energy nodes. In this embodiment of the invention, based on the global node carbon potential energy distribution characteristics and relying on the potential energy gradient driving mechanism, the adaptive optimization migration trajectory simulation of carbon particles is completed, generating a standardized carbon flow migration path set. A carbon particle migration driving model is constructed, and the model structure is as follows: ,in This represents the spatiotemporal migration vector of carbon particles, with its direction strictly conforming to the decreasing direction of the carbon potential energy gradient, matching the natural diffusion pattern of carbon emissions from high-aggregation areas to low-aggregation areas. A path probability generation model is introduced to quantify the migration probability; the model structure is as follows. ,in For nodes To node The migration probability, , The value represents the carbon potential energy of adjacent nodes. The normalization coefficients are fixed across the entire domain. The model training data consists of a dynamic propagation network node carbon potential energy dataset and a node spatiotemporal topology association dataset. The algorithm logic is to take high carbon potential energy nodes as the starting point of the path, and iteratively calculate the migration node by node along the gradient descent direction. Based on the potential energy difference between nodes, the corresponding migration probability is matched, and multiple carbon particle migration paths covering the entire spatiotemporal region with probability weights are iteratively generated to accurately restore the objective law of spontaneous spatiotemporal migration of carbon emissions in steel structure engineering.

[0040] The probabilistic migration paths are subjected to spatiotemporal consistency checks. Paths with spatiotemporal overlap exceeding a preset threshold are merged, and the expected carbon flux value of each merged path is calculated. The expected carbon flux value is obtained by the integral of the product of the net carbon emission equivalent of the nodes through which the path passes and its spatiotemporal connectivity strength. In this embodiment of the invention, by implementing spatiotemporal consistency verification and path fusion processing for all probabilistic migration paths, the expected carbon flux value of the merged paths is accurately calculated. A path spatiotemporal overlap discrimination model is constructed, and the model structure is as follows: ,in For path With path The degree of spatiotemporal overlap, , For the four-dimensional spatiotemporal voxel set covered by the two paths, a fixed overlap threshold is set. Independent paths with overlap below the threshold are retained, while homogeneous paths with overlap above the threshold are merged to eliminate redundant interference from duplicate paths. A carbon flux expectation integral model is constructed, with the following model structure: ,in This represents the expected value of the path carbon flux. For the complete path spatiotemporal domain after merging, This represents the net carbon emission equivalent of the nodes along the path. The spatiotemporal connectivity strength between nodes is represented by the model training data, which includes a global probabilistic transfer path dataset, a voxel net carbon emission equivalent dataset, and a spatiotemporal transfer weight dataset. The algorithm logic involves performing continuous integral operations along the spatiotemporal span of the path to quantify the carbon emission transmission carrying capacity of a single path and obtain the standardized carbon flux characteristics of all merged paths.

[0041] Based on the expected carbon flux of each path, a carbon flux competition relationship graph between paths is constructed, and key edges and key nodes that, when removed, lead to a large number of alternative paths detouring and significantly reduce the global carbon flux efficiency are identified in this graph. In this embodiment of the invention, a path-level carbon flow competition graph is constructed using the expected value of carbon flow flux for each path as the core feature, thereby completing the identification of global key carbon flow units. Each independent merging path is used as a competing network node, and the spatiotemporal overlap of paths and the carbon flow transport resource crowding relationship are used as directed edges for competition, thus constructing a topological graph of carbon flow competition between paths. A path competition intensity quantification model is constructed, with the following model structure: ,in For path With path The intensity of competition, , Let be the expected carbon flux values ​​for the two paths. The model training data consists of all path flow datasets and path spatiotemporal overlap feature datasets. The algorithm logic quantifies the degree of resource crowding and flow competition between paths, and performs removal perturbation analysis on each edge and node in the competition graph. It then statistically analyzes the total global carbon flow loss and path detour scale after the removal of the unit, identifying the topological edges and nodes that will cause a significant decrease in global carbon flow efficiency and generate large-scale path detours as the core key units of carbon flow transmission.

[0042] From the carbon flow competition relationship graph between the paths, the key edges and key nodes are extracted and reconstructed according to their original spatiotemporal connection relationships in the dynamic propagation network to form the key carbon flow path network describing the main migration paths of carbon emissions between stages and spatial units.

[0043] In this embodiment of the invention, by extracting all key edges and key nodes, and relying on the four-dimensional spatiotemporal topological relationships of the original dynamic propagation network, the reconstruction and shaping of the key carbon flow path network is completed. A topology reconstruction mapping model is constructed, and the model structure is as follows: ,in For the reconstructed set of key carbon flow networks, For the set of key nodes, For the set of key edges, The original spatiotemporal connection weights are used. The model training data consists of a key topological unit dataset and a dynamic propagation network original spatiotemporal topological dataset. The algorithm logic strictly replicates the three-dimensional spatial location, temporal association order, and spatiotemporal connection strength of key nodes, eliminates globally redundant, low-impact, and substitutable ordinary carbon flow paths, and retains only the core topological structure that is irreplaceable and dominates global carbon flow transmission. This reconstructs a key carbon flow path network that is structurally simplified, feature-focused, and perfectly matches the spatiotemporal carbon emission migration law of steel structure engineering, accurately representing the core spatial transmission channels and temporal migration paths of carbon emissions throughout the entire life cycle of the project.

[0044] Furthermore, step S3 includes the following steps: Step S31: For the set of high-energy-consuming spatiotemporal units, extract the corresponding engineering stage, process type, equipment type and material type features to form a high-energy-consuming unit feature vector; In this embodiment of the invention, the high-energy-consuming spatiotemporal unit set identified above is used as the core analysis object. Based on the full lifecycle engineering attributes bound to each four-dimensional spatiotemporal voxel unit, multi-dimensional feature extraction is completed and a standardized high-energy-consuming unit feature vector is constructed. Four core feature dimensions are extracted: the full lifecycle stage of the steel structure project to which the unit belongs, the on-site construction process type, the type of operating machinery and equipment, and the material and specification type of engineering materials, covering all core business attributes of high-energy-consuming and high-carbon-emission units. A structured feature vector construction model is constructed, with the model structure as follows: ,in This is the feature vector of a high-energy-consuming unit. Encode feature quantities for the engineering phase. Encode the characteristic quantities for construction process types, Encode the characteristic quantities for construction equipment types. The model encodes feature quantities for engineering material types. The training data consists of an engineering attribute dataset bound to a set of high-energy-consuming spatiotemporal units and a full-process technology and equipment material ledger dataset. The algorithm logic is to uniformly structure and encode unstructured stage, process, equipment, and material attributes, matching them with the specific technology and equipment material systems for each stage of steel structure manufacturing, transportation, construction, operation and maintenance, and demolition. This results in a unique, dimensionally regularized feature vector corresponding to each high-energy-consuming spatiotemporal unit, with full attribute coverage, providing standardized feature basis for subsequent accurate matching of low-carbon processes.

[0045] Step S32: From the feature library of low-carbon process modes of steel structures, match multiple low-carbon process modes that are aligned with the feature vector of the high-energy-consuming unit in terms of stage and process type, and parse the low-carbon technology package composition of each low-carbon process mode to generate a set of candidate low-carbon process modes. In this embodiment of the invention, by relying on the high-energy-consuming unit feature vector generated in S31, and connecting it with the previously constructed feature library of low-carbon steel structure process modes, accurate matching of homogeneous low-carbon process modes and construction of candidate mode sets are achieved. A two-layer rigid matching rule between stage and process is established, prioritizing the matching of low-carbon process modes in the feature library whose engineering stage, core process type, and feature vector are completely aligned, discarding unsuitable modes with misaligned stages or incompatible processes, and locking in multiple sets of suitable low-carbon process modes. A feature similarity matching model is constructed to complete accurate screening; the model structure is as follows: ,in To match similarity for process patterns, , Fixed weighting coefficients for stage and process dimensions. This is the feature consistency discrimination function. The model training data consists of a low-carbon process mode feature library full-domain feature dataset and a high-energy-consuming unit feature vector dataset. The algorithm logic is to calculate the matching similarity between each process mode in the feature library and the target feature vector, collect all qualified and adapted process modes, and analyze the low-carbon technology package consisting of equipment energy-saving technology, material emission reduction technology, process optimization technology, and energy management technology embedded in each mode layer by layer. Finally, it integrates all the analyzed and adapted process modes to form a candidate low-carbon process mode set that is dimensionally regular, scenario-adaptable, and specifically matched to the working conditions of high-energy-consuming units.

[0046] Step S33: Perform structural vulnerability analysis on the critical carbon flow path network, identify critical path nodes and edges in the network that are highly dependent on carbon flow and prone to blockage, and calculate their carbon flow resistance coefficient. In this embodiment of the invention, a global structural vulnerability quantification analysis is conducted on the key carbon flow path network reconstructed above to accurately identify the core weak units of carbon flow transmission and solve for the carbon flow resistance coefficient. Using the four-dimensional spatiotemporal topology of the key carbon flow path network as the analysis carrier, core nodes and topological edges within the network that bear the main carbon flow transmission flux, have low path substitution rates, and high flux carrying capacity are identified as key topological units that are highly dependent on carbon flow and prone to flux congestion. A carbon flow resistance calculation model is constructed, with the following structure: ,in The resistance coefficient of a single carbon flow. This represents the expected value of the real-time carbon flux of the unit. This is the maximum throughput threshold for the unit. The length of the four-dimensional spatiotemporal path topology. , Fixed weighting coefficients are applied to flux resistance and spatiotemporal resistance. The model training data consists of a dataset of key carbon flow path network topology, a dataset of carbon flow in each unit, and a dataset of spatiotemporal path lengths. The algorithm logic combines the unit flux saturation level with the spatiotemporal transmission span to quantify the passage resistance strength of each key node and edge. The higher the resistance coefficient value, the higher the probability of carbon flow congestion and the stronger the network vulnerability, thus completing the full coverage calculation of the resistance coefficient of key units across the entire domain.

[0047] Step S34: Apply the candidate low-carbon process mode set to the high-energy-consuming spatiotemporal unit set for simulation replacement, evaluate the reduction potential of each mode on the net carbon emission equivalent of the unit, and combine its influence on the carbon flow resistance coefficient of relevant nodes and edges in the key carbon flow path network to perform multi-objective collaborative optimization analysis to generate multi-objective collaborative optimization analysis results. In this embodiment of the invention, the obtained set of candidate low-carbon process modes is applied one by one to a set of high-energy-consuming spatiotemporal units to conduct operating condition simulation replacement, and multi-objective collaborative optimization analysis is carried out by combining the carbon emission reduction effect and the change in carbon flow resistance. A unit carbon reduction potential accounting model is constructed, and the model structure is as follows: ,in This represents the net carbon emission equivalent reduction after process replacement. To replace the original net carbon emission equivalent of the previous unit, Optimize the net carbon emission equivalent of the unit after low-carbon process replacement. Construct a quantitative model for carbon resistance improvement, with the following model structure: ,in This represents the improvement in carbon circulation resistance. , These represent the resistance coefficients of key units before and after process replacement. The model training data includes candidate low-carbon process parameter datasets, basic carbon emission data of high-energy-consuming units, and key carbon flow unit resistance coefficient datasets. The algorithm logic simultaneously calculates the carbon reduction benefits and carbon flow unblocking benefits of a single process, taking into account both static carbon emission reduction of units and dynamic carbon flow efficiency improvement of the network. It completes collaborative comparative analysis of multiple processes and multiple indicators, generating standardized multi-objective collaborative optimization analysis results that include the dual optimization benefits of each process mode.

[0048] Step S35: Based on the results of the multi-objective collaborative optimization analysis, select the optimal combination of process modes and their application strategies in terms of carbon reduction potential and improvement of system carbon flow, forming the process low-carbon optimization scheme cluster; at the same time, perform dynamic carbon resistance analysis and carbon flow bottleneck identification on the key carbon flow path network, and generate process reengineering, material substitution or energy dispatch strategies aimed at reducing carbon resistance at key nodes and edges, forming the carbon flow optimization and control strategy set.

[0049] In this embodiment of the invention, by relying on the results of multi-objective collaborative optimization analysis, the optimal process combination is screened and a carbon flow control strategy is generated, forming a complete low-carbon optimization system. A comprehensive benefit evaluation model is constructed to complete process optimization, and the model structure is as follows: ,in To comprehensively optimize the benefit value, To determine the weighting coefficient for carbon reduction benefits, the optimal combination of process modes with the best overall benefits is identified by ranking the benefit values ​​of all process modes across the entire domain. This combination is then matched with the corresponding unit's spatiotemporal operating conditions to form a cluster of adaptable low-carbon optimization process solutions. A dynamic carbon resistance analysis model is built to identify bottlenecks. The model structure is as follows: This study characterizes the temporal dynamic changes in carbon resistance of key units, identifying persistently high-resistance and dynamically congested carbon flow bottleneck units. The model training data includes a multi-objective optimization result dataset and a time-series dataset of dynamic resistance in the carbon flow network. The algorithm logic involves matching differentiated control measures such as process reengineering, low-carbon material replacement, and time-sharing energy scheduling to different bottleneck types, aggregating all targeted control strategies, and forming a standardized carbon flow optimization control strategy set. This achieves bidirectional global optimization of carbon reduction in high-energy-consuming units of steel structure engineering and efficient carbon flow network passage.

[0050] Furthermore, the dynamic carbon resistance analysis and carbon flow bottleneck identification for the key carbon flow path network described in step S35 includes the following steps: For each node and edge in the critical carbon flow path network, the corresponding process stage attribute, material attribute and energy type attribute are attached, and the unit carbon emission intensity parameter under the current attribute configuration is obtained according to the engineering spatiotemporal characteristic sequence, as the basic carbon resistance coefficient. In this embodiment of the invention, by using a pre-constructed key carbon flow path network as a carrier, fixed-dimensional inherent attribute labels are bound to all nodes and directed edges within the network. This comprehensively covers the attributes of the entire process stage of steel structure engineering, the attributes of construction materials, and the energy type attributes of operational energy consumption, achieving refined attribute labeling of carbon flow transmission units. Based on the pre-generated spatiotemporal feature sequence of the engineering project, the four-dimensional spatiotemporal working conditions corresponding to each node and edge are matched, and the standardized unit carbon emission intensity parameter under the current attribute configuration system is extracted to quantify the inherent carbon emission resistance characteristics of the unit. A basic carbon resistance coefficient calculation model is constructed, with the model structure as follows: ,in Based on the carbon resistivity, The parameter is the unit carbon emission intensity. , , These are the inherent resistance coefficients corresponding to the process, material, and energy properties, respectively. , , , The weighting ratio is fixed. The model training data consists of a dataset of key carbon flow network topology attributes, a dataset of engineering spatiotemporal feature sequences, and a dataset of carbon emission benchmark parameters for each attribute dimension. The algorithm logic integrates the inherent attributes of the unit with the real-time carbon emission intensity to quantify the static carbon flow obstruction capability of each spatiotemporal unit under no external disturbance, providing a standardized foundation for subsequent dynamic carbon obstruction superposition calculations.

[0051] Virtual carbon flow pulses are introduced into the critical carbon flow path network to simulate their propagation process from upstream nodes to downstream nodes. The delay and dissipation effects caused by property changes encountered by the virtual carbon flow on the propagation path are recorded, and the dynamic carbon resistance increment caused by property coupling interaction is quantified. In this embodiment of the invention, by embedding standardized virtual carbon flow pulses into a critical carbon flow path network with full attribute coverage, and relying on the network's four-dimensional spatiotemporal topology propagation rules, the continuous propagation process of the pulses from upstream spatiotemporal nodes to downstream spatiotemporal nodes is simulated, accurately capturing the dynamic carbon resistance change characteristics caused by attribute coupling interactions. A virtual carbon flow pulse propagation model is constructed, and the model structure is as follows: ,in The downstream spatiotemporal location pulse intensity, The initial pulse intensity, For the spatiotemporal joint gradient field, This represents a four-dimensional spatiotemporal propagation micro-path. The model training data includes a network global attribute distribution dataset, a spatiotemporal gradient field dataset, and a pulse propagation topology connectivity dataset. The algorithm logic tracks attribute change nodes encountered throughout the pulse propagation process, such as process switching, material replacement, and energy type conversion, and statistically analyzes the time delay duration and flux dissipation ratio generated during the propagation process. A dynamic carbon resistance incremental quantification model is constructed. ,in For dynamic carbon resistance increment, To delay the transmission duration, For flux dissipation rate, , To fix the conversion factor, the dynamic increase in passage obstacles caused by attribute coupling interaction is accurately quantified.

[0052] By integrating the basic carbon resistance coefficient and the dynamic carbon resistance increment, the comprehensive carbon resistance value of each node and edge in the path network is calculated, and the nodes and edges are classified according to the comprehensive carbon resistance value. Nodes and edges with a comprehensive carbon resistance value that is significantly higher than the average level of the same stage / type are identified and marked as preliminary carbon flow bottlenecks. In this embodiment of the invention, by integrating the obtained static basic carbon resistance coefficient and dynamic carbon resistance increment, the comprehensive carbon resistance value of all nodes and edges in the network is solved, achieving preliminary and accurate identification of carbon flow bottlenecks. A comprehensive carbon resistance fusion calculation model is constructed, and the model structure is as follows: ,in The final comprehensive carbon resistance value of a unit is determined by integrating static inherent resistance and dynamic interactive resistance through linear superposition, comprehensively characterizing the overall resistance level of carbon flow in a spatiotemporal unit. The model training data includes a global basic carbon resistance coefficient dataset, a dynamic carbon resistance increment dataset, and a network unit spatiotemporal classification dataset. The algorithm logic involves classifying and statistically analyzing units according to engineering stage, material type, and energy type, calculating the global average carbon resistance value for each category, and establishing a hierarchical threshold standard. The comprehensive carbon resistance value of each unit is compared horizontally with the average value of the same category. Spatiotemporal nodes and topological edges whose comprehensive carbon resistance values ​​consistently exceed the category average are identified and uniformly marked as preliminary carbon flow bottlenecks, accurately screening core spatiotemporal units that significantly hinder carbon flow transmission in steel structure engineering.

[0053] For the initial carbon flow bottleneck marked, the applicable process, material and energy alternatives in the process low-carbon optimization scheme cluster are traced back, the expected contribution of each alternative to reducing its overall carbon resistance is evaluated, and alternative optimization schemes with different priorities are generated according to the expected contribution. In this embodiment of the invention, for all marked preliminary carbon flow bottleneck units, adaptive optimization methods within the process low-carbon optimization scheme cluster are back-matched, covering three core optimization options: process iteration, material substitution, and energy dispatch. The drag reduction contribution of each optimization option is quantified one by one. An optimization contribution evaluation model is constructed, with the following model structure: ,in For the first The expected contribution of each optimization option. To optimize the overall carbon resistance value, This is a simulated comprehensive carbon drag value after the implementation of corresponding optimization options. The model training data consists of a preliminary bottleneck unit dataset, a low-carbon optimization scheme cluster process parameter dataset, and various optimization methods carbon drag correction parameter datasets. The algorithm logic simulates the independent implementation of a single optimization option, calculates the reduction magnitude of each option on the bottleneck carbon drag, and completes hierarchical division based on the magnitude of contribution value. It generates a gradient and differentiated queue of alternative optimization schemes by sorting them from high to low, ensuring that core drag reduction optimization methods are prioritized and adapting to the targeted optimization needs of carbon flow bottlenecks in steel structure engineering.

[0054] The network cascading effect of the proposed optimization schemes is evaluated. The optimization of a single bottleneck node is analyzed to mitigate or transfer the overall carbon resistance of adjacent nodes and edges. Schemes that can reduce the global carbon resistance of the network without triggering new high-order bottlenecks are selected. Specific process reengineering suggestions, material substitution lists, and energy scheduling sequences are formed, which together constitute the carbon flow optimization and control strategy set.

[0055] In this embodiment of the invention, by conducting a global extrapolation of the network cascading effect of gradient-based alternative optimization schemes, carbon flow transfer and secondary bottleneck problems caused by local optimization are avoided. The globally optimal optimization strategy is then selected and integrated to form a standardized set of control strategies. A carbon flow network cascading effect model is constructed, with the following structure: ,in This represents the global carbon resistance change in the network. This represents the change in carbon resistance between adjacent units. This is to reduce the carbon resistance of the bottleneck unit. This represents the global impact balance coefficient. The model training data includes candidate scheme parameter datasets, network-wide unit carbon resistance value datasets, and spatiotemporal topology association datasets. The algorithm logic deduce the interconnected changes in global carbon resistance after optimizing a single bottleneck, eliminating optimization schemes that cause a surge in carbon resistance in adjacent units or generate higher-order secondary bottlenecks, while retaining optimization strategies that can achieve an overall reduction in global carbon resistance. Finally, effective strategies are solidified, and standardized process reengineering suggestions, a low-carbon material substitution list, and time-sharing energy scheduling sequences are compiled. All effective control methods are integrated to form a complete set of carbon flow optimization and control strategies, achieving efficient and steady-state operation of the carbon flow network across the entire steel structure engineering.

[0056] Furthermore, the step of classifying the carbon resistance of nodes and edges based on the comprehensive carbon resistance value includes the following steps: Based on the process stage attributes and material / energy type attributes of nodes and edges in the critical carbon flow path network, an attribute condition distribution space is constructed. Within the attribute condition distribution space, the mean and standard deviation of the comprehensive carbon resistance values ​​of all nodes and edges of the same stage and type are calculated as the corresponding attribute condition statistical benchmark. In this embodiment of the invention, a multi-dimensional attribute conditional distribution space is constructed by relying on the process stage attribute, material type attribute, and energy type attribute bound to the global nodes and directed edges of the key carbon flow path network, thereby realizing the construction of a refined classification and statistical benchmark for carbon resistance data. This distribution space uses process stage, material category, and energy type as orthogonal dimensions to divide non-overlapping attribute subspaces. All spatiotemporal units within the network are assigned to their corresponding subspaces according to their inherent attributes, completing the attribute partitioning and aggregation of global units. An attribute conditional statistical model is constructed, with the following structure: , ,in This represents the average carbon resistivity of a single attribute subclass. The standard deviation of the carbon resistivity for a single attribute subclass. This represents the total number of spatiotemporal units under the corresponding attribute subclass. This represents the comprehensive carbon resistivity of each unit within a subclass. The model training data consists of a dataset of key carbon flow network unit attributes and a dataset of comprehensive carbon resistivity across the entire domain. The algorithm logic strictly follows clustering rules based on the same process stage, material type, and energy type, calculating statistical characteristics for each class to form a carbon resistivity distribution benchmark specific to each attribute subspace. This avoids benchmark bias caused by mixing attribute units under different operating conditions, providing an accurate reference standard for subsequent deviation calculations.

[0057] For each node and edge, calculate the degree of deviation of its comprehensive carbon resistance value relative to its corresponding attribute condition statistical benchmark, and obtain the local standardized deviation score; In this embodiment of the invention, by using the statistical benchmarks of each attribute subclass as a reference, the deviation level of carbon resistance values ​​of each network node and edge is quantified to generate a standardized local deviation score. A local standardized deviation calculation model is constructed, and the model structure is as follows: ,in For the local standardized deviation score of a single spatiotemporal unit, The measured comprehensive carbon resistance value of the unit. , The mean and standard deviation of the attribute subclass to which the unit belongs. The model training data consists of a comprehensive carbon resistance value dataset for each unit and a statistical benchmark dataset for each attribute subclass. The algorithm logic is to eliminate the interference of differences in carbon resistance benchmarks between different attribute subclasses by calculating the standardized deviation, and uniformly map the degree of carbon resistance deviation of all units to the standard quantization range. Positive values ​​represent that the unit's carbon resistance value is higher than the benchmark level of the same type, and negative values ​​represent that the unit's carbon resistance value is lower than the benchmark level of the same type. This accurately describes the abnormal carbon resistance deviation characteristics of a single unit under the same working conditions, and realizes the horizontal equivalent comparison of the degree of deviation of all units.

[0058] From the spatiotemporal feature sequence of the project, historical fluctuation features associated with each node and edge in the key carbon flow path network are extracted, including the historical variation coefficient of carbon emission intensity and the mutual influence intensity of adjacent path units, and the inherent noise level implied by the historical fluctuation features is calculated. In this embodiment of the invention, by relying on the spatiotemporal characteristic sequence of the engineering project, the temporal fluctuation and spatial correlation characteristics of each unit in the key carbon flow path network are extracted, and the inherent operating noise level of the unit is quantified. Two types of core fluctuation characteristics are fixedly extracted: the historical coefficient of variation of carbon emission intensity and the mutual influence intensity of adjacent path units. A historical coefficient of variation calculation model is constructed, and the model structure is as follows: ,in The historical variation coefficient of unit carbon emission intensity. , These represent the standard deviation and mean of the unit's time-series carbon emission intensity, respectively. A unit interaction intensity model is constructed, with the following structure: ,in The strength of the mutual influence between adjacent units, Assign spatiotemporal connection weights to the units. Construct an intrinsic noise level fusion model. ,in The inherent noise level of the unit. , The fusion weights are fixed. The model training data consists of a long-time series engineering spatiotemporal feature sequence dataset and a unit spatiotemporal adjacency topology dataset. The algorithm logic integrates the features of temporal fluctuations and spatial coupling perturbations to quantify the range of inherent fluctuation noise during the normal operation of the unit.

[0059] Based on the inherent noise level, the local standardized deviation score is subjected to noise adaptive correction to filter out normal deviations caused by historical inherent fluctuations and generate a corrected abnormal deviation index. In this embodiment of the invention, based on the inherent noise level of each unit, an adaptive correction is performed on the local standardized deviation score to eliminate invalid deviations caused by normal fluctuations and obtain true anomaly characteristics. A noise adaptive correction model is constructed, with the following structure: ,in This is the corrected abnormal deviation index. To ensure that the noise correction closely matches the actual deviation trend, the sign function of the deviation direction is used. The inherent noise level of the unit. The original standardized deviation scores are used. The model training data consists of a global unit local deviation score dataset and an inherent noise level dataset. The algorithm logic is to match the normal fluctuation range of each unit's differences, accurately deduct and correct the original deviation scores, completely filter out the normal numerical deviations caused by the construction sequence fluctuations of steel structure engineering and spatial unit coupling disturbances, and retain only the substantial carbon resistance anomalies caused by defects in process, materials, and energy configuration, thus achieving accurate purification of abnormal features.

[0060] The modified abnormal deviation index of all nodes and edges in the critical carbon flow path network is statistically sorted, and the nodes and edges that are ranked in the high percentile and whose absolute deviation exceeds the average level of the neighborhood are selected and marked as the initial carbon flow bottleneck.

[0061] In this embodiment of the invention, a global statistical ranking is performed on the abnormal deviation index of all nodes and edges in the critical carbon flow path network after correction. This, combined with a neighborhood comparison mechanism, completes the initial accurate labeling of carbon flow bottlenecks. A two-layer discrimination mechanism of global ranking and neighborhood verification is constructed, with the core discrimination model structure as follows: ,in As a bottleneck identification marker, To fix the high percentile interval The average anomaly deviation index of the unit's neighborhood is used. The model training data consists of a global unit anomaly deviation index dataset and a unit spatiotemporal neighborhood topology dataset. The algorithm logic is as follows: first, the global index is sorted in descending order to filter high-percentile high-anomaly units; then, units with small local fluctuations are eliminated by comparing the neighborhood mean, retaining core units whose anomaly level is significantly higher than the neighborhood level. Spatiotemporal nodes and topological edges that meet the dual discrimination conditions are uniformly judged as having substantial passage obstacles and preliminary carbon flow bottlenecks caused by abnormal fluctuations, accurately locating the core weak units in the carbon flow transport system of steel structure engineering.

[0062] Furthermore, step S4 includes the following steps: A coupling effect model is established to analyze the enhancement or weakening effect of the implementation of each scheme in the process low-carbon optimization scheme cluster on the effectiveness of related strategies in the carbon flow optimization and control strategy set, and to generate a scheme strategy interaction effect matrix. In this embodiment of the invention, a process-strategy coupling influence model is constructed to quantify the bidirectional coupling effect between each optimization scheme within the process low-carbon optimization scheme cluster and each control strategy within the carbon flow optimization control strategy set. This clarifies the enhancing and weakening effects of scheme implementation on the effectiveness of control strategies, and completes the structured construction of the interaction influence matrix. The coupling influence quantification model is built, and its structure is as follows: ,in For the first Type of process scheme and the first The interaction coefficient of similar regulatory strategies To assess the effectiveness of carbon flow control strategies, To reduce carbon emissions in the process, , These are respectively the intensity of process execution and the intensity of strategy implementation. , Fixed weight coefficients are used to represent the bidirectional influence. The model training data consists of a dataset of carbon reduction characteristics of process schemes, a dataset of carbon flow control strategy resistance reduction efficiency, and a dataset of full-time and space-time unit carbon resistance response. The algorithm logic involves solving the partial derivatives of process variables with respect to strategy effectiveness to determine the enhancement characteristics, and solving the partial derivatives of strategy variables with respect to process benefits to determine the constraint characteristics. Positive coefficients correspond to synergistic enhancement effects, while negative coefficients correspond to antagonistic weakening effects. All pairwise coupling coefficients of processes and strategies are regularly arranged to form a scheme-strategy interaction influence matrix with regular row and column dimensions and clear influence relationships, providing an interaction constraint basis for subsequent multi-objective optimization.

[0063] Based on the interaction matrix of the aforementioned scheme strategy, with the minimization of total carbon emissions throughout the entire life cycle as the core objective and engineering cost, construction period and technical requirements as constraints, a multi-objective optimization configuration model for low-carbon engineering is constructed. In this embodiment of the invention, based on the obtained scheme strategy interaction influence matrix, and with minimizing the total carbon emissions throughout the entire life cycle of steel structure engineering as the core optimization objective, a standardized low-carbon engineering multi-objective optimization configuration model is established, anchored to three types of hard constraints: engineering construction cost, construction period, and on-site technology adaptation. The core objective function structure is as follows: ,in This represents the total carbon emissions over the entire life cycle of the project. This represents the net carbon emission equivalent of a four-dimensional spatiotemporal unit. The constraint function includes three types of fixed constraints: cost constraint (total investment in the implementation of all processes and strategies does not exceed the preset cost control threshold), schedule constraint (the cumulative duration of each stage of the process does not exceed the preset total schedule threshold), and technical constraint (all configuration schemes are adapted to standardized technical conditions for steel structure component processing, hoisting, and operation and maintenance). The model training data consists of process energy consumption parameters, carbon flow control strategy operating parameters, and benchmark datasets of engineering quota costs and schedules. The algorithm logic embeds the interactive coupling relationship between schemes and strategies into the objective constraint system, avoiding optimization failure caused by scheme antagonism, and constructing a multi-objective optimization configuration model that balances carbon reduction goals, engineering feasibility, and construction standardization.

[0064] The process low-carbon optimization scheme cluster and carbon flow optimization control strategy set are used as decision variables and input into the low-carbon engineering multi-objective optimization configuration model. The model is solved by intelligent optimization algorithm to obtain the Pareto optimal solution set under the constraints. The Pareto optimal solution set is a series of non-dominated low-carbon engineering configuration schemes. In this embodiment of the invention, all elements of the process low-carbon optimization scheme cluster and the carbon flow optimization control strategy set are used as multi-dimensional decision variables of the model. These are input into a multi-objective optimization configuration model to perform global optimal solution, and a constrained multi-objective intelligent optimization algorithm is used to complete iterative optimization. The algorithm iteratively updates the model structure as follows: ,in For the first Allocation vector of decision variables, This represents the gradient reduction amount for the carbon emission target. Correct the offset for the interaction matrix. , The iteration step size is a fixed coefficient. The model training data consists of a dataset of all feasible process and strategy combinations, a dataset of multi-objective constraint thresholds, and a dataset of interaction influence matrix coefficients. The algorithm logic is to traverse all combinations of working conditions within the constraints of cost, schedule, and technical boundaries, continuously eliminating configuration combinations with high carbon emissions, high conflict, and low efficiency, and retaining individual solutions that are mutually independent and cannot be optimized unilaterally, ultimately converging to generate a steady-state Pareto optimal solution set. Each set of vectors within the solution set corresponds to an independent combination mode of process matching and carbon flow control, forming a set of low-carbon engineering configuration schemes that are fully covered, non-redundant, and satisfy all engineering constraints.

[0065] The Pareto optimal solution set is evaluated for engineering feasibility and screened for technology maturity. The configuration scheme with the best overall performance is selected, and its parameters are decoded to generate the low-carbon engineering optimization configuration parameters for the whole life cycle. In this embodiment of the invention, a two-layer screening mechanism is used to select the optimal configuration scheme based on the Pareto optimal solution set, thereby finalizing the low-carbon parameters. An engineering feasibility assessment model is constructed, with the following structure: ,in To score the feasibility of the plan, , , Scores were given for cost, schedule, and technology compatibility. , , A technology maturity scoring model is constructed using fixed weighting coefficients. This model quantifies and scores based on the adoption rate of the technology, the stability of the strategy operation, and the adaptability to different operating conditions. The model training data includes Pareto solution set configuration parameter datasets, engineering implementation evaluation benchmark datasets, and industry low-carbon technology maturity benchmark datasets. The algorithm logic involves quantitatively scoring the feasibility and maturity of each non-dominated solution within the solution set, and then aggregating the single optimal configuration solution with the highest overall score. The high-dimensional decision vector of the optimal solution is then decoded to obtain process ratio parameters, material replacement parameters, energy scheduling parameters, carbon flow control parameters, and time-series configuration parameters. These parameters are then integrated to form a complete, accurate, and directly implementable full-lifecycle low-carbon engineering optimization configuration parameter set.

[0066] The low-carbon construction simulation model is initialized based on the optimized configuration parameters of the low-carbon project. The model is driven to dynamically simulate the construction process, resource allocation, and carbon emissions in the subsequent stages of the project. Based on the gap between the carbon emission trajectory output by the simulation and the preset low-carbon target, the achievability assessment result of the low-carbon target is generated, and the dynamic optimization control command is generated based on the gap analysis feedback. The optimized configuration parameters of the low-carbon project are dynamically adjusted based on the dynamic optimization control command to realize closed-loop intelligent energy consumption management oriented towards the low-carbon target.

[0067] In this embodiment of the invention, by using low-carbon engineering optimized configuration parameters as the initialization base, the parameter assignment and working condition initialization of the low-carbon construction simulation model are completed, realizing dynamic full-domain simulation of the subsequent construction process, resource scheduling, and carbon emission evolution of steel structure engineering. A low-carbon simulation dynamic prediction model is constructed, with the following model structure: ,in To predict spatiotemporal carbon emission equivalents, For the optimal set of configuration parameters, This establishes a mapping relationship for the project's temporal evolution. The model training data includes an optimal configuration parameter dataset, a dataset of construction patterns throughout the project's timeline, and a benchmark dataset of carbon emission evolution across all stages. The algorithm logic involves generating a full-time carbon emission trajectory for the project, comparing it to preset low-carbon control target thresholds, solving for the temporal carbon emission difference across the entire domain, quantifying the differences in target achievability, and generating a low-carbon target achievability assessment result that includes periods and spatial areas of excess emissions. A closed-loop regulation and correction model is then constructed. ,in The difference. To adjust the coefficients, dynamic optimization control commands are generated in reverse based on the differences. Process parameters, energy scheduling and carbon flow control parameters are fine-tuned dimension by dimension. The simulated operating conditions are continuously iterated and corrected to form a closed-loop intelligent energy consumption management system with dynamic perception, dynamic evaluation and dynamic correction throughout the entire process.

[0068] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A smart energy consumption analysis method for steel structure engineering aimed at low-carbon goals, characterized in that, Includes the following steps: Step S1: Collect multi-dimensional data corresponding to the entire life cycle of the steel structure project, decouple the multi-dimensional data for low-carbon features, and generate a spatiotemporal feature sequence of the project under the constraint of low-carbon objectives; Acquire historical low-carbon benchmark project data and mine low-carbon process modes to generate a feature library of low-carbon process modes for steel structures. Step S2: Construct a spatiotemporal energy consumption field for the spatiotemporal feature sequence of the project to generate a spatiotemporal coupled energy consumption field distribution map; perform energy consumption hotspot and carbon flow path tracing analysis based on the spatiotemporal coupled energy consumption field distribution map to generate a set of high energy consumption spatiotemporal units and a network of key carbon flow paths; Step S3: Match the high-energy-consuming spatiotemporal unit set with the low-carbon process mode feature library of steel structure to generate a cluster of low-carbon process optimization schemes; perform dynamic carbon resistance analysis and carbon flow bottleneck identification on the key carbon flow path network to generate a set of carbon flow optimization and control strategies. Step S4: Integrate the process low-carbon optimization scheme cluster with the carbon flow optimization and control strategy set to generate low-carbon engineering optimization configuration parameters for the entire life cycle; drive the low-carbon construction simulation model based on the low-carbon engineering optimization configuration parameters to output the low-carbon target attainability assessment results and dynamic optimization control instructions; dynamically adjust the low-carbon engineering optimization configuration parameters based on the dynamic optimization control instructions to achieve closed-loop intelligent energy consumption management for low-carbon targets.

2. The intelligent energy consumption analysis method for steel structure engineering oriented towards low-carbon goals as described in claim 1, characterized in that, Step S1 includes the following steps: Step S11: Collect raw heterogeneous data from the design, material, manufacturing, transportation, construction, operation and maintenance and dismantling stages of steel structure engineering. The raw heterogeneous data shall include at least design drawing parameters, material list, processing energy consumption log, transportation trajectory, construction machinery operation data and environmental monitoring data. Step S12: Perform spatiotemporal alignment and semantic fusion on the original heterogeneous data to generate a spatiotemporally aligned engineering holographic data cube; perform feature dimension mapping on the engineering holographic data cube based on the low-carbon evaluation index system, and extract the carbon source features and carbon sink potential features corresponding to each stage; Step S13: Construct a graph of the relationship between carbon transfer between stages based on the carbon source characteristics and carbon sink potential characteristics, and analyze the topological structure of carbon element flow and transformation between stages; perform coupling correlation analysis on the carbon source characteristics and carbon sink potential characteristics based on the topological structure, and decouple independent carbon source factors, coupled carbon flow factors and potential carbon sink factors. Step S14: Integrate the independent carbon source factor, coupled carbon flow factor and potential carbon sink factor, and combine them with the spatiotemporal labels of each stage to generate the engineering spatiotemporal feature sequence under the low-carbon target constraint; Step S15: Retrieve historical low-carbon benchmark project data with excellent low-carbon performance from the historical engineering database and mine low-carbon process patterns. Construct the feature library of the steel structure low-carbon process patterns through pattern clustering and feature encoding.

3. The intelligent energy consumption analysis method for steel structure engineering oriented towards low-carbon goals as described in claim 1, characterized in that, Step S2 includes the following steps: Step S21: Perform spatiotemporal grid subdivision on the spatiotemporal feature sequence of the project to discretize the project time progress and three-dimensional spatial location into spatiotemporal voxel units; Step S22: Within each spatiotemporal voxel unit, aggregate the intensity of carbon source factor and carbon sink factor in the spatiotemporal characteristic sequence of its contained engineering elements, and calculate the unit's net carbon emission equivalent; Step S23: Based on the net carbon emission equivalent of all spatiotemporal voxel units and their spatial adjacency and temporal order, construct a spatiotemporal energy transfer weight matrix to characterize the spatiotemporal diffusion and accumulation effects of carbon emissions. Step S24: Based on the spatiotemporal energy transfer weight matrix and the net carbon emission equivalent of each unit, generate the spatiotemporal coupled energy consumption field distribution map describing the continuous distribution of carbon emission intensity in spatiotemporal space through the field strength calculation model; Step S25: Perform gradient analysis and contour line extraction on the spatiotemporal coupled energy consumption field distribution map to identify the peak carbon emission intensity region and its spatiotemporal boundary, and generate the high-energy-consuming spatiotemporal unit set; at the same time, perform streamline tracing and convergence point analysis on the carbon flow vector field in the spatiotemporal coupled energy consumption field distribution map to generate the key carbon flow path network describing the main migration paths of carbon emissions between stages and spatial units.

4. The intelligent energy consumption analysis method for steel structure engineering oriented towards low-carbon goals according to claim 3, characterized in that, Step S25, which involves gradient analysis and contour line extraction of the spatiotemporal coupled energy consumption field distribution map, includes the following steps: In the spatiotemporal coupled energy consumption field distribution map, a joint scalar field of carbon emission intensity in the spatial and temporal dimensions is defined, and the spatiotemporal joint gradient vector of each point in the joint scalar field is calculated to characterize the rate of change of carbon emission intensity in the three-dimensional spatial and temporal axis directions. Based on the magnitude and direction of the spatiotemporal joint gradient vector, spatiotemporal vector field streamline tracing is performed on the spatiotemporal coupled energy consumption field distribution map to generate potential diffusion main paths and convergence main paths of carbon emission intensity in spatiotemporal space, and regions where the streamline endpoints are scattered points are identified as initial carbon emission convergence areas. Based on the initial carbon emission accumulation zone, the spatiotemporal region through which its streamline passes is traced in reverse, and the spatiotemporal second derivative features of the joint scalar field in the spatiotemporal region are extracted, i.e., spatiotemporal curvature features. According to the sign and magnitude of the spatiotemporal curvature features, pseudo-accumulation points formed by local small fluctuations are filtered out from the initial accumulation zone, and stable carbon emission core accumulation zones with positive and significant spatiotemporal curvature are retained. Critical edge analysis is performed on the core accumulation area of ​​stable carbon emissions and in conjunction with the joint scalar field to generate spatiotemporal topological critical edges; The set of continuous spatiotemporal voxels defined by the core accumulation area of ​​stable carbon emissions and its corresponding spatiotemporal topological critical edge is defined as the set of high-energy-consuming spatiotemporal units.

5. The intelligent energy consumption analysis method for steel structure engineering oriented towards low-carbon goals according to claim 4, characterized in that, The critical edge analysis for the core accumulation zone of stable carbon emissions, combined with the joint scalar field, includes the following steps: Within the core accumulation zone of stable carbon emissions, the spatial-time geodesic path from each point on the edge of the accumulation zone to its geometric center is calculated based on the joint scalar field. Carbon emission intensity values ​​are sampled along each of the aforementioned space-time geodesic paths to construct a path carbon emission intensity decay sequence. The path carbon emission intensity decay sequence is then normalized according to the path length to generate a standard decay curve. Calculate the first derivative of each standard decay curve to obtain the decay rate curve, and further calculate the curvature of the decay rate curve on the space-time geodesic path to generate a decay acceleration sequence. The attenuation acceleration sequence is subjected to Fourier transform, and the harmonic components of the attenuation process are analyzed from the frequency domain perspective. The ratio of high-frequency harmonic energy representing nonlinear abrupt change characteristics to low-frequency harmonic energy representing smooth transition characteristics is extracted as the abrupt change characteristic parameter corresponding to the space-time geodesic path. The abrupt change characteristic parameters of each space-time geodesic path are mapped back to their origin position at the edge of the convergence zone, and cluster analysis is performed on the edge to connect the positions where the abrupt change characteristic parameters are significantly higher than the neighborhood mean, forming the spatiotemporal topological critical edge.

6. The intelligent energy consumption analysis method for steel structure engineering oriented towards low-carbon goals according to claim 3, characterized in that, Step S25, which involves streamline tracing and convergence point analysis of the carbon flow vector field in the spatiotemporal coupled energy consumption field distribution map, includes the following steps: In the spatiotemporal coupled energy consumption field distribution map, a dynamic propagation network of carbon particles in spatiotemporal space is constructed based on the carbon flow vector field, and carbon potential energy reflecting the dynamic balance between carbon emission absorption and release is introduced at the nodes of the dynamic propagation network. Based on the gradient direction of the carbon potential energy, the adaptive and optimal migration trajectory of carbon particles in the spatiotemporal network is simulated to generate multiple probabilistic migration paths from high carbon potential energy nodes to low carbon potential energy nodes. The probabilistic migration paths are subjected to spatiotemporal consistency checks. Paths with spatiotemporal overlap exceeding a preset threshold are merged, and the expected carbon flux value of each merged path is calculated. The expected carbon flux value is obtained by the integral of the product of the net carbon emission equivalent of the nodes through which the path passes and its spatiotemporal connectivity strength. Based on the expected carbon flux of each path, a carbon flux competition relationship graph between paths is constructed, and key edges and key nodes that, when removed, lead to a large number of alternative paths detouring and significantly reduce the global carbon flux efficiency are identified in this graph. From the carbon flow competition relationship graph between the paths, the key edges and key nodes are extracted and reconstructed according to their original spatiotemporal connection relationships in the dynamic propagation network to form the key carbon flow path network describing the main migration paths of carbon emissions between stages and spatial units.

7. The intelligent energy consumption analysis method for steel structure engineering oriented towards low-carbon goals according to claim 1, characterized in that, Step S3 includes the following steps: Step S31: For the set of high-energy-consuming spatiotemporal units, extract the corresponding engineering stage, process type, equipment type and material type features to form a high-energy-consuming unit feature vector; Step S32: From the feature library of low-carbon process modes of steel structures, match multiple low-carbon process modes that are aligned with the feature vector of the high-energy-consuming unit in terms of stage and process type, and parse the low-carbon technology package composition of each low-carbon process mode to generate a set of candidate low-carbon process modes. Step S33: Perform structural vulnerability analysis on the critical carbon flow path network, identify critical path nodes and edges in the network that are highly dependent on carbon flow and prone to blockage, and calculate their carbon flow resistance coefficient. Step S34: Apply the candidate low-carbon process mode set to the high-energy-consuming spatiotemporal unit set for simulation replacement, evaluate the reduction potential of each mode on the net carbon emission equivalent of the unit, and combine its influence on the carbon flow resistance coefficient of relevant nodes and edges in the key carbon flow path network to perform multi-objective collaborative optimization analysis to generate multi-objective collaborative optimization analysis results. Step S35: Based on the results of the multi-objective collaborative optimization analysis, select the optimal combination of process modes and their application strategies in terms of carbon reduction potential and improvement of system carbon flow, forming the process low-carbon optimization scheme cluster; at the same time, perform dynamic carbon resistance analysis and carbon flow bottleneck identification on the key carbon flow path network, and generate process reengineering, material substitution or energy dispatch strategies aimed at reducing carbon resistance at key nodes and edges, forming the carbon flow optimization and control strategy set.

8. The intelligent energy consumption analysis method for steel structure engineering oriented towards low-carbon goals according to claim 7, characterized in that, Step S35, which involves dynamic carbon resistance analysis and carbon flow bottleneck identification for the key carbon flow path network, includes the following steps: For each node and edge in the critical carbon flow path network, the corresponding process stage attribute, material attribute and energy type attribute are attached, and the unit carbon emission intensity parameter under the current attribute configuration is obtained according to the engineering spatiotemporal characteristic sequence, as the basic carbon resistance coefficient. Virtual carbon flow pulses are introduced into the critical carbon flow path network to simulate their propagation process from upstream nodes to downstream nodes. The delay and dissipation effects caused by property changes encountered by the virtual carbon flow on the propagation path are recorded, and the dynamic carbon resistance increment caused by property coupling interaction is quantified. By integrating the basic carbon resistance coefficient and the dynamic carbon resistance increment, the comprehensive carbon resistance value of each node and edge in the path network is calculated, and the nodes and edges are classified according to the comprehensive carbon resistance value. Nodes and edges with a comprehensive carbon resistance value that is significantly higher than the average level of the same stage / type are identified and marked as preliminary carbon flow bottlenecks. For the initial carbon flow bottleneck marked, the applicable process, material and energy alternatives in the process low-carbon optimization scheme cluster are traced back, the expected contribution of each alternative to reducing its overall carbon resistance is evaluated, and alternative optimization schemes with different priorities are generated according to the expected contribution. The network cascading effect of the proposed optimization schemes is evaluated. The optimization of a single bottleneck node is analyzed to mitigate or transfer the overall carbon resistance of adjacent nodes and edges. Schemes that can reduce the global carbon resistance of the network without triggering new high-order bottlenecks are selected. Specific process reengineering suggestions, material substitution lists, and energy scheduling sequences are formed, which together constitute the carbon flow optimization and control strategy set.

9. The intelligent energy consumption analysis method for steel structure engineering oriented towards low-carbon goals as described in claim 8, characterized in that, The carbon resistance classification of nodes and edges based on the comprehensive carbon resistance value includes the following steps: Based on the process stage attributes and material / energy type attributes of nodes and edges in the critical carbon flow path network, an attribute condition distribution space is constructed. Within the attribute condition distribution space, the mean and standard deviation of the comprehensive carbon resistance values ​​of all nodes and edges of the same stage and type are calculated as the corresponding attribute condition statistical benchmark. For each node and edge, calculate the degree of deviation of its comprehensive carbon resistance value from its corresponding attribute condition statistical benchmark, and obtain the local standardized deviation score; From the spatiotemporal feature sequence of the project, historical fluctuation features associated with each node and edge in the key carbon flow path network are extracted, including the historical variation coefficient of carbon emission intensity and the mutual influence intensity of adjacent path units, and the inherent noise level implied by the historical fluctuation features is calculated. Based on the inherent noise level, the local standardized deviation score is subjected to noise adaptive correction to filter out normal deviations caused by historical inherent fluctuations and generate a corrected abnormal deviation index. The modified abnormal deviation index of all nodes and edges in the critical carbon flow path network is statistically sorted, and the nodes and edges that are ranked in the high percentile and whose absolute deviation exceeds the average level of the neighborhood are selected and marked as the initial carbon flow bottleneck.

10. The intelligent energy consumption analysis method for steel structure engineering oriented towards low-carbon goals according to claim 1, characterized in that, Step S4 includes the following steps: A coupling effect model is established to analyze the enhancement or weakening effect of the implementation of each scheme in the process low-carbon optimization scheme cluster on the effectiveness of related strategies in the carbon flow optimization and control strategy set, and to generate a scheme strategy interaction effect matrix. Based on the interaction matrix of the aforementioned scheme strategy, with the minimization of total carbon emissions throughout the entire life cycle as the core objective and engineering cost, construction period and technical requirements as constraints, a multi-objective optimization configuration model for low-carbon engineering is constructed. The process low-carbon optimization scheme cluster and carbon flow optimization control strategy set are used as decision variables and input into the low-carbon engineering multi-objective optimization configuration model. The model is solved by intelligent optimization algorithm to obtain the Pareto optimal solution set under the constraints. The Pareto optimal solution set is a series of non-dominated low-carbon engineering configuration schemes. The Pareto optimal solution set is evaluated for engineering feasibility and screened for technology maturity. The configuration scheme with the best overall performance is selected, and its parameters are decoded to generate the low-carbon engineering optimization configuration parameters for the whole life cycle. The low-carbon construction simulation model is initialized based on the optimized configuration parameters of the low-carbon project. The model is driven to dynamically simulate the construction process, resource allocation, and carbon emissions in the subsequent stages of the project. Based on the gap between the carbon emission trajectory output by the simulation and the preset low-carbon target, the achievability assessment result of the low-carbon target is generated, and the dynamic optimization control command is generated based on the gap analysis feedback. The optimized configuration parameters of the low-carbon project are dynamically adjusted based on the dynamic optimization control command to realize closed-loop intelligent energy consumption management oriented towards the low-carbon target.

Images