Building construction evaluation method and system based on multi-source data and intelligent regulation and control

Abstract

The invention provides a building construction evaluation method and system based on multi-source data and intelligent regulation and control, and is applied to the field of data processing application. According to the method, feature extraction is performed on current situation data of a building construction stage, and a multi-source heterogeneous data set including energy, carbon emission and water resource utilization conditions is generated; processing the multi-source heterogeneous data set based on the data association relationship and the data conversion logic among the target systems, and generating a target directed multi-view dynamic graph representing the relationship between all links of building construction green monitoring and resource cycle management; performing modeling processing on the target directed multi-view dynamic graph to generate node embedding information; injecting the node embedding information into each layer of the initial evaluation model, optimizing the model by adopting a self-adaptive adjustment strategy, and generating integrated information after fusion; and processing the current situation data of the building construction stage based on the initial evaluation model in combination with the fused comprehensive information, and generating building construction evaluation result information.

Description

Technical Field

[0001] The present invention relates to the field of data processing applications, and in particular to a construction evaluation method and system based on multi-source data and intelligent control. Background Art

[0002] Traditional construction relies heavily on grid power, resulting in high and unstable energy consumption from temporary power equipment. During construction, a large number of temporary power equipment, such as tower cranes and concrete mixers, suffer from inefficient energy utilization, leading to excessive energy consumption. This not only increases construction costs but also contradicts the development philosophy of energy conservation and emission reduction. The lack of effective real-time monitoring during the construction process makes it difficult to dynamically optimize high-carbon emission processes. The inability to obtain timely and accurate carbon emission data hinders the implementation of targeted measures for high-carbon emission processes, making it difficult to effectively control carbon emissions during construction and placing significant pressure on the environment. Direct discharge of construction wastewater is common, with a recycling rate of less than 30%. Large amounts of construction wastewater are discharged without effective treatment and recycling, not only wasting water resources but also potentially polluting the surrounding environment. This falls far short of the requirements for water recycling and sustainable development. Traditional power supply methods, such as diesel generators, are widely used in construction, but these devices are noisy and polluting. Near construction sites, the noise generated by diesel generators disrupts the daily lives of nearby residents, and their exhaust emissions negatively impact air quality.

[0003] Existing technologies, such as standalone photovoltaic systems or simple sedimentation tanks, cannot achieve a closed-loop management of energy, emissions, and water resources. While standalone photovoltaic systems can provide a certain amount of clean energy, they lack energy storage and coordination with other energy systems. Simple sedimentation tanks have limited capacity to treat construction wastewater and cannot meet the demand for efficient water recycling during construction. These technological limitations pose significant challenges to green development in construction.

[0004] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore includes information that does not constitute prior art known to ordinary technicians in this field. Summary of the Invention

[0005] The purpose of this application is to provide a construction assessment method and system based on multi-source data and intelligent control, which, at least to some extent, overcomes the problems of existing technologies. This method, specifically targeting the construction phase, acquires current data and training sample sets, and generates assessment results through multi-step data processing and model optimization. It offers significant advantages in clean energy supply, precise emission reduction, efficient recycling, and cloud-based collaboration. It also provides alternative solutions for different construction scenarios, effectively promoting the green and efficient development of the construction industry.

[0006] Other features and advantages of the present application will become apparent from the following detailed description, or may be learned in part by practice of the invention.

[0007] According to one aspect of the present application, a construction assessment method based on multi-source data and intelligent control is provided, including: obtaining current data and a training sample set of the construction stage; performing feature extraction on the current data of the construction stage to generate a multi-source heterogeneous data set containing energy, carbon emissions, and water resource utilization; processing the multi-source heterogeneous data set based on the data association relationship and data conversion logic between each target system to generate a target directed multi-view dynamic graph that represents the relationship between each link of green monitoring and resource recycling management of construction; modeling the target directed multi-view dynamic graph to generate node embedding information; injecting the node embedding information into each layer of the initial assessment model, optimizing the model with an adaptive adjustment strategy, and generating comprehensive information that integrates the relationship between each link of the system and the connotation of multi-source data; processing the current data of the construction stage based on the initial assessment model combined with the integrated comprehensive information to generate construction assessment result information.

[0008] Another aspect of the present application is a construction evaluation device based on multi-source data and intelligent control, characterized in that it includes: an acquisition module for acquiring current data and a training sample set of the construction stage; a processing module for extracting features of the current data of the construction stage to generate a multi-source heterogeneous data set containing energy, carbon emissions, and water resource utilization; based on the data association relationship and data conversion logic between each target system, the multi-source heterogeneous data set is processed to generate a target directed multi-view dynamic graph that represents the relationship between each link of green monitoring and resource recycling management of construction; the target directed multi-view dynamic graph is modeled to generate node embedding information; the node embedding information is injected into each layer of the initial evaluation model, and the model is optimized using an adaptive adjustment strategy to generate comprehensive information that integrates the relationship between each link of the system and the connotation of multi-source data; the current data of the construction stage is processed based on the initial evaluation model combined with the integrated comprehensive information to generate construction evaluation result information.

[0009] According to another aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a second processor, the computer program implements the above-mentioned construction evaluation method based on multi-source data and intelligent control.

[0010] This application provides a construction assessment method and system based on multi-source data and intelligent control. The server implements green monitoring and resource recycling management assessment through a series of data processing and model optimization operations. First, the current status data and training sample set of the construction phase are obtained. After feature extraction, a multi-source heterogeneous data set is generated. Then, based on data association and conversion logic processing, a target directed multi-view dynamic graph is obtained to show the relationship between each link. The graph is then modeled to generate node embedding information, which is injected into the initial assessment model and optimized to obtain integrated comprehensive information. Finally, the current status data is processed by combining the model and comprehensive information to generate the assessment results.

[0011] In terms of clean energy supply, the photovoltaic-mobile power system significantly reduces electricity costs and noise pollution. For precise emission reduction, the carbon emission monitoring platform effectively reduces carbon emission intensity. In terms of efficient recycling, it achieves the coordinated utilization of multiple resources, reducing dependence on municipal water and electricity. Cloud-based collaboration enables closed-loop data management, improving overall efficiency. Furthermore, it provides alternative solutions, such as changing power supply methods, to expand the possibilities for flexible application of technology in different construction scenarios, helping the construction industry develop in a green and efficient direction.

[0012] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 A flowchart of a construction evaluation method based on multi-source data and intelligent control provided by an embodiment of the present application is shown;

[0014] Figure 2 A schematic structural diagram of a construction evaluation device based on multi-source data and intelligent control provided by an embodiment of the present application is shown. DETAILED DESCRIPTION

[0015] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0016] The following combination Figure 1 To describe the construction evaluation method based on multi-source data and intelligent control according to the exemplary embodiment of the present application. In one embodiment, the present application also proposes a construction evaluation method and system based on multi-source data and intelligent control. Figure 1 As shown, including:

[0017] S101, obtaining the current status data and training sample set of the building construction stage.

[0018] In one implementation, smart meters and energy consumption monitoring equipment are used to collect data on the real-time power consumption and daily power consumption of various electrical equipment on the construction site. For example, the power consumption of equipment such as tower cranes, concrete mixers, and electric welders is recorded down to the hourly level. This allows for the acquisition of current energy usage data during the construction phase, including total energy consumption, consumption distribution by time period, and the contribution of energy consumption by each type of equipment. Gas monitoring sensors installed at the exhaust outlets of construction machinery collect data on greenhouse gas emissions such as carbon dioxide and nitrogen oxides generated during the construction process. For transport vehicles, mileage, load, and exhaust emission concentration are recorded; for concrete mixing plants at the construction site, carbon emissions data from the production process is monitored. Through these methods, a comprehensive understanding of the current status of carbon emissions during the construction phase is achieved, including total carbon emissions and the contribution of different construction activities.

[0019] Smart water meters are installed on the water supply and drainage pipes at the construction site to monitor total construction water usage and water consumption in different construction stages (such as concrete curing, dust suppression, and domestic water use). Information on the generation and discharge destination of construction wastewater is also collected to understand the current status of water resource utilization during the construction phase, including water resource consumption levels, wastewater generation, and existing wastewater treatment and reuse practices. Data on energy use, carbon emissions, and water resource utilization during the construction phase of various historical construction projects is collected. These projects cover buildings of varying sizes, structural types (such as residential, commercial, and industrial plants), and construction techniques. For example, 10 completed construction projects of different types and regions were selected to obtain energy consumption records, carbon emission monitoring data, and water resource utilization reports throughout the construction process to construct a historical construction database. Benchmark green construction projects within the industry were researched to obtain data on their energy self-sufficiency rates, carbon emission reductions, and water resource recycling rates. For example, by referring to projects that have obtained green building construction certification, we can analyze their energy, emissions, and water utilization data after adopting advanced technologies and management measures, and use these data as part of the training sample set to provide optimization targets and reference standards for current projects.

[0020] Using construction simulation software, we set up different construction scenarios and parameters to simulate energy, emissions, and water utilization during construction, generating simulated data. By adjusting parameters such as construction schedules, equipment selection, and water recycling plans, we simulated construction data under different circumstances, expanded the training sample set, and enabled the model to learn more potential construction situations and data characteristics.

[0021] S102, feature extraction is performed on the current status data of the building construction phase to generate a multi-source heterogeneous data set including energy, carbon emissions, and water resource utilization.

[0022] In one implementation, historical data on the construction project is collected, such as power consumption data for equipment such as tower cranes, concrete mixers, and welders during different construction periods. Key variables such as equipment type, usage duration, and power are filtered out to generate sample data on energy usage. For example, during the foundation construction phase, a tower crane consumes 200 kWh of power, operating 8 hours per day; during the main construction phase, a concrete mixer consumes 150 kWh of power, operating 10 hours per day.

[0023] Data on mileage, load, and exhaust emission concentrations of transport vehicles during construction, as well as carbon emission data from concrete batching plants during production, are collected. Variables such as transport distance, load weight, and emission concentration are filtered out to generate sample data on carbon emission values. For example, a transport vehicle traveling 50 kilometers and carrying 8 tons during a material transport operation would have a carbon dioxide emission concentration of 3 grams per cubic meter. Data on water consumption at the construction site for concrete curing and dust suppression operations, as well as construction wastewater generation and recycling, are collected. Variables such as curing water consumption, dust suppression water consumption, wastewater generation, and wastewater recycling are extracted to generate sample data on water resource recycling. For example, during the concrete curing phase, daily water consumption is 50 cubic meters, construction wastewater generation is 30 cubic meters, and recycling is 10 cubic meters.

[0024] Energy usage sample data is aggregated and calculated over a certain period (e.g., weekly). Assuming the tower crane's power consumption over a week is 1400 kWh, 1500 kWh, 1600 kWh, 1450 kWh, 1550 kWh, 1650 kWh, and 1400 kWh, the formula: Mean = (1400 + 1500 + 1600 + 1450 + 1550 + 1650 + 1400) ÷ 7 = 1507.14 kWh yields the average energy usage for the tower crane for that week. For example, if the carbon dioxide emission concentrations for different transport vehicles over the week are 3 g / m³, 3.2 g / m³, 2.8 g / m³, 3.1 g / m³, and 2.9 g / m³, the mean is calculated as: (3 + 3.2 + 2.8 + 3.1 + 2.9) ÷ 5 = 3 g / m³, which represents the average carbon emissions for transport vehicles for that week. Regarding the amount of construction wastewater recycled, if the recycling volume for four consecutive weeks is 40 cubic meters, 50 cubic meters, 35 cubic meters, and 45 cubic meters respectively, the average value is: (40+50+35+45)÷4=42.5 cubic meters, which means the average information on water resource recycling and utilization is obtained.

[0025] Let's assume that the energy usage (taking the power consumption of a tower crane as an example, unit: kWh) data is X = [1400, 1500, 1600, 1450, 1550], and the corresponding carbon emission value (taking the carbon emission during tower crane operation as an example, unit: kilogram) data is Y = [300, 320, 350, 310, 330]. First, calculate the mean of X Mean of Y According to the covariance formula Calculation yields: Cov(X,Y) = ((1400-1500)(300-322)+(1500-1500)(320-322)+(1600-1500)(350-322)+(1450-1500)(310-322)+(1550-1500)(330-322)) / 5 = 180. This is the energy-carbon emissions covariance information.

[0026] If the energy consumption (taking the total power consumption of the construction site as an example, unit: kWh) data is A = [5000, 5500, 6000, 5300, 5700], and the water resource recycling and utilization (unit: cubic meters) data is B = [20, 25, 30, 23, 27]. Calculate the mean of A The mean of B According to the covariance formula, we can calculate: Cov(A,B)=

[0027] ((5000-5500)(20-25)+(5500-5500)(25-25)+(6000-5500)(30-25)+(5300-5500)(23-25)+(5700-5500)(27-25)) / 5=380, which is the energy-water resources covariance information.

[0028] Assume that the carbon emission values ​​(taking the carbon emission of a concrete mixing station as an example, unit: kilogram) are C = [400, 420, 450, 410, 430], and the water resource recycling volume (the amount of water used for dust reduction in the concrete mixing station, unit: cubic meter) is D = [15, 18, 20, 16, 19]. Calculate the mean of C The mean of D According to the formula, Cov(C,D)=((400-422)(15-17.6)+(420-422)(18-17.6)+(450-422)(20-17.6)+(410-422)(16-17.6)+(430-422)(19-17.6)) / 5=22.4, which is the carbon emission-water resource covariance information.

[0029] According to the standard deviation formula Taking the previous tower crane power consumption data X as an example, the calculation is: This is the standard deviation of energy usage. Taking the carbon emission concentration data Y of transportation vehicles as an example, calculate the standard deviation: This is the standard deviation of carbon emissions. For construction wastewater recovery data, such as [40, 50, 35, 45], the calculation is: That is, the standard deviation information of water resource recycling and utilization.

[0030] According to the energy-carbon emission covariance information, energy-water resource covariance information, carbon emission-water resource covariance information, energy usage standard deviation information, carbon emission value standard deviation information, water resource recycling standard deviation information, the Pearson correlation coefficient calculation formula is used. Processing is performed to generate Pearson correlation coefficient information between energy usage and carbon emission values, Pearson correlation coefficient information between energy usage and water resource recycling and utilization, and Pearson correlation coefficient information between carbon emission values ​​and water resource recycling and utilization, where Cov(x,y) is the covariance of xy class resources, S D (x) is the standard deviation of resource usage of type x, S D (y) is the standard deviation of the value of category y resources;

[0031] According to the Pearson correlation coefficient calculation formula The previously calculated energy-carbon emission covariance Cov(X,Y) = 180, the energy usage standard deviation S D (X)≈54.77, the standard deviation of carbon emissions S D Substituting (Y)≈0.14 into the formula, we can get This value indicates that there is a strong positive correlation between energy usage and carbon emissions, that is, when energy usage increases, carbon emissions also tend to increase.

[0032] The energy-water resource covariance Cov(A,B) = 380, the standard deviation of energy use (calculated based on the total electricity consumption of the construction site) S D (A) (calculated to be 500 degrees), standard deviation of water resource recycling S D Substitute (B) (calculated to be 3 cubic meters) into the formula, This shows that there is a certain degree of positive correlation between energy usage and water resource recycling and utilization, but the correlation is relatively weak.

[0033] Substitute the carbon emissions-water resources covariance Cov(C,D) = 22.4, the standard deviation of carbon emissions (calculated to be 15 kilograms), and the standard deviation of water resource recycling (calculated to be 2 cubic meters) into the formula, It shows that there is a moderate positive correlation between carbon emissions and water resource recycling.

[0034] This approach integrates sample data on energy usage, carbon emissions, and water recycling, along with their corresponding means, covariances, standard deviations, and Pearson correlation coefficients. This generates a multi-source, heterogeneous dataset encompassing a variety of data types and features, providing a comprehensive data foundation for subsequent analysis of the relationships between energy, carbon emissions, and water utilization during the construction phase.

[0035] S103, based on the data association relationship and data conversion logic between each target system, the multi-source heterogeneous data set is processed to generate a target directed multi-view dynamic graph representing the relationship between each link of green monitoring and resource recycling management of construction.

[0036] In one implementation, energy usage data, such as power consumption, operating hours, and power consumption of various types of construction equipment (tower cranes, concrete mixers, and electric welders), are extracted from multi-source heterogeneous datasets. Data with a significant impact on energy self-sufficiency is selected, such as the daily power consumption of tower cranes during different construction phases (foundation construction, main construction, etc.), to form a subset of energy usage data. For example, during the foundation construction phase, the tower crane consumes 200 kWh of electricity per day, operating for 8 hours; during the main construction phase, the crane consumes 300 kWh of electricity per day, operating for 10 hours.

[0037] During the construction process, data on mileage, loads, and exhaust emission concentrations of transport vehicles, as well as carbon emissions from equipment such as concrete mixing plants, were collected. Data closely related to emission reduction optimization was selected, such as the CO2 emission concentration of a transport vehicle with a single transport distance of 50 kilometers and a load of 8 tons, to form a carbon emissions data subset. Data on water consumption at the construction site for concrete curing, dust suppression, and domestic water use, as well as the generation, recycling, and reuse of construction wastewater, were compiled. Data representative of efficient water resource utilization, such as 50 cubic meters of daily water consumption during the concrete curing phase, 30 cubic meters of daily construction wastewater generation, and 10 cubic meters of daily recycling, were selected to form a water resource utilization data subset. Data on operational procedures, construction time, and resource consumption (energy, materials, etc.) for different construction processes (such as concrete pouring and rebar processing) was extracted. For example, data on energy consumption and construction efficiency under different pouring methods (pumping, chute, etc.) during the concrete pouring process formed a data subset related to construction processes.

[0038] Factors that directly contribute to energy self-sufficiency, such as "photovoltaic panel power generation," can be used as source nodes, as it directly increases clean energy supply and reduces reliance on the traditional power grid. Regarding carbon emissions, "using low-grade cement" can be used as a source node, as this construction process adjustment can directly reduce carbon emissions. Regarding efficient water resource utilization, "rainwater harvesting system" can be used as a source node, as it is directly related to water recycling and reuse. The affected indicators corresponding to the source nodes are used as target nodes. For the source node "photovoltaic panel power generation," "grid electricity consumption at the construction site" can be used as a target node, as increased photovoltaic panel power generation reduces grid electricity consumption. The target node for "using low-grade cement" could be "carbon emission intensity of the concrete mixing process," as the use of low-grade cement affects this intensity. The target node for "rainwater harvesting system" could be "municipal water consumption at the construction site," as increased rainwater harvesting reduces municipal water consumption. Each data subset is processed, and the relevant data for the source and target nodes is organized into node information. For example, the "photovoltaic panel power generation" node information includes the photovoltaic panel power (5kW), daily power generation (10 degrees), etc.; the "construction site power grid electricity consumption" node information includes daily power consumption (2000 degrees in the foundation construction phase, 3000 degrees in the main construction phase), etc.

[0039] There's a close connection between energy use and carbon emissions. For example, when a tower crane's electricity consumption increases, its carbon emissions also increase accordingly. By analyzing subsets of energy use and carbon emissions data, we can determine that for every 100 kWh of electricity consumed by a tower crane, carbon emissions increase by 50 kg. This relationship can be used as edge information to indicate the impact of energy use on carbon emissions. Certain construction equipment consumes both energy and water during operation. For example, a concrete mixer requires water for cooling during the mixing process, which also consumes electricity. Assuming that for every 10 kWh of electricity consumed by a concrete mixer, water usage increases by 5 cubic meters, this data relationship constitutes edge information between energy use and water resource utilization.

[0040] Some construction activities may affect both carbon emissions and water use. For example, adopting a water-saving curing process during concrete curing not only reduces water consumption but also reduces energy consumption by shortening curing time, thereby reducing carbon emissions. Assuming that the new curing process reduces water consumption by 20% and carbon emissions by 10%, this change relationship can be used as side information. Different construction techniques have an impact on energy, carbon emissions, and water use. For example, pump pouring uses more energy than chute pouring, but also offers higher construction efficiency. If pump pouring consumes 20% more energy than chute pouring but also shortens construction time by 10% (construction time is also correlated with carbon emissions and water use), these data relationships constitute side information between the construction technique and other factors.

[0041] Integrate the node and edge information generated above to construct an initial directed multi-view dynamic graph. Layout source nodes such as "Photovoltaic Panel Power Generation," "Use of Low-Grade Cement," and "Rainwater Recycling System," along with target nodes such as "Construction Site Grid Power Consumption," "Carbon Emission Intensity of the Concrete Mixing Process," and "Construction Site Municipal Water Consumption," according to their interrelationships. Directed edges connect source and target nodes. The direction of the edge indicates the direction of influence. For example, a line from "Photovoltaic Panel Power Generation" to "Construction Site Grid Power Consumption" indicates the impact of photovoltaic panel power generation on construction site grid power consumption. Edge weights can be initially set based on the degree of influence from the edge information. For example, the edge weight between "Photovoltaic Panel Power Generation" and "Construction Site Grid Power Consumption" can be set based on the actual impact of photovoltaic panel power generation on grid power consumption. Assuming that every kilowatt-hour of electricity generated reduces grid power consumption by 0.8 kWh, the edge weight can be set to 0.8.

[0042] Adjust node positions based on the closeness of the connections between them. Closely connected nodes (such as "Photovoltaic Panel Power Generation" and "Construction Site Grid Power Consumption") are placed closer together in the graph, making the graph structure more clear and intuitive in reflecting the relationships between factors. Further data analysis optimizes edge weights. For example, the impact of photovoltaic panel power generation on grid power consumption may change with the construction phase. In the later stages of construction, as more construction equipment is added, the proportion of photovoltaic panel power generation in total power consumption may decrease, and its impact on grid power consumption will also change. In this case, the weights of the edges between "Photovoltaic Panel Power Generation" and "Construction Site Grid Power Consumption" should be adjusted accordingly.

[0043] Check whether there are redundant edges in the graph, that is, edges that have no substantial effect on the overall relationship expression. For example, if it is found that a certain construction process has a very small impact on energy, carbon emissions, and water resource utilization in the current project, and the edges between it and other nodes are negligible in the overall relationship, this edge can be removed. For abnormal nodes, such as nodes with obvious errors in certain data or nodes that are seriously inconsistent with the overall data trend (possibly caused by data collection errors), they are removed from the graph. After these processes, the target directed multi-view dynamic graph is finally generated, which accurately reflects the degree of influence and mutual relationship of various construction factors on energy self-sufficiency, emission reduction optimization, and efficient use of water resources.

[0044] S104: Modeling the target directed multi-view dynamic graph to generate node embedding information.

[0045] In one embodiment, the target directed multi-view dynamic graph is divided into three layers. The bottom layer is the basic data layer, which includes basic nodes such as various types of construction equipment (such as tower cranes, concrete mixers, electric welders, etc.), energy supply sources (power grids, photovoltaic panels), water resource-related facilities (rainwater collection devices, wastewater treatment equipment), and construction process links (concrete pouring, steel bar processing). The middle layer is the process layer, which covers process nodes such as energy consumption and conversion, carbon emission generation and control, and water resource utilization and circulation. The top layer is the target layer, which has overall target nodes such as energy self-sufficiency rate, carbon emission reduction, and water resource recycling rate. This layering clarifies the logical relationship of data from basic generation to the impact on the overall goal, forming hierarchical information.

[0046] The tower crane's energy consumption node is connected to the grid power supply and photovoltaic panel power generation nodes, indicating the crane's energy sources. The tower crane's carbon emission node is connected to the concrete pouring process's carbon emission control node, reflecting the impact of construction technology on carbon emissions. The rainwater collection device node is connected to the concrete curing and dust suppression water use nodes, reflecting the flow of water resources. By recording the direction of these connections (e.g., from the energy supply node to the equipment energy consumption node, indicating the direction of energy supply) and the connection strength (e.g., determining the connection strength based on the proportion of the tower crane's energy consumption to total energy consumption), node connection relationship information is generated.

[0047] The construction progress is divided into phases, such as foundation construction, main construction, and decoration and renovation. During the foundation construction phase, the tower crane's weekly energy consumption increased from 1,000 kWh to 1,200 kWh. During the main construction phase, the concrete mixing plant's carbon emissions gradually decreased from 45 kilograms of carbon dioxide per cubic meter of concrete per week to 40 kilograms. Water recovery at the construction site increased from 15 cubic meters to 30 cubic meters per day during the rainy season (during the main construction phase). These time-varying data were organized into time series to form time series feature information.

[0048] Hierarchical structure information, node connectivity information, and time series feature information are integrated to form a complete target graph structure. This information comprehensively demonstrates the interrelationships between various construction factors at different levels and times, providing a foundational framework for subsequent analysis. Analysis of the data within the target graph structure revealed significant fluctuations in energy consumption across different construction phases, with energy demand during the main construction phase significantly higher than during the foundation construction phase. Carbon emissions were higher in the concrete pouring and transportation phases, while water utilization was higher in the concrete curing and dust reduction phases, and was also affected by weather (e.g., increased rainwater recovery during the rainy season).

[0049] Taking the neural network model as an example, for node connection weights related to energy consumption fluctuations, larger initial values ​​are used to highlight the impact of energy consumption changes on the model. Node A (crane energy consumption) is connected to node B (total energy supply). Based on the degree of fluctuation in energy consumption data, the connection weight is initialized to 0.7. Bias parameters are initialized based on the node's average data value. For example, if the average recovery volume of a water resource recovery node is 20 cubic meters, the bias parameter of this node is initialized to 15, ensuring that the model has a certain basis for responding to water resource recovery data in the initial stage.

[0050] The initialized weights and bias parameters are organized into model initialization information, and the weights of each node connection and the bias values ​​of each node are recorded to determine the initial state of the model when it starts processing data. Based on the weights and biases in the model initialization information, the data in the target graph structure information is calculated. Taking energy supply as an example, for grid power supply nodes and photovoltaic power generation nodes, the connection weights between them and various electrical equipment (such as tower cranes and concrete mixers) are combined to calculate the proportion of energy obtained by each device from different energy supply sources. Assuming that the weight of the tower crane obtaining energy from the grid is 0.6 and the weight of obtaining energy from photovoltaic power generation is 0.4, when the tower crane consumes 100 kWh of electricity, it is calculated that 60 kWh is obtained from the grid and 40 kWh is obtained from photovoltaic power generation.

[0051] For energy supply, information such as the energy distribution ratio and supply stability between various energy supply sources and equipment is collated to form energy supply characteristics. For carbon emission control, the carbon emission intensity of different construction processes (such as concrete mixing and transportation processes) and the correlation between carbon emission monitoring data and process adjustments are calculated as carbon emission control characteristics. For water resource recycling, information such as the matching degree between rainwater recovery and wastewater treatment volume and each water use link (such as concrete curing and dust reduction) is analyzed to form water resource recycling characteristics. These characteristics are integrated to generate initial node feature information that includes energy supply, carbon emission control, and water resource recycling characteristics.

[0052] From the initial node feature information, we extract information that changes over time, such as the change in energy supply proportions across different construction phases, the time series variation in carbon emission intensity, and the time trend in water resource recycling efficiency. For example, in the early stages of construction, photovoltaic power generation accounted for 15% of the total energy supply. As photovoltaic panels were added to the construction site, this proportion increased to 25% in the later stages of construction. Furthermore, after adopting a new mixing process, the carbon emission intensity of the concrete mixing process decreased from 45 kg of CO2 per cubic meter of concrete to 40 kg, and this change was sustained across different construction phases.

[0053] Time series analysis methods, such as moving averages or autoregressive models, are used to fuse these time series features. Taking the moving average method as an example, assume that the PV power generation share over the past three construction phases is calculated using a moving average to smooth data fluctuations. If the PV power generation share over the past three phases was 15%, 20%, and 25%, respectively, the fused value for the current phase is calculated using the three-term moving average method as (15% + 20% + 25%) / 3 = 20%. This method fuses the time series features of energy supply, carbon emission control, and water resource recycling to generate fused time series feature information that reflects the comprehensive changes in each factor over time. The synergistic relationships between these factors are analyzed from multiple perspectives, including energy, carbon emissions, water resources, and construction technology. In the energy perspective, the impact of changes in the energy supply structure (such as an increase in the proportion of PV power generation) on construction costs and energy stability is examined. In the carbon emissions perspective, the environmental impact of reduced carbon emission intensity (such as the adoption of new construction technologies) is studied. In the water resources perspective, the impact of improved water resource recycling efficiency (such as increased rainwater recovery) on construction water costs and sustainability is explored. At the same time, the interrelationships between these views are analyzed, such as changes in energy supply structure may affect carbon emissions, and water resource recycling may be related to improvements in construction technology.

[0054] Through multi-view collaborative correlation analysis, the combined impact of various factors on construction is represented in vector form, generating node embedding information. For example, for a tower crane node, its embedded information may include the degree of cost reduction in energy supply (e.g., an increase in the proportion of photovoltaic power generation reduces weekly energy costs by 80 yuan), the effect of carbon emission reduction (e.g., the use of new energy-saving equipment reduces carbon emissions by 40 kilograms per week), and the indirect impact on water resource utilization (e.g., cooling water consumption is reduced by 8 cubic meters due to energy-saving equipment). This is represented by a multidimensional vector [80, -40, -8], which comprehensively reflects the comprehensive impact of the tower crane on construction and provides key information support for subsequent construction evaluation and decision-making.

[0055] S105, injecting the node embedding information into each layer of the initial evaluation model, optimizing the model with an adaptive adjustment strategy, and generating comprehensive information that integrates the relationships between various links of the system and the connotations of multi-source data.

[0056] In one embodiment, it is assumed that the generated node embedding information is a high-dimensional vector, such as a 10-dimensional vector containing multiple influencing factors such as energy supply, carbon emissions, water resource utilization, and construction technology, but the initial evaluation model requires an input vector dimension of 8 dimensions. At this time, dimensionality adaptation is performed through a dimensionality reduction algorithm, such as principal component analysis (PCA). For example, after PCA analysis, it is determined that the first 8 principal components can explain more than 95% of the information of the original 10-dimensional vector, thereby converting the 10-dimensional node embedding information into 8 dimensions so that its dimension matches the initial evaluation model. If the node embedding information was originally stored in the form of a floating-point array, and the initial evaluation model requires the input format to be a tensor, use the corresponding programming tool (such as PyTorch or TensorFlow library in Python) to convert the floating-point array into a tensor format. Assuming that the node embedding information is [0.2, 0.5, 0.3, 0.8, 0.6, 0.4, 0.7, 0.1], PyTorch can convert it into a tensor through the torch.tensor() function, that is, adapted_embedding = torch.tensor([0.2, 0.5, 0.3, 0.8, 0.6, 0.4, 0.7, 0.1]), which completes the format conversion and obtains the adapted node embedding information.

[0057] Load the pretrained weights into the initial evaluation model using the model loading function (such as torch.load() in PyTorch). For example, model = torch.load('pretrained_model.pth'). Then, freeze some of the pretrained weights, such as the weights of the first few layers of the model, so that they are not updated during subsequent training. In PyTorch, this is achieved by setting requires_grad = False, such as in forparaminmodel.parameters()[:3]:param.requires_grad = False, where the weights of the first three layers are frozen. Initialize key parameters of the adaptive adjustment strategy: A common adaptive adjustment strategy is the learning rate adjustment strategy, such as the learning rate in the Adam optimizer. Assuming the Adam optimizer is used, initialize the learning rate lr = 0.001, and the β1 and β2 parameters to 0.9 and 0.999, respectively. Record the model state after loading the pretrained weights, freezing some weights, and initializing the key parameters of the adaptive adjustment strategy to generate the model's initial state information, including the model structure, weight values, and optimizer parameters.

[0058] The initial evaluation model is a simple multilayer perceptron (MLP) consisting of an input layer, two hidden layers, and an output layer. The input layer receives the adapted node embeddings. In PyTorch, the adapted node embeddings are passed as input to the input layer, for example, input_layer = nn.Linear(8,16) (where the input layer maps 8-dimensional vectors to 16 dimensions). The node embeddings are fed into the model's input layer via input = input_layer(adapted_embedding) . The output of the input layer serves as the input to the next hidden layer. For the first hidden layer, hidden_layer1 = nn.Linear(16,32), the output of the input layer is passed to the first hidden layer via hidden1 = hidden_layer1(input) . In the same way, the output of the first hidden layer is passed to subsequent layers until the information transmission of all layers is completed. Finally, the propagation result of the model input information with node embedding in the entire model is obtained. This result includes the feature representation after processing by each layer, providing a data basis for subsequent model training and evaluation.

[0059] Forward propagation is performed to obtain the model's prediction results. Assuming that the model's prediction results are predictions, and the actual construction green management benefit data labels are labels, a loss function (such as the mean square error loss function MSE) is used to calculate the loss between the prediction results and the actual labels. Based on the calculated loss, an adaptive adjustment strategy (such as the Adam optimizer) is used for backpropagation to update the model parameters. In PyTorch, the previous gradient is cleared through optimizer.zero_grad(), and then the gradient is calculated through loss.backward(), and finally the model parameters are updated through optimizer.step(). Assume that after one backpropagation and parameter update, a weight parameter weight of the model changes from the initial value w0 to w1. All these updated parameters are sorted and recorded to generate adjusted model parameter information, which reflects the state changes of the model after training based on the current input data.

[0060] The adjusted model parameters incorporate learning results from data related to the green management benefits of construction. Graph data reflecting the relationships between various aspects of green monitoring and resource recycling management in construction (e.g., graph data containing nodes and edges related to energy consumption, carbon emissions, and water resource utilization) is input into the adjusted model. The model processes this graph data based on the adjusted parameters, mining the semantic relationships between various aspects of the graph data through operations and feature extraction at each layer. For example, the model discovered a potential connection between increased photovoltaic power generation and reduced carbon emissions in energy supply, as well as a synergistic relationship between improved water recycling efficiency and reduced energy consumption.

[0061] After model processing, the output is comprehensive information that integrates the relationships between various system links and the connotations of multi-source data. This comprehensive information can be a vector that contains multiple evaluation indicators for the green management benefits of construction projects, such as the potential for increasing energy self-sufficiency, the degree of carbon emission reduction, and the optimization effect of water resource recycling. Assuming the output vector is [0.7, -0.5, 0.8], it represents a 70% potential increase in energy self-sufficiency, a 50% reduction in carbon emissions (the negative sign indicates a reduction), and an 80% optimization effect on water resource recycling. This provides a comprehensive basis for subsequent green management assessments and decision-making for construction projects.

[0062] S106 , processing the current status data of the building construction stage based on the initial assessment model and the integrated comprehensive information to generate building construction assessment result information.

[0063] In one implementation, feature extraction is performed on the fused comprehensive information to generate fused data feature information. This fused comprehensive information is a complex dataset containing information on multiple aspects, including energy supply, carbon emissions, and water resource utilization. Energy supply includes the proportion and total amount of different energy sources (such as photovoltaics and the power grid), carbon emissions include the carbon emission intensity and total amount of different construction activities, and water resource utilization includes data on water consumption and wastewater recovery. Feature extraction can be performed using a combination of principal component analysis (PCA) and feature selection algorithms.

[0064] Use PCA to process the energy supply data in the comprehensive information and identify the main components that affect the energy supply situation. For example, after analysis, it was found that the two components of photovoltaic power supply ratio and total energy supply can explain more than 90% of the changes in energy supply data. These two components are regarded as key features of energy supply. For carbon emission data, through feature selection algorithms (such as correlation-based feature selection), the carbon emission intensity of the concrete mixing and transportation process is determined to be the feature with the greatest impact on overall carbon emissions. For water resource utilization data, the two key features of total water consumption at the construction site and wastewater recycling rate are extracted. These key features extracted from different aspects are combined to generate fused data feature information. These features can more concisely represent the fused comprehensive information and highlight the factors that have a greater impact on the green management benefits of construction.

[0065] The energy usage, carbon emission, and water resource utilization data sequences from the current construction phase are parsed and processed to generate target event feature information. The energy usage data sequence records the changes in electricity consumption over time for various equipment during the construction process, such as the hourly power consumption of tower cranes, concrete mixers, and other equipment. This data is parsed to calculate the average power consumption, peak power consumption, and power consumption trends for different equipment during different construction phases. For example, during the foundation construction phase, the tower crane's average hourly power consumption is 25 kWh, rising to 35 kWh during the main construction phase, with power consumption showing a gradual upward trend. These data features can reflect the energy consumption patterns of the equipment and serve as target event feature information for energy usage.

[0066] The carbon emission data series includes data on exhaust emissions from transport vehicles and concrete mixing plant production, among other processes. This data is analyzed to determine the contribution of different construction activities to carbon emissions and how carbon emissions change over time. For example, carbon emissions from transport vehicles during material transportation account for 30% of total construction carbon emissions. As construction progresses, total carbon emissions gradually increase due to increased transport distances and the number of trips. This information constitutes characteristic information about target carbon emission events.

[0067] The water resource utilization data series covers water consumption, wastewater generation, and recycling at construction sites. By analyzing this data, we can determine the water consumption share of different construction stages (such as concrete curing and dust suppression) and the ratio of wastewater recycling to water consumption. For example, concrete curing accounts for 40% of total construction water consumption, while the ratio of wastewater recycling to water consumption during the current construction phase is 20%. These data characteristics reflect water resource utilization and serve as characteristic information for target events related to water resource utilization.

[0068] Based on the fused data feature information, an association analysis is performed on the target event feature information to generate association analysis results. An association rule mining algorithm (such as the Apriori algorithm) is used to analyze the relationship between the fused data feature information and the target event feature information. Taking energy supply and energy use as an example, a support threshold of 0.3 and a confidence threshold of 0.6 are set. Algorithmic analysis reveals that when the photovoltaic power supply ratio exceeds 30% (fused data feature information), there is a 65% probability that the average power consumption of the tower crane during the foundation construction phase (target event feature information) will decrease by 10%-20%. This association rule meets the set threshold conditions. Similarly, for carbon emissions and water resource utilization, it was found that when the wastewater recycling rate increases to 40% (fused data feature information), there is a 70% probability that the carbon emission intensity of the concrete mixing process (target event feature information) will decrease by approximately 15%. These satisfying associations are organized into association analysis results, which demonstrate the potential connections between different factors and provide a basis for subsequent evaluation.

[0069] The association analysis results are input into the evaluation module of the initial evaluation model, where they are processed using the model's evaluation capabilities to generate preliminary evaluation information. The initial evaluation model is a machine learning-based scoring model that receives the association analysis results as input. The previously obtained association data, such as the relationship between the photovoltaic power supply ratio and tower crane power consumption, and the relationship between wastewater recycling rate and carbon emission intensity of concrete mixing, are formatted and input into the model according to the model's required format. The model performs calculations based on the input association analysis results and its internal evaluation logic (e.g., based on pre-trained weights and algorithms). For example, the model evaluates and scores energy self-sufficiency, carbon emission reduction effectiveness, and water resource utilization efficiency. If the model sets the maximum score for energy self-sufficiency to 100, based on the input associations, the current project's energy self-sufficiency performance would be 70 points; carbon emission reduction effectiveness would be 80 points, with the current project receiving 60 points; and water resource utilization efficiency would be 90 points, with the current project receiving 75 points. These scores are combined to generate preliminary evaluation information, providing a preliminary assessment of the construction project's performance in terms of green management benefits.

[0070] The preliminary construction green management benefit assessment information is verified and adjusted to generate the construction assessment results. The preliminary assessment information is compared with industry standards or actual data from similar projects to verify the rationality of the assessment results. For example, the average score for energy self-sufficiency for similar projects in the industry is 75, while the preliminary assessment of this project is 70, which is within a reasonable range. However, if certain assessment results deviate significantly from the actual situation, such as the water resource utilization efficiency assessment score for this project being significantly higher than the actual observed water resource waste, further review of the data and assessment process is required.

[0071] Based on the verification results, if errors are found in certain data or factors are not fully considered during the evaluation process, the evaluation results will be adjusted. For example, when calculating the wastewater recycling rate, it was found that the number of reuses of some recycled water was incorrectly calculated. After recalculation, the wastewater recycling rate was reduced, and the water resource utilization efficiency score was adjusted accordingly. After adjustment, the final energy self-sufficiency score was determined to be 65 points, the carbon emission reduction effect score was 62 points, and the water resource utilization efficiency score was 70 points. These adjusted scores were combined to generate the construction evaluation results information, providing project managers with more accurate and reliable construction green management benefit evaluation conclusions, so that they can make targeted decisions and improvement measures.

[0072] This application obtains the current status data and training sample set of the construction phase, extracts features from the current status data, and generates a multi-source heterogeneous data set containing energy, carbon emissions, and water resource utilization. Then, based on the data association and conversion logic between the systems, the multi-source heterogeneous data set is processed to generate a target directed multi-view dynamic graph to characterize the relationship between each link of green monitoring and resource cycle management of construction. Then, the target graph is modeled and processed to generate node embedding information, which is injected into the initial evaluation model and optimized using an adaptive adjustment strategy to obtain comprehensive information that integrates the relationship between each link of the system and the connotation of multi-source data. Finally, the initial evaluation model and comprehensive information are combined to process the current status data and generate construction evaluation result information.

[0073] In terms of clean energy supply, the photovoltaic-mobile power system reduces electricity costs by 40% and noise pollution by 90%. For targeted emission reduction, the carbon emission monitoring platform optimizes processes in real time, reducing carbon emission intensity by 12-18%. In terms of efficient recycling, rainwater collection, photovoltaic power generation, water resources, and mobile energy storage are implemented to reduce municipal water and electricity dependence. Cloud collaboration enables closed-loop management of energy, emissions, and water resource data, increasing overall efficiency by 30%. Furthermore, alternative solutions, such as switching from photovoltaic power generation to diesel generators, are proposed, providing more options for technology application.

[0074] In one embodiment, Figure 2 As shown, the present application also provides a construction evaluation device based on multi-source data and intelligent control, comprising:

[0075] Acquisition module 201, used to obtain the current status data of the building construction stage and the training sample set;

[0076] The processing module 202 is used to extract features from the current data of the construction stage and generate a multi-source heterogeneous data set containing energy, carbon emissions, and water resource utilization; based on the data association relationship and data conversion logic between the target systems, the multi-source heterogeneous data set is processed to generate a target directed multi-view dynamic graph that represents the relationship between each link of green monitoring and resource circulation management of construction; the target directed multi-view dynamic graph is modeled to generate node embedding information; the node embedding information is injected into each layer of the initial evaluation model, and the model is optimized using an adaptive adjustment strategy to generate comprehensive information that integrates the relationship between each link of the system and the connotation of multi-source data; based on the initial evaluation model and the integrated comprehensive information, the current data of the construction stage is processed to generate construction evaluation result information.

[0077] Each embodiment in this application is described in a related manner, and similar parts between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences from other embodiments. In particular, with respect to the construction evaluation method based on multi-source data and intelligent control, the electronic device, the electronic device, and the readable storage medium embodiment are substantially similar to the above-described construction evaluation method based on multi-source data and intelligent control, so the description is relatively simple. For related parts, refer to the partial description of the above-described construction evaluation method based on multi-source data and intelligent control embodiment.

Claims

1. A construction evaluation method based on multi-source data and intelligent control, characterized in that: include: Obtain current data and training sample sets during the construction phase; Extract features from the current status data of the building construction phase to generate a multi-source heterogeneous dataset containing energy, carbon emissions, and water resource utilization; Based on the data association relationship and data conversion logic between target systems, multi-source heterogeneous data sets are processed to generate a target-directed multi-view dynamic graph that represents the relationship between each link of green monitoring and resource recycling management in construction. Model the target directed multi-view dynamic graph and generate node embedding information; Inject node embedding information into each layer of the initial evaluation model, optimize the model using an adaptive adjustment strategy, and generate comprehensive information that integrates the relationships between various links in the system and the connotations of multi-source data; Based on the initial assessment model and the integrated comprehensive information, the current status data of the construction stage is processed to generate construction assessment result information.

2. The method according to claim 1, wherein Feature extraction is performed on the current status data of the building construction phase to generate a multi-source heterogeneous dataset containing energy, carbon emissions, and water resource utilization, including: Extract and filter the variable data of energy usage, carbon emissions, and water resource recycling in the historical data of the construction phase to generate sample data information for each variable; Process the sample data of each variable, such as energy usage, carbon emissions, and water resource recycling, to generate average information on energy usage, average information on carbon emissions, and average information on water resource recycling; Process the sample data information and corresponding mean information of each variable of energy usage, carbon emission values, and water resource recycling and utilization to generate energy-carbon emission covariance information, energy-water resource covariance information, carbon emission-water resource covariance information, energy usage standard deviation information, carbon emission value standard deviation information, and water resource recycling and utilization standard deviation information; According to the energy-carbon emission covariance information, energy-water resource covariance information, carbon emission-water resource covariance information, energy usage standard deviation information, carbon emission value standard deviation information, water resource recycling standard deviation information, the Pearson correlation coefficient calculation formula is used. Processing is performed to generate Pearson correlation coefficient information between energy usage and carbon emission values, Pearson correlation coefficient information between energy usage and water resource recycling and utilization, and Pearson correlation coefficient information between carbon emission values ​​and water resource recycling and utilization, where Cov(x,y) is the covariance of xy class resources, S D (x) is the standard deviation of resource usage of type x, S D (y) is the standard deviation of the value of category y resources; The Pearson correlation coefficient information between energy usage and carbon emissions, between energy usage and water resource recycling and utilization, and between carbon emissions and water resource recycling and utilization is processed to generate a multi-source heterogeneous data set.

3. The method according to claim 1, wherein Based on the data association and data conversion logic between target systems, the multi-source heterogeneous data sets are processed to generate a target-directed multi-view dynamic graph representing the relationships between each link of green monitoring and resource recycling management in construction, including: Classify and screen the energy usage data, carbon emission data, water resource utilization data, and construction process-related data in multi-source heterogeneous data sets to generate energy usage data subsets, carbon emission data subsets, water resource utilization data subsets, and construction process data subsets; The factors that affect construction are considered as nodes, the factors that directly promote energy self-sufficiency, emission reduction optimization, and efficient use of water resources are considered as source nodes, and the related indicators affected by them are considered as target nodes. Each data subset is processed to generate node information. Process each data subset based on the relationship affecting building construction to generate edge information; Generate an initial directed multi-view dynamic graph based on the generated node information and edge information; The initial directed multi-view dynamic graph is integrated and optimized. By using the graph optimization algorithm, the node positions and edge weights are adjusted to make the graph structure reflect the influence of various factors on construction and their mutual relationships. At the same time, redundant edges and abnormal nodes are removed to generate the target directed multi-view dynamic graph.

4. The method according to claim 1, wherein Model the target directed multi-view dynamic graph and generate node embedding information, including: Perform layered decomposition on the target directed multi-view dynamic graph to generate graph hierarchical structure information, node connection relationship information, and time series feature information, thereby generating target graph structure information. Based on the target graph structure information and data characteristics, the adaptive parameter initialization algorithm is used to initialize the model parameters and generate model initialization information; The target graph structure information is processed based on the model initialization information to generate initial node feature information including energy supply, carbon emission control, and water resource recycling features; Process the initial node feature information to generate time series fusion feature information; A multi-view collaborative correlation analysis mechanism is used to process the time series fusion feature information to generate node embedding information that reflects the impact of various factors on building construction.

5. The method according to claim 4, wherein Inject node embedding information into each layer of the initial evaluation model, and use adaptive adjustment strategies to optimize the model to generate comprehensive information that integrates the relationships between various links in the system and the connotations of multi-source data, including: Perform dimension adaptation and format conversion on the node embedding information to generate adapted node embedding information; Load and freeze pre-trained weights for the initial evaluation model, initialize key parameters in the adaptive adjustment strategy, and generate the model's initial state information; Based on the initial state information of the model, the adapted node embedding information is injected layer by layer according to the layer structure of the initial evaluation model to generate model input information with node embedding; Adopting an adaptive adjustment strategy, the model input information with node embedding is trained and adjusted to generate adjusted model parameter information; Based on the adjusted model parameter information, the input graph data reflecting the relationship between each link of green monitoring and resource recycling management in construction is processed by joint semantic modeling to generate comprehensive information that integrates the relationship between each link of the system and the connotation of multi-source data.

6. The method according to claim 1, wherein Based on the initial assessment model and the integrated comprehensive information, the current status data of the construction stage is processed to generate construction assessment result information, including: Perform feature extraction on the fused comprehensive information to generate fused data feature information; Analyze and process the energy usage data series, carbon emission data series, and water resource utilization data series in the current status data of the building construction phase to generate target event feature information; Perform correlation analysis on target event feature information based on fused data feature information to generate correlation analysis result information; Input the association analysis result information into the evaluation module of the initial evaluation model, process it using the evaluation capability of the model, and generate preliminary evaluation information; The preliminary construction green management benefit evaluation information is verified and adjusted to generate construction evaluation result information.

7. A construction evaluation device based on multi-source data and intelligent control, characterized in that: The device comprises: The acquisition module is used to obtain the current status data and training sample sets of the building construction stage; The processing module is used to extract features from the current data of the construction phase and generate a multi-source heterogeneous data set containing energy, carbon emissions, and water resource utilization; based on the data association relationship and data conversion logic between the target systems, the multi-source heterogeneous data set is processed to generate a target directed multi-view dynamic graph that represents the relationship between each link of green monitoring and resource circulation management of construction; the target directed multi-view dynamic graph is modeled and processed to generate node embedding information; the node embedding information is injected into each layer of the initial evaluation model, and the model is optimized using an adaptive adjustment strategy to generate comprehensive information that integrates the relationship between each link of the system and the connotation of multi-source data; based on the initial evaluation model and the integrated comprehensive information, the current data of the construction phase is processed to generate construction evaluation result information.

8. An electronic device, characterized in that: include: a first processor; and a memory for storing executable instructions of the first processor; The first processor is configured to execute the construction evaluation method based on multi-source data and intelligent control according to any one of claims 1 to 6 by executing the executable instructions.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the second processor, the construction evaluation method based on multi-source data and intelligent control as described in any one of claims 1 to 6 is implemented.

Images