Construction site energy consumption intelligent monitoring method and system based on BIM and edge calculation

Abstract

The invention discloses a construction site energy consumption intelligent monitoring method and system based on BIM and edge calculation, and relates to the field of building information modeling and Internet of Things fusion application. The method comprises the following steps: acquiring BIM model data and field energy consumption time sequence data through an acquisition terminal with edge computing capability, and generating an initial energy consumption data set; forming a feature data set through multi-scale time and space division; and generating an energy consumption trend prediction result based on space-time modeling, identifying a high energy consumption point, generating an optimization suggestion in combination with historical data, and pushing the optimization suggestion to a management terminal. According to the invention, space-time fusion of BIM structure information and energy consumption data is realized, and monitoring precision and positioning are improved; identifying regional energy consumption difference by a multi-scale feature extraction enhancement model; the energy consumption trend prediction capability is improved through LSTM modeling; the recognition accuracy of the high-energy consumption point is improved through threshold value and continuity analysis; energy-saving suggestions are generated based on the strategy template, a closed-loop process from prediction to optimization is constructed, and the energy efficiency management level of the construction site is improved.

Description

Technical Field

[0001] This invention relates to the field of integrated application of building information modeling and Internet of Things (IoT) technologies, specifically to a method and system for intelligent monitoring of energy consumption at construction sites based on BIM and edge computing. Background Technology

[0002] With the deepening development of smart construction site construction and green building concepts, energy consumption management at construction sites is gradually shifting from traditional manual monitoring to data-driven intelligent monitoring. Building Information Modeling (BIM), as a crucial support for building a fully digital construction process, can provide detailed structural, spatial, and functional data, while the introduction of edge computing provides strong computational support for on-site perception and real-time processing. Currently, more and more research and engineering practices are beginning to explore the integration of BIM and IoT technologies to achieve data integration and energy consumption optimization during the construction phase, providing a technical path for building energy conservation and carbon emission control.

[0003] However, existing construction site energy consumption monitoring systems still face numerous technical bottlenecks. Firstly, in terms of data acquisition, BIM model data and energy consumption sensor data are often independent, lacking a unified spatiotemporal alignment mechanism, resulting in low data fusion efficiency and severe information silos. Secondly, in terms of feature construction, most existing systems employ single-scale or coarse-grained statistical methods, making it difficult to capture the coupling patterns of energy consumption across different time levels and spatial regions, thus failing to provide high-quality input for prediction models. Furthermore, energy consumption trend predictions largely rely on shallow models or static rules, lacking the ability to deeply model dynamic trends, resulting in insufficient prediction accuracy and timeliness. Especially in the identification of high-energy-consumption points and the generation of control suggestions, existing solutions generally rely on manual analysis or simple rule matching, failing to achieve attribution reasoning and strategy generation based on multi-dimensional features. Therefore, current technology struggles to meet the comprehensive needs of complex construction sites in terms of "spatiotemporal accuracy, prediction capability, and intelligent feedback."

[0004] To overcome the above problems, this invention proposes an intelligent energy consumption monitoring method and system that integrates BIM and edge computing capabilities, possesses multi-scale spatiotemporal feature construction, in-depth energy consumption trend prediction, and high energy consumption attribution analysis. It can achieve a closed-loop response from on-site data collection to strategy suggestion push, improve the energy efficiency management level during construction, and has significant engineering practicality and technological advancement. Summary of the Invention

[0005] To achieve the aforementioned objectives and address the aforementioned technical problems, this invention provides a method for intelligent monitoring of energy consumption at construction sites based on BIM and edge computing, comprising: BIM model data and construction site energy consumption time-series data are collected by a data acquisition terminal with edge computing capabilities to generate an initial energy consumption dataset. The initial energy consumption dataset is subjected to multi-scale temporal and spatial data partitioning to generate a feature dataset; Based on the aforementioned feature dataset, a space-time model is constructed to generate overall energy consumption trend prediction results. High energy consumption points are identified based on the overall energy consumption trend prediction results. Based on the overall energy consumption trend prediction results and high energy consumption point data, optimization suggestions are generated and pushed to the management terminal.

[0006] Preferably, generating the initial energy consumption dataset includes: Data acquisition terminals with edge computing capabilities at the construction site establish data interface connections with the BIM system to collect BIM model data; The data acquisition terminal collects energy consumption time-series data by communicating with field sensors; The BIM model data and energy consumption time series data are aligned with a unified timestamp and matched according to spatial location to generate an initial energy consumption dataset.

[0007] Preferably, the BIM model data includes structural layer data and regional layer data; The structural layer data includes the geometric and attribute information of building components; The regional layer data includes spatial partitioning structure. The energy consumption time-series data includes continuous time records of electricity, water resources, gas, and equipment energy consumption.

[0008] Preferably, the step of generating the feature dataset includes: The energy consumption time series data is divided into sliding windows on the scale of minutes, hours, and days, and the time characteristics of total energy consumption, peak value, and fluctuation within each window are calculated. Specifically, in each sliding window Internally, define the energy consumption function with time variable. Introducing regional unit energy density factor Exponential rate of change control factor and normalized regulation factor The following integral expression is used to construct a window-level composite time feature model:

[0009] in, For the first The composite energy consumption characteristic value of a region or component under the current time window expresses the overall intensity and dynamic trend of energy consumption in that region; The sampling normalization coefficient within the time window; For the first in the sliding window Standardized energy consumption function per unit time for the region; This indicates the energy density per unit area of ​​the region; This represents the exponential growth rate of energy consumption changes within the region. To adjust the normalized scaling parameter of the overall response sensitivity; the above integral expression is valid over a finite interval. The above will be carried out. A spatial mapping is established based on structural layer data and regional layer data to extract spatial characteristics of energy consumption per unit area, load density, and equipment activity in each region. The temporal and spatial features are concatenated to output a feature dataset.

[0010] Preferably, the generation of overall energy consumption trend prediction results includes: The feature dataset is divided into multiple spatial-temporal series samples based on region ID and time series. Each series sample contains composite temporal feature values ​​and spatial feature vectors of the corresponding spatial unit within a continuous time window. By encoding region attributes in the spatial dimension and preserving the sliding window order in the temporal dimension, a model input format with temporal dynamics and spatial distribution is constructed, providing high-dimensional structured data support for subsequent predictive modeling.

[0011] Based on this, a Long Short-Term Memory (LSTM) network model is used for training and modeling. This model utilizes its internal control structure to remember and retrieve long-term and short-term energy consumption changes, effectively capturing the periodic, abrupt, and trend characteristics in construction site energy consumption data, thereby more accurately simulating energy consumption evolution behavior under complex working conditions.

[0012] For each sequence sample, the following energy consumption integral expression model driven by activation function is introduced to construct the time prediction mapping relationship:

[0013] in, Within the time period, the first Energy consumption forecasts for each region Total energy consumption measured within the window; Energy consumption amplitude; Energy consumption fluctuation intensity; tanh , Activation coefficient, amplitude weight, smoothing factor.

[0014] This expression serves as the logic for constructing feature vectors, providing a stable non-linear feature embedding mechanism for model training.

[0015] Finally, the network parameters are optimized through backpropagation and iterative training to minimize the mean squared error loss, outputting the overall energy consumption trend prediction results for all spatial regions over different time periods. This model not only integrates historical dependency information from time series data but also preserves the differences in energy consumption across spatial distributions, significantly improving prediction accuracy and timeliness compared to traditional statistical methods.

[0016] Preferably, identifying high-energy-consumption points includes: A threshold judgment is performed on the overall energy consumption trend prediction results to filter out the spatial areas where energy consumption exceeds the preset value. Among them, based on the following formula

[0017] in, For the first A spatial region in continuous The probability value of being judged as high energy consumption within a time window. Indicates the first The region in the first Energy consumption prediction values ​​for each window. The energy consumption warning threshold can be set. For indicator functions (i.e., when) (If the value is 1, it is 0 otherwise). This formula is used to quantify the frequency of excessive energy consumption in each region and to achieve preliminary screening of anomalies.

[0018] Based on this, for the selected spatial regions where energy consumption exceeds the preset value, a continuous analysis of the energy consumption results in these regions is further performed over multiple time windows to eliminate short-term abnormal results and output stable high-energy-consumption points. To ensure the persistence and stability of anomalies, the following expression can be further used:

[0019] in, Indicates the first A spatial region in any continuous Energy consumption consistently exceeds the threshold within a given time window. The number of events. When When the condition is met, the region can be identified as a stable high-energy-consumption point. This continuity verification mechanism effectively reduces the interference of instantaneous fluctuations and improves the accuracy and reliability of high-energy-consumption identification.

[0020] Preferably, the step of generating optimization suggestions based on the overall energy consumption trend prediction results and high energy consumption points includes: Read the historical energy consumption sequence and spatial characteristics of high energy consumption points, and perform causal attribution analysis; Identifying high-frequency fluctuations and peak load patterns specifically includes: analyzing a unit spatial region over a time interval. Energy consumption time series data within After extracting trend and volatility features, the following indicators are suggested for generation:

[0021] Where S represents the suggested metrics, and T represents the time window width. Let be the energy consumption per unit area at time t. The rate of change of energy consumption This represents the average energy consumption within that time window. For high-frequency fluctuation weighting coefficients, This is the peak load fluctuation weighting coefficient. This formula is used to simultaneously identify the rapid change trend of energy consumption per unit time and the overall degree of deviation, thereby quantitatively determining whether there are obvious anomalies in the regional load pattern.

[0022] Based on the calculated range of S values, the system automatically generates energy-saving optimization suggestions using a tiered strategy: if If the fluctuation and load pattern are stable, no suggestions are generated; otherwise... Generate general optimization suggestions; if If the data is identified as a high-priority, high-energy-consumption fluctuation range, specific energy-saving suggestion parameters are generated. These suggestion parameters are bound to the high-energy-consumption area identifier and the abnormal time period. Finally, structured energy-saving optimization suggestion data is output and pushed to the construction site management terminal.

[0023] This invention also provides a construction site energy consumption intelligent monitoring system based on BIM and edge computing, used to execute the above-mentioned construction site energy consumption intelligent monitoring method based on BIM and edge computing, including: The data acquisition module is used to collect BIM model data and construction site energy consumption time series data through an acquisition terminal with edge computing capabilities, and to align and match the data according to timestamps and spatial locations to generate an initial energy consumption dataset. The feature construction module is used to perform multi-scale time and spatial dimension data partitioning on the initial energy consumption dataset, extract time and spatial features, and splice them together to form a feature dataset. The trend prediction module is used to construct a space-time prediction model based on the feature dataset and generate overall energy consumption trend prediction results for different spatial regions. The high energy consumption identification module is used to perform threshold judgment and continuity analysis based on the overall energy consumption trend prediction results to identify stable high energy consumption points; The optimization suggestion module is used to combine historical feature data and spatial distribution characteristics of high energy consumption points to generate structured energy-saving optimization suggestions and push them to the management terminal at the construction site.

[0024] The beneficial effects of the technical solution provided by this invention are as follows: This approach effectively integrates dynamic energy consumption data from the construction site with structural information from the static BIM model, giving the energy consumption data both timestamp and spatial location dimensions. This step improves the accuracy and contextual semantic relevance of the energy consumption data, enabling subsequent analysis to have higher positioning capabilities and scene reconstruction accuracy. Ultimately, it ensures that the intelligent monitoring system has scalability and high compatibility from the data source.

[0025] It achieves detailed decomposition of energy consumption data at different time granularities (minutes, hours, days) and different spatial scales (component layer, region layer), thereby constructing a high-dimensional feature vector containing dimensions such as total energy consumption, peak value, fluctuation, and density. This step helps to reveal the differences in energy consumption patterns in different time periods and regions, ultimately enabling the model to have the ability to finely identify energy consumption patterns and providing strong feature support for deep modeling.

[0026] It achieves deep learning modeling by combining information on temporal dynamics and spatial structure; specifically, the LSTM model can memorize and fit complex sequential energy consumption data, capture implicit relationships such as periodic loads and abnormal trends, and thus obtain the future energy consumption evolution trend at the regional level; this step enhances the predictability and foresight of the system, and provides a scientific basis for early detection of potential energy waste at construction sites.

[0027] It enables intelligent screening and dynamic positioning of areas with abnormal energy consumption. Furthermore, by using time continuity judgment and threshold identification methods, it eliminates false alarm interference caused by short-term fluctuations, thereby outputting true and stable high-energy-consuming areas. This step effectively improves monitoring accuracy, enabling managers to accurately locate key energy-saving control areas without increasing the burden of on-site inspections, ultimately significantly improving the targeting and response efficiency of energy efficiency management.

[0028] It has achieved a closed-loop management mechanism from "problem discovery" to "intelligent solution recommendation". By reading historical data and spatial characteristics, it automatically performs cause analysis and identifies characteristic patterns (such as load peaks and high-frequency fluctuations). Then, it generates structured optimization suggestions based on strategy templates, ensuring that the suggestions have specific time, region and operation dimensions. This step enhances the system's decision support function, reduces the cost of manual analysis, and ultimately helps construction sites achieve continuous energy efficiency optimization and intelligent management upgrades. Attached Figure Description

[0029] Figure 1 This is a diagram illustrating the steps of the method of the present invention; Figure 2 This is a system module diagram of the present invention. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. Of course, the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0031] Example 1 This invention provides a method for intelligent monitoring of energy consumption at construction sites based on BIM and edge computing, comprising: S1 collects BIM model data and construction site energy consumption time-series data through a data acquisition terminal with edge computing capabilities, generating an initial energy consumption dataset: Data acquisition terminals with edge computing capabilities at the construction site establish data interface connections with the BIM system to collect BIM model data; The data acquisition terminal collects energy consumption time-series data by communicating with field sensors; The BIM model data and energy consumption time series data are aligned with a unified timestamp and matched according to spatial location to generate an initial energy consumption dataset.

[0032] The BIM model data includes building information model data for the structural layer and the regional layer, wherein... Structural layer data, including geometric and attribute information of building components; Regional layer data, including spatial partitioning structure, Energy consumption time-series data, including continuous time records of electricity, water resources, gas, and equipment energy consumption.

[0033] In step S1, BIM model data and energy consumption time-series data are aligned with a unified timestamp and matched spatially to form an initial energy consumption dataset. This process enables traditional static building information to acquire dynamic energy consumption attributes, thus overcoming the limitation of BIM in the construction phase, which is limited to structural information management. The alignment mechanism ensures the synchronous fusion of multi-source heterogeneous data in both time and spatial dimensions, providing a high-quality data foundation for subsequent deep feature extraction. In practical engineering, this alignment operation effectively avoids analytical biases caused by sensor drift and data loss, laying a unified coordinate system foundation for spatiotemporal modeling.

[0034] S2 performs multi-scale temporal and spatial data partitioning on the initial energy consumption dataset to generate a feature dataset: The energy consumption time series data is divided into sliding windows on the scale of minutes, hours, and days, and the time characteristics of total energy consumption, peak value, and fluctuation within each window are calculated. Specifically, in each sliding window Internally, define the energy consumption function with time variable. Introducing regional unit energy density factor Exponential rate of change control factor and normalized regulation factor The following integral expression is used to construct a window-level composite time feature model:

[0035] in, For the first The composite energy consumption characteristic value of a region or component under the current time window expresses the overall intensity and dynamic trend of energy consumption in that region; The sampling normalization coefficient within the time window; For the first in the sliding window Standardized energy consumption function per unit time for the region; This indicates the energy density per unit area of ​​the region; This represents the exponential growth rate of energy consumption changes within the region. To adjust the normalized scaling parameter of the overall response sensitivity; the above integral expression is valid over a finite interval. The above will be carried out. Composite energy consumption characteristic value is used to comprehensively measure the energy intensity, spatial density, and dynamic growth trend of a region per unit time. It is an important characteristic indicator expressing the energy consumption behavior pattern of the region. This characteristic value not only reflects the instantaneous energy consumption state, but also captures the rate of change and amplification trend of energy consumption over time, and has strong time sensitivity and spatial adaptability.

[0036] In subsequent steps, this composite energy consumption feature value will be used as one of the core variables, together with the spatial structure features, to form a high-dimensional feature vector, which will be input into the space-time modeling module for energy consumption trend prediction and high-energy consumption area identification.

[0037] A spatial mapping is established based on the structural layer and the regional layer to extract the spatial characteristics of energy consumption per unit area, load density and equipment activity of each region. The temporal and spatial features are concatenated to output a feature dataset.

[0038] Step S2 introduces a sliding window and an integral expression to construct a window-level composite time feature model. This expression reflects the functional coupling between energy consumption fluctuation trends and response sensitivity, and is an expression of high-order nonlinear feature extraction, which cannot be achieved in existing average-based statistical analysis. Its function is to simultaneously quantify the total energy consumption and the rate of change, making the time series data exhibit stronger discriminative power during model training.

[0039] S3 constructs a space-time model based on the aforementioned feature dataset to generate overall energy consumption trend prediction results: The feature dataset is divided into multiple spatial-temporal series samples based on region ID and time series. Each series sample contains composite temporal feature values ​​and spatial feature vectors of the corresponding spatial unit within a continuous time window. By encoding region attributes in the spatial dimension and preserving the sliding window order in the temporal dimension, a model input format with temporal dynamics and spatial distribution is constructed, providing high-dimensional structured data support for subsequent predictive modeling.

[0040] Based on this, a Long Short-Term Memory (LSTM) network model is used for training and modeling. This model utilizes its internal control structure to remember and retrieve long-term and short-term energy consumption changes, effectively capturing the periodic, abrupt, and trend characteristics in construction site energy consumption data, thereby more accurately simulating energy consumption evolution behavior under complex working conditions.

[0041] For each sequence sample, the following energy consumption integral expression model driven by activation function is introduced to construct the time prediction mapping relationship:

[0042] in, Within the time period, the first Energy consumption forecasts for each region Total energy consumption measured within the window; Energy consumption amplitude; Energy consumption fluctuation intensity; tanh , Activation coefficient, amplitude weight, smoothing factor.

[0043] This expression serves as the logic for constructing feature vectors, providing a stable non-linear feature embedding mechanism for model training.

[0044] Finally, the network parameters are optimized through backpropagation and iterative training to minimize the mean squared error loss, outputting the overall energy consumption trend prediction results for all spatial regions over different time periods. This model not only integrates historical dependency information from time series data but also preserves the differences in energy consumption across spatial distributions, significantly improving prediction accuracy and timeliness compared to traditional statistical methods.

[0045] In S3, an LSTM model is used to model high-dimensional space-time feature sequences, enabling effective learning of complex, nonlinear, and highly periodic energy consumption behaviors at construction sites. Combined with input expressions including variables such as amplitude, fluctuation, and normalized energy consumption, the model not only possesses the ability to memorize long-term dependent information but also responds quickly to sudden changes in energy consumption. Compared to traditional linear models, LSTM is better suited for capturing the cyclical changes in load and potential anomalies in energy consumption sequences, thereby improving the timeliness and accuracy of predictions.

[0046] S4 identifies high-energy-consuming points based on the overall energy consumption trend prediction results: A threshold judgment is performed on the overall energy consumption trend prediction results to filter out the spatial areas where energy consumption exceeds the preset value. Among them, based on the following formula

[0047] in, For the first A spatial region in continuous The probability value of being judged as high energy consumption within a time window. Indicates the first The region in the first Energy consumption prediction values ​​for each window. The energy consumption warning threshold can be set. For indicator functions (i.e., when) (If the value is 1, it is 0 otherwise). This formula is used to quantify the frequency of excessive energy consumption in each region and to achieve preliminary screening of anomalies.

[0048] For the selected spatial regions where energy consumption exceeds the preset value, further continuous analysis is performed on the results within multiple time windows to eliminate short-term anomalies and output stable high-energy-consumption points. To ensure the persistence and stability of anomalies, the following expression can be further used:

[0049] in, Indicates the first A spatial region in any continuous Energy consumption consistently exceeds the threshold within a given time window. The number of events. When When the condition is met, the region can be identified as a stable high-energy-consumption point. This continuity verification mechanism effectively reduces the interference of instantaneous fluctuations and improves the accuracy and reliability of high-energy-consumption identification.

[0050] The threshold judgment function and continuity verification function designed in step S4 embody a two-layer mechanism from anomaly identification to stability screening. The threshold judgment function quantifies the degree of anomaly in a region by accumulating high-energy-consumption judgment frequencies across multiple time windows, while the continuity verification function further verifies whether the anomaly is persistent, thereby effectively eliminating transient spike interference. This method has dynamic adaptability, improving practicality while ensuring identification accuracy, and is suitable for deployment needs under actual construction site conditions.

[0051] S5 generates optimization suggestions based on the overall energy consumption trend prediction results and high energy consumption point data, and pushes the optimization suggestions to the management terminal: Read the historical energy consumption sequence and spatial characteristics of high energy consumption points, and perform causal attribution analysis; Identifying high-frequency fluctuations and peak load patterns specifically includes: analyzing a unit spatial region over a time interval. Energy consumption time series data within After extracting trend and volatility features, the following indicators are suggested for generation:

[0052] Where S represents the suggested metrics, and T represents the time window width. Let be the energy consumption per unit area at time t. The rate of change of energy consumption This represents the average energy consumption within that time window. For high-frequency fluctuation weighting coefficients, This is the peak load fluctuation weighting coefficient. This formula is used to simultaneously identify the rapid change trend of energy consumption per unit time and the overall degree of deviation, thereby quantitatively determining whether there are obvious anomalies in the regional load pattern.

[0053] Based on the calculated range of S values, the system automatically generates energy-saving optimization suggestions using a tiered strategy: if If the fluctuation and load pattern are stable, no suggestions are generated; otherwise... Generate general optimization suggestions; if If the data is identified as a high-priority, high-energy-consumption fluctuation range, specific energy-saving suggestion parameters are generated. These suggestion parameters are bound to the high-energy-consumption area identifier and the abnormal time period. Finally, structured energy-saving optimization suggestion data is output and pushed to the construction site management terminal.

[0054] The suggestion generation index S in S5, built upon high-frequency fluctuations and load anomalies, integrates energy consumption change rate, average deviation, and weighting coefficients to effectively characterize the complex features of energy consumption behavior. By classifying the S value range, a logical hierarchy of suggestion output is achieved, enabling differentiated response mechanisms for optimization suggestions. This closed-loop path from feature extraction to suggestion delivery overcomes the bottleneck of traditional data analysis results being unable to be implemented, constructing an intelligent energy efficiency management chain of "perception—prediction—optimization."

[0055] Example 2 This invention also provides a construction site energy consumption intelligent monitoring system based on BIM and edge computing, used to execute the above-mentioned construction site energy consumption intelligent monitoring method based on BIM and edge computing, including: The data acquisition module is used to collect BIM model data and construction site energy consumption time series data through an acquisition terminal with edge computing capabilities, and to align and match the data according to timestamps and spatial locations to generate an initial energy consumption dataset. The feature construction module is used to perform multi-scale time and spatial dimension data partitioning on the initial energy consumption dataset, extract time and spatial features, and splice them together to form a feature dataset. The trend prediction module is used to construct a space-time prediction model based on the feature dataset and generate overall energy consumption trend prediction results for different spatial regions. The high energy consumption identification module is used to perform threshold judgment and continuity analysis based on the overall energy consumption trend prediction results to identify stable high energy consumption points; The optimization suggestion module is used to combine historical feature data and spatial distribution characteristics of high energy consumption points to generate structured energy-saving optimization suggestions and push them to the management terminal at the construction site.

[0056] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent monitoring of energy consumption at construction sites based on BIM and edge computing, characterized in that, include: BIM model data and construction site energy consumption time-series data are collected by a data acquisition terminal with edge computing capabilities to generate an initial energy consumption dataset. The initial energy consumption dataset is subjected to multi-scale temporal and spatial data partitioning to generate a feature dataset; Based on the aforementioned feature dataset, a space-time model is constructed to generate overall energy consumption trend prediction results. High energy consumption points are identified based on the overall energy consumption trend prediction results. Based on the overall energy consumption trend prediction results and high energy consumption point data, optimization suggestions are generated and pushed to the management terminal.

2. The intelligent monitoring method for construction site energy consumption based on BIM and edge computing according to claim 1, characterized in that, The initial energy consumption dataset generated includes: Data acquisition terminals with edge computing capabilities at the construction site establish data interface connections with the BIM system to collect BIM model data; The data acquisition terminal collects energy consumption time-series data by communicating with field sensors; The BIM model data and energy consumption time series data are aligned with a unified timestamp and matched according to spatial location to generate an initial energy consumption dataset.

3. The intelligent monitoring method for construction site energy consumption based on BIM and edge computing according to claim 2, characterized in that, The BIM model data includes structural layer data and area layer data; The structural layer data This includes the geometric and attribute information of building components; The regional layer data includes a spatial partitioning structure. The energy consumption time-series data includes continuous time records of electricity, water resources, gas, and equipment energy consumption.

4. The intelligent monitoring method for construction site energy consumption based on BIM and edge computing according to claim 3, characterized in that, The steps for generating the feature dataset include: The energy consumption time series data is divided into sliding windows on the scale of minutes, hours, and days, and the time characteristics of total energy consumption, peak value, and fluctuation within each window are calculated. A spatial mapping is established based on structural layer data and regional layer data to extract spatial characteristics of energy consumption per unit area, load density, and equipment activity in each region. The temporal and spatial features are concatenated to output a feature dataset.

5. The intelligent monitoring method for construction site energy consumption based on BIM and edge computing according to claim 4, characterized in that, The overall energy consumption trend prediction results include: The feature dataset is divided into sequence samples according to region and time. A long short-term memory network model is used to map sequence samples to energy consumption prediction values ​​for the corresponding time periods; By minimizing the mean square error through iterative training, the overall energy consumption trend prediction results for all regions are output.

6. The intelligent monitoring method for construction site energy consumption based on BIM and edge computing according to claim 5, characterized in that, The identification of high-energy-consuming points includes: A threshold judgment is performed on the overall energy consumption trend prediction results to filter out spatial areas where energy consumption exceeds the preset value; For the selected spatial regions where energy consumption exceeds the preset value, a continuous analysis of the energy consumption results of the spatial region within multiple time windows is further performed to eliminate short-term abnormal results and output stable high-energy-consumption points.

7. The intelligent monitoring method for construction site energy consumption based on BIM and edge computing according to claim 6, characterized in that, The optimization suggestions include: Read the historical energy consumption sequence and spatial characteristics of high energy consumption points, and perform causal attribution analysis; Identify high-frequency fluctuations and peak load patterns; Based on the matching strategy template, suggested parameters are generated, the region identifier and time period are bound, structured energy-saving optimization suggestion data is output, and pushed to the management terminal.

8. A smart energy consumption monitoring system for construction sites based on BIM and edge computing, characterized in that, The method for intelligent monitoring of construction site energy consumption based on BIM and edge computing as described in any one of claims 1-7 includes: The data acquisition module is used to collect BIM model data and construction site energy consumption time series data through an acquisition terminal with edge computing capabilities, and to align and match the data according to timestamps and spatial locations to generate an initial energy consumption dataset. The feature construction module is used to perform multi-scale time and spatial dimension data partitioning on the initial energy consumption dataset, extract time and spatial features, and splice them together to form a feature dataset. The trend prediction module is used to construct a space-time prediction model based on the feature dataset and generate overall energy consumption trend prediction results for different spatial regions. The high energy consumption identification module is used to perform threshold judgment and continuity analysis based on the overall energy consumption trend prediction results to identify stable high energy consumption points; The optimization suggestion module is used to combine historical feature data and spatial distribution characteristics of high energy consumption points to generate structured energy-saving optimization suggestions and push them to the management terminal at the construction site.

Images