Building carbon emission analysis method and system combining virtual simulation with knowledge graph

Abstract

The present application relates to the technical field of carbon emission analysis, and particularly relates to a building carbon emission analysis method and system combining virtual simulation and knowledge graph. The method comprises the following steps: obtaining multi-source data streams in building operation, generating interaction feature vectors through clustering algorithm, and constructing carbon emission knowledge graph and attribute dependency path; secondly, calculating the Euclidean distance between different clustering clusters, identifying potential collision areas and generating conflict identification codes, further calculating the synergistic effect value, and generating a fusion point set; then, updating the initial data set through the fusion point set, obtaining a scenario parameter set, and constructing an emission path mapping model according to the scenario parameter set to obtain emission simulation parameters; finally, based on the emission simulation parameters, the dynamic distribution of building carbon emission is generated. The present application can provide efficient and accurate building carbon emission management, and provide strong support for building energy efficiency optimization and low-carbon emission reduction.

Description

Technical Field

[0001] This invention relates to the field of carbon emission analysis technology, and in particular to a building carbon emission analysis method and system that combines virtual simulation and knowledge graph. Background Technology

[0002] Carbon emission management in building operations is a crucial area for achieving low-carbon goals and is key to promoting sustainable development and addressing climate change. With the increasing demand for intelligent and green buildings, accurately simulating carbon emissions during building operations has become central to optimizing energy use and reducing environmental impact. However, current technologies face significant challenges in processing building operation data, making it difficult to meet the needs of dynamic and precise management.

[0003] Existing methods often fail to effectively handle the complexity of multi-source heterogeneous data when integrating building operation data. Building operations involve multiple data sources, including environmental, equipment, and usage patterns. These data vary in format and update frequency, leading to difficulties in data integration. For example, ambient temperature data collected by sensors may differ from equipment operating status data in timestamps or units, affecting the accuracy of the analysis. Furthermore, existing technologies often overlook the dynamic interactions between attributes when analyzing dependencies between data, resulting in an inability to accurately capture key factors influencing carbon emissions.

[0004] The core technical challenge lies in accurately extracting attribute dependency paths from multi-source data. Data from building operations not only comes from diverse sources but also contains complex interactions. Furthermore, the dependencies between multi-source data are difficult to define clearly. Without accurately identifying these dependency paths, it's difficult to determine which factors have the greatest impact on carbon emissions in specific scenarios. For example, in office buildings, air conditioning operation during peak summer periods can lead to a surge in carbon emissions, but existing methods struggle to dynamically analyze the causal relationship between temperature increases and equipment operation, and cannot identify potential conflict areas, such as contradictions between equipment operation and personnel usage patterns.

[0005] Therefore, how to extract and construct clear attribute dependency paths from multi-source heterogeneous data in order to dynamically reveal the influencing factors of carbon emissions and their interactions has become a key issue in building carbon emission simulation. Summary of the Invention

[0006] This invention provides a building carbon emission analysis method and system combining virtual simulation and knowledge graph, used for high-precision simulation and optimization of the dynamic distribution of building carbon emissions. In a first aspect, this invention provides a building carbon emission analysis method combining virtual simulation and knowledge graph, comprising:

[0007] Step S1: Acquire multi-source data streams during building operation and preprocess them to obtain an initial dataset; based on the initial dataset, use a clustering algorithm to generate interactive feature vectors and construct a carbon emission knowledge graph and corresponding attribute dependency paths;

[0008] Step S2: Calculate the Euclidean distance between the interaction feature vectors corresponding to different clusters. If the Euclidean distance is less than a preset collision threshold, it is determined to be a potential collision region and a conflict identification code is generated. Based on the conflict identification code, calculate the synergy effect value and generate a set of fusion points.

[0009] Step S3: Update the initial dataset through the fusion point set to obtain the scenario parameter set; construct the emission path mapping model based on the scenario parameter set to obtain emission simulation parameters;

[0010] Step S4: Generate the dynamic distribution of building carbon emissions using the emission simulation parameters.

[0011] As a preferred embodiment of the present invention, step S1, obtaining the multi-source data stream during building operation to obtain the initial dataset, includes:

[0012] Environmental monitoring data, equipment status data, and usage pattern data are collected by multi-source sensors to obtain a raw data set. Data cleaning technology is used to preprocess the raw data set to remove noisy data and missing values, resulting in a standardized data set. The environmental monitoring data includes at least temperature, humidity, air quality, and light intensity; the equipment status data includes the operating status and energy consumption information of the equipment; and the usage pattern data includes the usage pattern information of the equipment in the building.

[0013] As a preferred embodiment of the present invention, step S1 involves constructing a carbon emission knowledge graph and obtaining the corresponding attribute dependency paths, including:

[0014] The K-means clustering algorithm is used to group the environmental monitoring data and equipment status data of buildings in the initial dataset to obtain a set of similar data points; the Euclidean distance between each point in the set of similar data points is calculated to generate an interaction feature vector;

[0015] Based on the interaction feature vector, a feature representation matrix is ​​constructed to determine the correlation strength between data points. If the correlation strength exceeds a preset threshold, node links are constructed and established in the carbon emission knowledge graph to obtain preliminary attribute dependencies.

[0016] By analyzing the topological structure of the node links, the dependency paths between attributes are extracted to obtain a path set. A graph traversal algorithm is used to parse the hierarchical structure of the path set to determine the contextual logical relationships between attributes. Based on the contextual logical relationships, the weights of the interaction feature vectors are updated to obtain an optimized feature representation.

[0017] As a preferred embodiment of the present invention, step S2, determining the potential collision region and generating a collision identification code, includes:

[0018] Calculate the Euclidean distance between the interaction feature vectors corresponding to different clusters. If the Euclidean distance is less than a preset collision threshold, use a clustering algorithm to extract a collision candidate set within the potential collision region. Using the collision candidate set, use a decision tree algorithm to analyze the association between data attributes, generate attribute association rules, and obtain an attribute association set.

[0019] Based on the attribute association set and combined with the node link structure of the knowledge graph, the association paths between nodes are mapped to obtain a set of node association paths. If the path length in the set of node association paths is less than a preset path threshold, it is determined to be a strong association path, and a conflict identification code is generated to obtain a set of conflict identification codes.

[0020] As a preferred embodiment of the present invention, step S2, based on the conflict identification code, calculates the synergy effect value and generates a set of fusion points, including:

[0021] The conflict identification code is matched using a preset rule set to obtain the rule matching degree; based on the rule matching degree, a synergistic effect value is generated using a forward calculation method, and the effect value weight is determined.

[0022] The entity relationship chain is extracted by conflict generation rules to obtain the relationship chain strength; the relationship chain strength and the effect value weight are integrated by the inference chain fusion method to generate a fusion feature set; the coordinates of the cooperating points are calculated based on the fusion feature set to obtain the coordinate positioning accuracy.

[0023] The preset rule set is adjusted based on the coordinate positioning accuracy, and the conflict identification code is updated; the preset rule set is repeatedly matched according to the updated conflict identification code to obtain the optimized rule matching degree.

[0024] As a preferred embodiment of the present invention, step S3 involves updating the initial dataset using the fusion point set to obtain a scenario parameter set, including:

[0025] If the synergistic effect value is greater than a set value, corresponding data points are obtained from the initial dataset, and fusion points are generated through a fusion mechanism to obtain a fusion point set; the initial dataset is updated through the fusion point set to obtain an updated dataset; scene parameters are obtained according to the virtual scene, and the updated dataset is optimized through a parameter adjustment mechanism to obtain an optimized dataset; if the accuracy analysis result of the optimized dataset meets a preset threshold, a scenario parameter set is obtained.

[0026] The scenario construction process is adjusted by adjusting the scenario parameter set, and the virtual scenario is optimized using the support vector machine algorithm to obtain optimized scenario parameters. The fusion mechanism is updated according to the optimized scenario parameters, and the fusion points are re-evaluated using the random forest algorithm to obtain an updated fusion point set. If the synergistic effect value of the updated fusion point set is greater than zero, the optimized dataset is iteratively updated by the data processing module to obtain the final scenario parameters.

[0027] As a preferred embodiment of the present invention, in step S3, an emission path mapping model is constructed based on the scenario parameter set to obtain emission simulation parameters, including:

[0028] An initial dataset of the building environment is generated using virtual simulation technology, containing information on building structure and energy consumption, to obtain a multi-dimensional scenario dataset. Point data is extracted from the multi-dimensional scenario dataset, and multi-source information is integrated using a data flow fusion method to obtain an integrated scenario parameter set.

[0029] If the correlation between parameters in the integrated scenario parameter set exceeds a preset threshold, key scenario features are extracted using principal component analysis to determine carbon emission influencing factors; based on the key scenario features and the carbon emission factors, an emission path mapping model is constructed to obtain the dynamic distribution of building carbon emissions and obtain the path mapping matrix.

[0030] Emission simulation parameters are extracted from the path mapping matrix, and the carbon emission values ​​of buildings under different scenarios are predicted using the random forest algorithm to determine the emission prediction results. Based on the emission prediction results, the parameter configuration of the virtual simulation scenario is adjusted, the initial dataset is updated, and optimized building carbon emission simulation parameters are obtained. For the optimized building carbon emission simulation parameters, a dynamic visualization of the emission path is generated.

[0031] As a preferred embodiment of the present invention, generating a dynamic distribution of building carbon emissions using the emission simulation parameters includes:

[0032] The initial dataset is adjusted using a preset parameter optimization method, and an iteratively updated dataset is obtained through a feedback loop. The iteratively updated dataset is then processed using path mapping technology to obtain carbon emission path features. Scenario parameters are extracted from these carbon emission path features to generate a scenario parameter set. An extraction chain is constructed based on the scenario parameter set to determine the input conditions for emission simulation. Random forest is used to process the input conditions in the extraction chain to generate emission simulation vector groups. The emission simulation vector groups are decomposed using path analysis technology to obtain micro-emission path distributions. Cluster analysis is used to group the micro-emission path distributions to obtain refined building emission paths.

[0033] Secondly, the present invention also provides a building carbon emission analysis system combining virtual simulation and knowledge graph, for implementing the above-mentioned method, the system comprising:

[0034] The graph construction unit is used to acquire multi-source data streams during building operation and preprocess them to obtain an initial dataset; based on the initial dataset, a clustering algorithm is used to generate interactive feature vectors to construct a carbon emission knowledge graph and corresponding attribute dependency paths.

[0035] The collaborative computing unit is used to calculate the Euclidean distance between the interaction feature vectors corresponding to different clusters. If the Euclidean distance is less than a preset collision threshold, it is determined to be a potential collision region and a conflict identification code is generated. Based on the conflict identification code, the collaborative effect value is calculated and a set of fusion points is generated.

[0036] The parameter acquisition unit is used to update the initial dataset through the fusion point set to obtain the scenario parameter set; and to construct an emission path mapping model based on the scenario parameter set to obtain emission simulation parameters.

[0037] The distribution generation unit is used to generate a dynamic distribution of building carbon emissions using the emission simulation parameters.

[0038] Thirdly, the present invention also provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the above-described method.

[0039] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0040] This invention acquires multi-source data streams from building operations using virtual simulation technology. Combined with environmental monitoring data, equipment status data, and usage pattern data, an initial dataset is formed, providing a foundation for subsequent analysis and optimization. Next, a clustering algorithm is used to process the initial data, generating interaction feature vectors. Based on these vectors, a carbon emission knowledge graph is constructed. The clustering algorithm groups similar building environment and equipment data, identifying potential interaction relationships by calculating Euclidean distance between data points. This provides a clear logical structure for identifying carbon emission pathways, enhancing the correlation between data points and laying a solid foundation for subsequent path mapping and simulation. Through Euclidean distance calculation and cluster analysis, potential collision regions can be identified and conflict identification codes generated. These conflict identification codes are further used to calculate synergistic effect values ​​to determine which data points have the most significant impact on carbon emissions, thus forming a set of fusion points that can optimize the data. The relationships between points improve the accuracy of subsequent simulations and analyses. By acquiring and optimizing scenario parameter sets, combined with parameter adjustment mechanisms in virtual scenarios, the simulation parameters for building carbon emissions can be continuously optimized. The random forest algorithm is used to adjust the path mapping model, enabling building carbon emission simulations to adapt to different scenario conditions and accurately predict carbon emission values ​​under various simulation scenarios, ensuring the reliability and accuracy of data processing. Path analysis and cluster analysis are used to conduct in-depth analysis of micro-emission paths, classifying building carbon emission paths into low, medium, and high emission groups. This process refines the identification of carbon emission paths, helping to accurately identify and optimize high-emission areas. Furthermore, the dynamic visualization function of emission paths generated based on optimized simulation parameters helps relevant personnel monitor and manage building carbon emission status in real time, providing strong support for optimizing building energy efficiency and achieving low-carbon goals. Through the above technical solutions, each step forms an effective synergy and interaction, making the building carbon emission analysis process more accurate and dynamic, and effectively supporting the implementation of carbon reduction strategies. Attached Figure Description

[0041] Figure 1 This is a flowchart of a building carbon emission analysis method combining virtual simulation and knowledge graph in an embodiment of the present invention;

[0042] Figure 2 This is a structural diagram of a building carbon emission analysis system that combines virtual simulation and knowledge graph in an embodiment of the present invention. Detailed Implementation

[0043] To further understand the content of this invention, a detailed description of the invention is provided in conjunction with the accompanying drawings and embodiments. The specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0044] like Figure 1 This embodiment presents a building carbon emission analysis method combining virtual simulation and knowledge graph, specifically including:

[0045] Step S1: Obtain multi-source data streams during building operation to obtain an initial dataset; based on the initial dataset, use a clustering algorithm to generate interactive feature vectors, construct a carbon emission knowledge graph, and obtain attribute dependency paths in the carbon emission knowledge graph;

[0046] In step S1, obtaining the initial dataset by acquiring multi-source data streams during building operation includes:

[0047] Environmental monitoring data, equipment status data, and usage pattern data are collected by multi-source sensors to obtain a raw data set. Data cleaning technology is used to preprocess the raw data set to remove noisy data and missing values, resulting in a standardized data set. The environmental monitoring data includes at least temperature, humidity, air quality, and light intensity; the equipment status data includes the operating status and energy consumption information of the equipment; and the usage pattern data includes the usage pattern information of the equipment in the building.

[0048] Specifically, by deploying multi-source sensors inside and around the building, environmental monitoring data, equipment status data, and usage pattern data are collected. Environmental monitoring data includes at least temperature, humidity, air quality, and light intensity, which helps reflect the environmental conditions inside and outside the building. Equipment status data includes the operating status and energy consumption information of equipment such as air conditioning, lighting, and heating systems. Usage pattern data includes equipment usage pattern information, which helps assess the actual operating mode of the building.

[0049] The original dataset is preprocessed using data cleaning techniques. The data cleaning process includes removing redundant information, eliminating noisy data, and removing missing values. This process ensures data quality and accuracy by setting thresholds or using interpolation methods to fill in missing data and eliminate outliers. The cleaned dataset is then standardized, converting data from different sources and organizations into a unified standard format. Standardization ensures that different types of data can be analyzed and compared uniformly, providing a high-quality, standardized dataset for subsequent processing.

[0050] Further, in step S1, a carbon emission knowledge graph is constructed, and the attribute dependency paths in the carbon emission knowledge graph are obtained, including:

[0051] The K-means clustering algorithm is used to group the environmental monitoring data and equipment status data of buildings in the initial dataset to obtain a set of similar data points; the Euclidean distance between each point in the set of similar data points is calculated to generate an interaction feature vector;

[0052] Based on the interaction feature vector, a feature representation matrix is ​​constructed to determine the correlation strength between data points. If the correlation strength exceeds a preset threshold, node links are constructed and established in the carbon emission knowledge graph to obtain preliminary attribute dependencies.

[0053] By analyzing the topological structure of the node links, the dependency paths between attributes are extracted to obtain a path set. A graph traversal algorithm is used to parse the hierarchical structure of the path set to determine the contextual logical relationships between attributes. Based on the contextual logical relationships, the weights of the interaction feature vectors are updated to obtain an optimized feature representation.

[0054] Specifically, the K-means clustering algorithm is used to perform multi-dimensional clustering analysis on the environmental monitoring data of buildings in the initial dataset, such as temperature, humidity, and light intensity, and equipment status data, such as air conditioning power, lighting energy consumption, and elevator operation status. Based on the feature similarity between data points, the dataset is divided into multiple clusters, thus obtaining a set of similar data points representing different operating states. Each of these data points includes the environmental monitoring data and equipment status data at the corresponding time point. For each set of similar data points, the Euclidean distance between any two data points within a cluster is calculated in the multi-dimensional feature space. This distance is obtained by taking the square root of the sum of the squares of the differences in each dimension. These distance values ​​are then combined in order to form an interaction feature vector, which quantifies the degree of interaction between data points within a cluster. The magnitude of this vector directly reflects the dispersion of the data within the cluster; the smaller the magnitude of the interaction feature vector, the closer the points are to each other. The stronger the interaction, the stronger the correlation, and the more consistent the behavior. Conversely, the larger the modulus, the greater the deviation between data points; the weaker the interaction, the weaker the correlation, and the less consistent the behavior. Further, based on the generated interaction feature vectors, a feature representation matrix is ​​constructed, where the matrix elements are the reciprocals of the corresponding distance values. This transforms distance information into a correlation strength metric. When the correlation strength exceeds a preset threshold set according to building type, the data attributes corresponding to the data in the aforementioned interaction vectors are used as nodes in the carbon emission knowledge graph. The correlations between each node are analyzed and linked to establish preliminary attribute dependencies. For example, the temperature node is linked to the air conditioning energy consumption node to identify strong dependency paths between them. Based on this, by analyzing the topological structure of node links in the knowledge graph, dependency paths between attributes are extracted and a path set is formed. The analysis of the topological structure helps identify dependency paths between different attributes. For example, the path from the temperature node to the emission node (equipment node) through the energy consumption node can reveal how temperature changes affect equipment energy consumption, thereby affecting carbon emissions. Based on the path set, a graph traversal algorithm is used to parse the hierarchical structure of the path and further determine the contextual logical relationships between attributes. Specifically, a depth-first search algorithm is used to traverse the path set from the starting node to the terminal node of the path to parse the hierarchical dependency structure, identify first-level and second-level dependencies, and determine the causal or conditional relationships between attributes.

[0055] Based on the contextual logic, the weights of the interaction feature vectors are updated to obtain an optimized feature representation. Specifically, the elements in the interaction feature vectors are adjusted according to the weights based on the parsed causal or conditional relationships. For example, if the causal relationship between temperature and equipment energy consumption is strong, the weight of the temperature node in the feature vector can be increased accordingly to optimize the feature representation. In this way, the updated feature representation can more accurately reflect the dynamic interaction between different data points during building operation, providing a precise data foundation for subsequent dynamic simulation and path optimization of carbon emissions.

[0056] Step S2: Calculate the Euclidean distance between the interaction feature vectors corresponding to different clusters. If the Euclidean distance is less than a preset collision threshold, it is determined to be a potential collision region and a conflict identification code is generated. Based on the conflict identification code, calculate the synergy effect value and generate a set of fusion points.

[0057] In step S2, determining potential collision regions and generating collision identification codes includes:

[0058] Calculate the Euclidean distance between the interaction feature vectors corresponding to different clusters. If the Euclidean distance is less than a preset collision threshold, use a clustering algorithm to extract a collision candidate set within the potential collision region. Using the collision candidate set, use a decision tree algorithm to analyze the association between data attributes, generate attribute association rules, and obtain an attribute association set.

[0059] Based on the attribute association set, combined with the node link structure of the carbon emission knowledge graph, the association paths between nodes are mapped to obtain a node association path set. If the path length in the node association path set is less than a preset path threshold, it is determined to be a strong association path, and a conflict identification code is generated to obtain a conflict identification code set.

[0060] Using an entity relationship reasoning chain, logical relationships between attributes are extracted from the conflict identification code set to generate an attribute mapping relationship set. The accuracy of the attribute mapping relationship is verified by combining the attribute mapping relationship set with the node links of the knowledge graph to obtain a verified attribute mapping set. Conflict identification results are generated based on the verified attribute mapping set.

[0061] Specifically, in the scenario of building carbon emission analysis, it is first necessary to acquire environmental monitoring data of the building, such as temperature and humidity, and equipment status data, such as air conditioning energy consumption, lighting energy consumption, and the on / off status of corresponding equipment, through multi-source sensors. These data may have certain similarities in different areas of the building. Therefore, the relationship between different data points is evaluated by calculating the Euclidean distance between the corresponding vectors of environmental monitoring data and equipment energy consumption data points in the initial dataset. The Euclidean distance is the straight-line distance between data points and is used to measure the similarity between two data points. Assuming that environmental data and equipment status data are represented in vector form, the Euclidean distance between each pair of data points reflects their proximity in the feature space. If the Euclidean distance between a pair of data points is less than a preset collision threshold, it is considered that there may be a potential conflict area between the two data points.

[0062] For example, if a certain area has a high temperature and the air conditioner consumes a lot of energy, while an adjacent area has a high temperature and fluctuating air conditioner energy consumption, these similar data points may form a potential collision zone, indicating that there is a potential conflict of carbon emissions within the building.

[0063] After identifying potential collision regions, K-means clustering is further applied to group similar data points. K-means clustering is an unsupervised learning algorithm that iteratively calculates the distance from each data point to the cluster center, minimizing the distance from each data point to the cluster center, thereby dividing the data points into multiple clusters. The cluster centers are initialized by randomly selecting data points. After multiple iterations, the cluster allocation stabilizes, and each cluster contains a group of similar data points.

[0064] In the context of building carbon emissions, K-means clustering can divide data points such as temperature, humidity, and energy consumption into multiple clusters based on their similarity. Each cluster represents a building area or equipment status. For example, cluster 1 may contain data points with high temperature and high air conditioning energy consumption, while cluster 2 contains areas with high humidity and high energy consumption. Through clustering, potential collision areas can be effectively extracted to form a collision candidate set.

[0065] For the collision candidate set extracted from the clustering algorithm, the association between data attributes is analyzed by decision tree analysis to generate attribute association rules. Through decision tree analysis, the potential association between different attributes such as temperature and air conditioning energy consumption can be revealed.

[0066] The decision tree calculates the information gain of each attribute and selects the optimal split attribute. The information gain is measured by the change in entropy, i.e., the change in data uncertainty, before and after the attribute split. Each time the split occurs, the decision tree selects the attribute that maximizes the information gain, thus recursively generating branches of the tree. In building carbon emission analysis, possible rules include: when the temperature is higher than the set temperature value, the air conditioning energy consumption increases sharply, or when the humidity increases, the air conditioning energy consumption also increases, etc.

[0067] The aforementioned attribute association rules help reveal potential conflicts related to carbon emissions within buildings. For example, if there are significant association rules between temperature and air conditioning energy consumption, these rules can be used to detect early on situations where increased temperature leads to increased air conditioning energy consumption, which may in turn lead to increased carbon emissions. The generated attribute association rules are combined with the carbon emission knowledge graph to map the association paths between nodes. As a data structure, the knowledge graph uses nodes to represent data attributes, such as environmental monitoring nodes and equipment energy consumption nodes, and links to represent the dependencies between these attributes. The rules in the attribute association set are mapped to the nodes of the carbon emission knowledge graph, forming dependencies between attribute nodes.

[0068] Dijkstra's algorithm can be used to calculate the shortest path between different nodes in a knowledge graph. Dijkstra's algorithm prioritizes the path with the lowest cumulative weight, which is usually based on the strength of the links between nodes, such as the correlation between temperature and air conditioning energy consumption. The above path helps to reveal the implicit dependencies between different attributes, such as the temperature node affecting the air conditioning energy consumption node through the humidity node. Through such mapping, we can clearly understand the interaction between different building areas, environmental conditions and carbon emissions, and thus optimize carbon emission path analysis.

[0069] If the path length in the set of associated paths of a node is less than a preset path threshold, it is judged as a strongly associated path, and a conflict identification code is generated for this path. There may be a relatively close mutual dependency relationship, thus generating a unique conflict identification code. For example, if the path of temperature, humidity and air conditioning energy consumption is short, it means that the relationship between temperature and air conditioning energy consumption is very close. This path will be marked as a strongly associated path and assigned a conflict identification code, namely a carbon emission risk identification code. The entity relationship reasoning chain is used to extract the logical relationship between attributes from the set of conflict identification codes and generate an attribute mapping relationship set. The reasoning chain can gradually extract the logical relationship between different attributes through forward reasoning rules, such as if A is associated with B and B is associated with C, then A is associated with C. In the scenario of building carbon emissions, the reasoning chain can help deduce the logical relationship between temperature and air conditioning energy consumption, and how these factors affect the overall path of carbon emissions.

[0070] Based on the verified attribute mapping relationship set, combined with the node links of the carbon emission knowledge graph, the accuracy of the mapping relationship is verified. Through graph traversal algorithm, the validity of the path is ensured and invalid mappings are eliminated, thereby obtaining the verified attribute mapping set and obtaining the above-mentioned conflict identification results. The above technical solution, by verifying the carbon emission path under different building scenarios, can more accurately reflect the actual impact of factors such as temperature, humidity, and equipment energy consumption on building carbon emissions, and reduce simulation errors caused by data collisions.

[0071] Further, in step S2, based on the conflict identification code, a synergy effect value is calculated to generate a set of fusion points, including:

[0072] The conflict identification code is matched using a preset rule set to obtain the rule matching degree; based on the rule matching degree, a synergistic effect value is generated using a forward calculation method, and the effect value weight is determined.

[0073] The entity relationship chain is extracted by conflict generation rules to obtain the relationship chain strength; the relationship chain strength and the effect value weight are integrated by the inference chain fusion method to generate a fusion feature set; the coordinates of the cooperating points are calculated based on the fusion feature set to obtain the coordinate positioning accuracy.

[0074] The preset rule set is adjusted based on the coordinate positioning accuracy, and the conflict identification code is updated; the preset rule set is repeatedly matched according to the updated conflict identification code to obtain the optimized rule matching degree.

[0075] Specifically, in the scenario of building carbon emission analysis, conflict identification codes are compared with building energy efficiency pattern templates through a preset rule set. For example, a rule set is set to identify excessive air conditioning energy consumption under high temperature and humidity conditions. A preliminary matching value is obtained by calculating path similarity. Then, the cosine similarity algorithm is used to further calculate the fit between the conflict identification code and the rule set. Cosine similarity can measure the similarity between two vectors by the product of the vector inner product and the vector modulus, thereby obtaining the rule matching degree.

[0076] In building carbon emission analysis, synergy values ​​reflect the mutual influence and correlation strength between different attributes such as temperature and air conditioning energy consumption. The forward calculation method constructs effect values ​​by accumulating rule matching degree components layer by layer. For example, assuming a rule matching degree of 0.8 and a maximum value of 1, different weighting factors are gradually added from the initial matching degree value to accumulate and generate synergy value. Then, based on the generated synergy value, the effect value weight is calculated. The effect value weight is quantified by dividing the synergy value by the sum of the total matching degrees to ensure that the weight can accurately reflect the relative importance between the attributes. For example, if temperature has a greater impact on energy consumption, the effect value weight of temperature will be higher, while the weight of humidity will be lower.

[0077] Based on the generated conflict identification code and rule matching degree, entity relationship chains are extracted through conflict generation rules. In the context of building carbon emissions, entity relationship chains refer to the interdependent paths between different attributes such as temperature and energy consumption. By mapping the relationships between these attributes to the node links of the knowledge graph, inference chains are formed. Inference chains can not only reveal the causal relationships between attributes, but also help identify potential peaks in building carbon emissions; for example, an increase in temperature will lead to an increase in air conditioning energy consumption, thereby driving up carbon emissions.

[0078] Next, the strength value is obtained by weighted summation of the Euclidean distances of each link in the chain. Euclidean distance is used to quantify the tightness of attribute dependency paths; the smaller the distance, the stronger the relationship chain between attributes. For example, the relationship chain strength between temperature and air conditioning energy consumption may be high because these two attributes are closely linked in building carbon emissions. Calculating the relationship chain strength helps determine the impact of these attribute combinations on carbon emissions, thereby improving the accuracy of subsequent calculations and optimizations. Through the inference chain fusion method, the relationship chain strength and effect value weights are integrated to generate a fusion feature set. The inference chain fusion method combines the relationship chain strength and effect value weights through matrix multiplication to obtain a fusion vector, which reflects the comprehensive impact of various attributes in building energy efficiency analysis.

[0079] Based on this fusion vector, data points are grouped using a clustering algorithm to generate a fusion feature set. K-means clustering iteratively assigns similar feature vectors to the same cluster center, ensuring that the feature set can effectively capture the dynamic interaction features between different data attributes. For example, in the scenario of building carbon emissions, the fusion feature set can comprehensively consider the interaction of attributes such as temperature, humidity, and energy consumption to obtain a more comprehensive feature representation. After generating the fusion feature set, the cluster centers are iteratively adjusted to minimize the intra-cluster variance, further optimizing the accuracy of the fusion feature set, thereby improving the accuracy of building carbon emission analysis and avoiding simulation bias caused by data conflicts.

[0080] Based on the fused feature set, the coordinates of the collaborative point are calculated. The collaborative point coordinates are obtained by weighted averaging of the key vectors in the fused feature set. The calculation method sums the feature values ​​of each dimension in a weighted manner to obtain a coordinate point that represents the interaction of building energy efficiency. This coordinate point can reflect the relative importance of different attributes in building carbon emissions. The accuracy of coordinate positioning is evaluated by calculating the percentage deviation between the coordinate point and the actual collision zone threshold. Ensuring that the positioning accuracy is higher than the preset threshold can help make the building energy efficiency simulation more accurate, thereby optimizing the building carbon emission path mapping.

[0081] Based on the calculated coordinate positioning accuracy, the preset rule set is adjusted. If the positioning accuracy is lower than the preset threshold, the number of matching templates in the rule set will increase, thus enabling the rule set to better adapt to the needs of the building carbon emission model. Subsequently, a conflict identification code is regenerated based on the adjusted rule set, and the code value is optimized through an updated entity relationship reasoning chain to ensure that it accurately reflects the updated dependency path.

[0082] The above-mentioned technical solutions can accurately identify potential conflict areas in building carbon emissions, optimize energy efficiency analysis and emission simulation, thereby providing accurate data support for achieving building energy conservation and emission reduction.

[0083] Step S3: Update the initial dataset through the fusion point set to obtain the scenario parameter set; construct the emission path mapping model based on the scenario parameter set to obtain emission simulation parameters;

[0084] In step S3, updating the initial dataset using the fusion point set to obtain the scenario parameter set includes:

[0085] If the synergistic effect value is greater than a set value, corresponding data points are obtained from the initial dataset, and fusion points are generated through a fusion mechanism to obtain a fusion point set; the initial dataset is updated through the fusion point set to obtain an updated dataset; scene parameters are obtained according to the virtual scene, and the updated dataset is optimized through a parameter adjustment mechanism to obtain an optimized dataset; if the accuracy analysis result of the optimized dataset meets a preset threshold, a scenario parameter set is obtained.

[0086] The scenario construction process is adjusted by adjusting the scenario parameter set, and the virtual scenario is optimized using the support vector machine algorithm to obtain optimized scenario parameters. The fusion mechanism is updated according to the optimized scenario parameters, and the fusion points are re-evaluated using the random forest algorithm to obtain an updated fusion point set. If the synergistic effect value of the updated fusion point set is greater than zero, the optimized dataset is iteratively updated by the data processing module to obtain the final scenario parameters.

[0087] Specifically, in one embodiment, when the synergistic effect value is greater than a set effect value, the corresponding data points related to the conflict identification code are first extracted from the initial dataset. These data points are similar data points that have been grouped by a clustering algorithm and match the dependency paths between attributes identified by the carbon emission knowledge graph node links. For example, in building carbon emission analysis, the data points include temperature, humidity, and equipment energy consumption. These data points are considered to have similarity and may have some inherent interaction characteristics. These characteristics are grouped by a clustering algorithm, and the dependencies between data points are determined by carbon emission knowledge graph mapping.

[0088] These data points are integrated through a fusion mechanism, which here refers to a vector merging process based on Euclidean distance calculation. The purpose is to add and normalize the interaction feature vectors between data points to generate new fusion points. These fusion points represent the optimized representation of the dependency paths between initially identified attributes. In the building carbon emission scenario, these fusion points may represent the merging of data from multiple similar areas, such as merging temperature and equipment energy consumption data from multiple floors into one fusion point, representing the carbon emission characteristics of these areas. Forming the above fusion points into a fusion point set can more accurately reflect the interaction between the environment and energy consumption within the building, providing accurate data input for subsequent optimization and simulation.

[0089] Mapping each fusion point in the fusion point set back to its corresponding position in the initial dataset—that is, overwriting the original data point with the value of the fusion point—ensures that the update of the initial dataset reflects the latest attribute dependency paths and provides an accurate basis for subsequent analysis. In the building carbon emission scenario, this update mechanism, by replacing the original temperature, humidity, and energy consumption data points, enables the updated dataset to effectively reflect the latest interactions between these data. For example, if the relationship between air conditioning energy consumption and temperature on a certain floor changes, i.e., is reflected through fusion points, then new fusion points will replace the original data points so that the updated energy efficiency model more accurately reflects the actual carbon emissions.

[0090] Scenario parameters are obtained through virtual scenarios. These parameters include environmental monitoring data during building operation (such as temperature and humidity) and equipment status data (such as energy consumption of air conditioning and lighting). Virtual scenarios typically represent the building's operating status and environment, simulating the impact of different factors (e.g., the relationship between indoor temperature and external weather) on energy consumption. By extracting parameters from these virtual scenarios, more accurate data input can be provided for building carbon emission models.

[0091] For example, in a virtual scenario of a high-energy-consuming building, the changes in indoor air conditioning energy consumption under different external temperatures are simulated. The extracted scenario parameters, such as the air conditioning operating status at different temperatures, can serve as the basis for subsequent calculations and simulations. The scenario parameters provide data on the actual environment and equipment status that the building may encounter during operation, providing a basis for subsequent optimization and parameter adjustments.

[0092] The updated dataset is optimized through a parameter tuning mechanism, an algorithm based on gradient descent. This mechanism applies scenario parameter weights to each data point in the dataset and iteratively adjusts these weights to minimize the Euclidean distance between data points. Euclidean distance calculates the straight-line distance between different data points, ensuring that the similarity and interaction relationships between data points are accurately reflected in the updated dataset.

[0093] In building carbon emission scenarios, this process allows for the dynamic adjustment of energy consumption models for different floors or areas, making the relationships between parameters such as temperature, humidity, and equipment energy consumption more precise. For example, if the air conditioning equipment in a certain area has low energy efficiency under certain environmental conditions, optimization using the gradient descent method can improve the energy efficiency model for that area, thereby reducing the overall carbon emissions of the building.

[0094] Whether to extract scenario parameters is determined by calculating the accuracy of the optimized dataset. The accuracy analysis evaluates the update effect of the dataset by comparing it with the mean squared error of the initial dataset. If the error between the optimized dataset and the initial dataset is less than a preset threshold, the updated dataset is considered to be accurate enough, and scenario parameters can be extracted from the optimized dataset to form a scenario parameter set.

[0095] These scenario parameter sets include various data such as building interior temperature, humidity, and equipment energy consumption. These optimized and adjusted data are more representative and can provide more accurate input data for simulating building carbon emission pathways. These scenario parameter sets will be used in subsequent steps to further optimize building energy efficiency analysis and carbon emission models.

[0096] In summary, by updating the initial dataset through the fusion point set and combining virtual scene optimization with the efficient application of support vector machine and random forest algorithms, dynamic optimization and accurate simulation in building carbon emission analysis can be achieved, thus providing strong data support and decision-making basis for carbon emission path mapping and building energy efficiency optimization.

[0097] Further, in step S3, an emission path mapping model is constructed based on the scenario parameter set to obtain emission simulation parameters, including:

[0098] An initial dataset of the building environment is generated using virtual simulation technology, containing information on building structure and energy consumption, to obtain a multi-dimensional scenario dataset. Point data is extracted from the multi-dimensional scenario dataset, and multi-source information is integrated using a data flow fusion method to obtain an integrated scenario parameter set.

[0099] If the correlation between parameters in the integrated scenario parameter set exceeds a preset threshold, key scenario features are extracted using principal component analysis to determine carbon emission influencing factors; based on the key scenario features and the carbon emission factors, an emission path mapping model is constructed to obtain the dynamic distribution of building carbon emissions and obtain the path mapping matrix.

[0100] Emission simulation parameters are extracted from the path mapping matrix, and the carbon emission values ​​of buildings under different scenarios are predicted using the random forest algorithm to determine the emission prediction results. Based on the emission prediction results, the parameter configuration of the virtual simulation scenario is adjusted, the initial dataset is updated, and optimized building carbon emission simulation parameters are obtained. For the optimized building carbon emission simulation parameters, a dynamic visualization of the emission path is generated.

[0101] Specifically, this invention generates an initial dataset of a building environment using virtual simulation technology. This dataset includes building structure data such as wall thickness and window area, as well as energy consumption information such as lighting equipment power and air conditioning system energy consumption. The multidimensional data in the initial dataset are combined, with each dimension representing a data type, such as structural dimension and energy consumption dimension, to form a multidimensional scenario dataset, which provides a foundation for subsequent data processing.

[0102] After acquiring the initial dataset, multi-source information is integrated using a data stream fusion method. By selecting energy consumption readings at specific time points and fusing them with sensor data such as temperature and humidity, redundancy is eliminated and an optimized scenario parameter set is obtained. This scenario parameter set is then further analyzed. If the correlation between parameters exceeds a preset threshold, the principal component analysis (PCA) algorithm is activated. By calculating the covariance matrix, eigenvalues, and eigenvectors, the high-dimensional data is reduced in dimensionality, and the main factors affecting carbon emissions, such as energy intensity and structural density, are extracted.

[0103] Based on these key scenario characteristics, an emission path mapping model was constructed. This model maps these characteristics onto carbon emission paths and uses linear transformations and nonlinear activation functions to generate a path mapping matrix representing the dynamic distribution of carbon emissions. In this process, if there are potential dependent paths between data points, such as path segments whose Euclidean distance is less than a set threshold, they are identified by linking knowledge graph nodes. The relationship between attributes is analyzed in conjunction with decision tree branch conditions to avoid distribution bias and further refine the path mapping model.

[0104] After obtaining the path mapping matrix, the extracted path features are further processed using the random forest algorithm to predict the carbon emission values ​​of buildings under different scenarios. Specifically, multiple decision trees are constructed, and the voting results generate emission simulation vector groups that reflect the probability of building carbon emissions under different scenarios. To refine the prediction, the vector groups are decomposed using path analysis technology to obtain the micro-emission path distribution. Path points are then fused based on synergistic effect values, such as values ​​greater than a set value, to ensure coverage of all areas of the building and improve simulation accuracy.

[0105] Next, the K-means algorithm was used to cluster the micro-emission pathways, identifying low, medium, and high emission pathway groups. This provides detailed data support for building carbon emission management, helps identify priority optimization areas, and promotes the effective formulation of emission control strategies. To further improve the simulation effect, the parameter configuration of the virtual simulation scenario was adjusted based on the simulation results, and optimized building carbon emission simulation parameters were obtained by updating the initial dataset. During the adjustment process, the virtual scenario was optimized using the support vector machine algorithm to obtain more accurate scenario parameters. The adjusted parameter configuration was then re-evaluated using the random forest algorithm to re-evaluate the fusion point set. If the synergistic effect value of the updated fusion point set was greater than a set value, the optimized dataset was iteratively updated to finally generate an accurate scenario parameter set.

[0106] The above-mentioned technical solution, by combining virtual simulation with knowledge graphs, achieves high-precision and dynamic visualization in the simulation of building carbon emissions, providing a scientific basis and technical support for building energy efficiency optimization and low-carbon emission reduction.

[0107] Step S4: Generate the dynamic distribution of building carbon emissions using the emission simulation parameters; specifically including:

[0108] The initial dataset is adjusted using a preset parameter optimization method, and an iteratively updated dataset is obtained through a feedback loop. The iteratively updated dataset is then processed using path mapping technology to obtain carbon emission path features. Scenario parameters are extracted from these carbon emission path features to generate a scenario parameter set. An extraction chain is constructed based on the scenario parameter set to determine the input conditions for emission simulation. Random forest is used to process the input conditions in the extraction chain to generate emission simulation vector groups. The emission simulation vector groups are decomposed using path analysis technology to obtain micro-emission path distributions. Cluster analysis is used to group the micro-emission path distributions to obtain refined building emission paths.

[0109] Specifically, in this embodiment, multi-dimensional data such as building environment, equipment status and usage patterns are collected through multi-source data streams to form an initial dataset. The gradient descent algorithm is also used to iteratively optimize the data points in the initial dataset. By calculating the gradient of the loss function, the error between the predicted value and the actual environmental monitoring data is minimized, and the parameters of the dataset are gradually adjusted until the dataset converges to form an iteratively updated dataset.

[0110] After obtaining the updated dataset, path mapping technology is applied for further processing. Specifically, path mapping technology maps the updated dataset to the nodes of the carbon emission knowledge graph to identify the dependency paths between data points, and extracts path features by calculating Euclidean distance. If the Euclidean distance between path segments is less than a preset threshold, it is marked as a potential path segment, and then the path feature vector is extracted. Combined with the entity relationship reasoning chain, carbon emission path features are further generated. Through logical reasoning rules, relevant attributes such as equipment status and usage pattern data are connected to reveal the key carbon emission path features in building operation.

[0111] Based on the aforementioned carbon emission pathway characteristics, a set of scenario parameters is extracted and generated, including environmental variables such as temperature and humidity during building operation. These parameters then form an extraction chain, linked together by a set of rule definitions to create an ordered sequence of parameters that serves as input conditions for emission simulation. During the construction of the extraction chain, a conflict identification code mechanism is used to verify the compatibility between parameters, ensuring the rationality of the input conditions. Through this extraction chain, the input conditions for emission simulation are determined, providing accurate data support for subsequent simulations.

[0112] The random forest algorithm is used to process the input conditions in these extraction chains to generate emission simulation vector sets. The random forest integrates multiple decision trees to process different input features and generate multiple possibilities for emission simulation, thus forming an emission prediction vector set. Furthermore, the emission simulation vector set is decomposed through path analysis such as depth-first search to obtain the micro-emission path distribution. The path branch points are identified through graph traversal algorithm, the dependencies of each path branch are analyzed, and the emission contribution of each micro-path is calculated to further refine the carbon emission sources of each area within the building.

[0113] Finally, cluster analysis, such as K-means, is used to group the distribution of micro-emission pathways, thereby obtaining refined building emission pathways. The above grouping is used to identify pathways with different emission intensities, such as low-emission pathways, medium-emission pathways, and high-emission pathways, by iteratively calculating the cluster centers. This provides a basis for the refined management of building carbon emissions. Specifically, by identifying priority optimization areas through cluster analysis, different carbon reduction strategies can be adopted for pathways with different emission levels, thereby improving the overall efficiency of emission management.

[0114] The above-mentioned technical solution effectively overcomes the challenges posed by the complexity and dynamism of data in existing building carbon emission analysis. By combining virtual simulation technology with knowledge graphs, it accurately simulates and analyzes the dynamic distribution of building carbon emissions. Through multidimensional mapping and analysis of data, it not only improves the accuracy and robustness of carbon emission simulation, but also provides strong technical support for building energy optimization and carbon emission reduction.

[0115] This invention also provides a building carbon emission analysis system that combines virtual simulation and knowledge graph to implement the above-mentioned methods, such as... Figure 2 As shown, the system includes:

[0116] The graph construction unit is used to acquire multi-source data streams during building operation and preprocess them to obtain an initial dataset; based on the initial dataset, a clustering algorithm is used to generate interactive feature vectors to construct a carbon emission knowledge graph and corresponding attribute dependency paths.

[0117] The collaborative computing unit is used to calculate the Euclidean distance between the interaction feature vectors corresponding to different clusters. If the Euclidean distance is less than a preset collision threshold, it is determined to be a potential collision region and a conflict identification code is generated. Based on the conflict identification code, the collaborative effect value is calculated and a set of fusion points is generated.

[0118] The parameter acquisition unit is used to update the initial dataset through the fusion point set to obtain the scenario parameter set; and to construct an emission path mapping model based on the scenario parameter set to obtain emission simulation parameters.

[0119] The distribution generation unit is used to generate a dynamic distribution of building carbon emissions using the emission simulation parameters.

[0120] The present invention also provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the above-described method.

[0121] In summary, this invention acquires multi-source data streams from building operations using virtual simulation technology. Combined with environmental monitoring data, equipment status data, and usage pattern data, an initial dataset is formed, providing a foundation for subsequent analysis and optimization. Next, a clustering algorithm is used to process the initial data, generating interaction feature vectors. Based on these vectors, a carbon emission knowledge graph is constructed. The clustering algorithm groups similar building environment and equipment data, identifying potential interaction relationships by calculating the Euclidean distance between data points. This provides a clear logical structure for identifying carbon emission pathways, enhancing the correlation between data points and laying a solid foundation for subsequent path mapping and simulation. Through Euclidean distance calculation and cluster analysis, potential collision regions can be identified and conflict identification codes generated. These conflict identification codes are further used to calculate synergistic effect values ​​to determine which data points have the most significant impact on carbon emissions, thus forming a set of fusion points that can optimize... The relationships between data points improve the accuracy of subsequent simulations and analyses. By acquiring and optimizing scenario parameter sets, combined with parameter adjustment mechanisms in virtual scenarios, the simulation parameters for building carbon emissions can be continuously optimized. The random forest algorithm is used to adjust the path mapping model, enabling building carbon emission simulations to adapt to different scenario conditions and accurately predict carbon emission values ​​under various simulation scenarios, ensuring the reliability and accuracy of data processing. Path analysis and cluster analysis are used to conduct in-depth analysis of micro-emission paths, classifying building carbon emission paths into low, medium, and high emission groups. This process refines the identification of carbon emission paths, helping to accurately identify and optimize high-emission areas. Furthermore, the dynamic visualization function of emission paths generated based on optimized simulation parameters helps relevant personnel monitor and manage building carbon emission status in real time, providing strong support for optimizing building energy efficiency and achieving low-carbon goals. Through the above technical solutions, each step forms an effective synergy and interaction, making the building carbon emission analysis process more accurate and dynamic, and effectively supporting the implementation of carbon reduction strategies.

[0122] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A building carbon emission analysis method combining virtual simulation and knowledge graph, characterized in that, include: Step S1: Acquire multi-source data streams during building operation and preprocess them to obtain an initial dataset; Based on the initial dataset, a clustering algorithm is used to generate interactive feature vectors, and a carbon emission knowledge graph and corresponding attribute dependency paths are constructed; wherein, the multi-source data stream includes environmental monitoring data, equipment status data, and usage pattern data; Step S2: Calculate the Euclidean distance between the interaction feature vectors corresponding to different clusters. If the Euclidean distance is less than a preset collision threshold, it is determined to be a potential collision region, and a conflict identification code is generated. Based on the conflict identification code, calculate the synergy effect value and generate a fusion point set. If the synergy effect value is greater than a set value, obtain the corresponding data points from the initial dataset, generate fusion points through a fusion mechanism, and obtain a fusion point set. Update the initial dataset with the fusion point set to obtain an updated dataset. Obtain scene parameters based on the virtual scene, optimize the updated dataset through a parameter adjustment mechanism, and obtain an optimized dataset. If the accuracy analysis result of the optimized dataset meets a preset threshold, a scene parameter set is obtained. Step S3: Update the initial dataset through the fusion point set to obtain a scenario parameter set; construct an emission path mapping model based on the scenario parameter set to obtain emission simulation parameters; wherein, obtaining emission simulation parameters includes: generating an initial dataset of the building environment through virtual simulation technology, containing building structure and energy consumption information, to obtain a multi-dimensional scenario dataset; extracting point data from the multi-dimensional scenario dataset, integrating multi-source information using a data flow fusion method to obtain an integrated scenario parameter set; if the correlation between parameters in the integrated scenario parameter set exceeds a preset threshold, extracting key scenario features through principal component analysis to determine carbon emission influencing factors; constructing an emission path mapping model based on the key scenario features and the carbon emission influencing factors to obtain the dynamic distribution of building carbon emissions, and obtaining a path mapping matrix; extracting emission simulation parameters from the path mapping matrix, using a random forest algorithm to predict the carbon emission values ​​of buildings under different scenarios, and determining the emission prediction results; adjusting the parameter configuration of the virtual simulation scenario based on the emission prediction results, updating the initial dataset, and obtaining optimized building carbon emission simulation parameters; generating dynamic visualization of emission paths for the optimized building carbon emission simulation parameters; Step S4: Generate the dynamic distribution of building carbon emissions using the emission simulation parameters.

2. The method as described in claim 1, characterized in that, In step S1, obtaining the multi-source data stream during building operation to obtain the initial dataset includes: A raw dataset is obtained by collecting environmental monitoring data, equipment status data, and usage pattern data from multiple sources of sensors. The raw dataset is then preprocessed using data cleaning techniques to remove noisy data and missing values, resulting in a standardized dataset. The environmental monitoring data from the multiple sources of sensors includes at least temperature, humidity, air quality, and light intensity. The equipment status data includes the operating status and energy consumption information of air conditioning, lighting, and heating systems. The usage pattern data includes equipment usage pattern information.

3. The method as described in claim 2, characterized in that, In step S1, a carbon emission knowledge graph is constructed, and the corresponding attribute dependency paths are obtained, including: The K-means clustering algorithm is used to group the environmental monitoring data and equipment status data of buildings in the initial dataset to obtain a set of similar data points; the Euclidean distance between each point in the set of similar data points is calculated to generate an interaction feature vector; Based on the interaction feature vector, a feature representation matrix is ​​constructed to determine the correlation strength between data points. If the correlation strength exceeds a preset threshold, node links are constructed and established in the carbon emission knowledge graph to obtain preliminary attribute dependencies. By analyzing the topological structure of the node links, the dependency paths between attributes are extracted to obtain a path set. A graph traversal algorithm is used to parse the hierarchical structure of the path set to determine the contextual logical relationships between attributes. Based on the contextual logical relationships, the weights of the interaction feature vectors are updated to obtain an optimized feature representation.

4. The method as described in claim 1, characterized in that, In step S2, potential collision regions are determined and collision identification codes are generated, including: Calculate the Euclidean distance between the interaction feature vectors corresponding to different clusters. If the Euclidean distance is less than a preset collision threshold, use a clustering algorithm to extract a collision candidate set within the potential collision region. Using the collision candidate set, use a decision tree algorithm to analyze the association between data attributes, generate attribute association rules, and obtain an attribute association set. Based on the attribute association set and combined with the node link structure of the knowledge graph, the association paths between nodes are mapped to obtain a set of node association paths. If the path length in the set of node association paths is less than a preset path threshold, it is determined to be a strong association path, and a conflict identification code is generated to obtain a set of conflict identification codes.

5. The method as described in claim 1, characterized in that, In step S3, the initial dataset is updated using the fusion point set to obtain the scenario parameter set, including: The scenario construction process is adjusted by adjusting the scenario parameter set, and the virtual scenario is optimized using the support vector machine algorithm to obtain optimized scenario parameters. The fusion mechanism is updated according to the optimized scenario parameters, and the fusion points are re-evaluated using the random forest algorithm to obtain an updated fusion point set. If the synergistic effect value of the updated fusion point set is greater than zero, the optimized dataset is iteratively updated by the data processing module to obtain the final scenario parameter set.

6. The method as described in claim 1, characterized in that, The process of generating a dynamic distribution of building carbon emissions using the emission simulation parameters includes: The initial dataset is adjusted using a preset parameter optimization method, and an iteratively updated dataset is obtained through a feedback loop. The iteratively updated dataset is then processed using path mapping technology to obtain carbon emission path features. Scenario parameters are extracted from these carbon emission path features to generate a scenario parameter set. An extraction chain is constructed based on the scenario parameter set to determine the input conditions for emission simulation. Random forest is used to process the input conditions in the extraction chain to generate emission simulation vector groups. The emission simulation vector groups are decomposed using path analysis technology to obtain micro-emission path distributions. Cluster analysis is used to group the micro-emission path distributions to obtain refined building emission paths.

7. A building carbon emission analysis system combining virtual simulation and knowledge graph, used to implement the method as described in any one of claims 1-6, characterized in that, The system includes: The graph construction unit is used to acquire multi-source data streams during building operation and preprocess them to obtain an initial dataset. Based on the initial dataset, a clustering algorithm is used to generate interactive feature vectors and construct a carbon emission knowledge graph and corresponding attribute dependency paths. The multi-source data streams include environmental monitoring data, equipment status data, and usage pattern data. The collaborative computing unit is used to calculate the Euclidean distance between the interaction feature vectors corresponding to different clusters. If the Euclidean distance is less than a preset collision threshold, it is determined to be a potential collision region and a conflict identification code is generated. Based on the conflict identification code, the collaborative effect value is calculated and a set of fusion points is generated. The parameter acquisition unit is used to update the initial dataset through the fusion point set to obtain the scenario parameter set; and to construct an emission path mapping model based on the scenario parameter set to obtain emission simulation parameters. The distribution generation unit is used to generate a dynamic distribution of building carbon emissions using the emission simulation parameters.

8. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the method as described in any one of claims 1-6.

Images