A building energy consumption simulation optimization method and system based on BIM

Abstract

The application provides a BIM-based building energy consumption simulation optimization method and system, which comprises the following steps: introducing a semantic segmentation algorithm to perform component-level semantic labeling on a BIM model, and collecting physical sensing data of building entity components; performing virtual simulation and reverse deduction according to the physical sensing data through a digital twin to generate a corresponding traceability report, and mining cross-domain correlation rules of component functions and energy consumption parameters at different stages through a federated transfer learning algorithm; creating a dynamic mapping relationship between building components and building energy consumption based on the cross-domain correlation rules, generating a dynamic energy consumption simulation deduction framework according to the dynamic mapping relationship and the traceability report; outputting energy consumption evolution trends under different function adjustment scenarios through the dynamic energy consumption simulation deduction framework to generate full-time-series energy consumption simulation results, and outputting an energy consumption optimization scheme according to the full-time-series energy consumption simulation results. The application can improve optimization efficiency.

Description

Technical Field

[0001] This invention relates to the field of building simulation technology, and in particular to a BIM-based method and system for simulating and optimizing building energy consumption. Background Technology

[0002] In the field of building energy conservation, BIM-based building energy consumption simulation and optimization is the core technology for realizing energy consumption management throughout the building's entire life cycle. The BIM model provides accurate digital component / space information for energy consumption simulation, while sensors collect real-time energy consumption data to support simulation optimization. However, the two identification systems are independent of each other: BIM components / spaces use digital geometric identification (such as "west-facing exterior window-1F-01"), while sensor collection points use physical location identification (such as "next to the west-facing window on the 1st floor"), making it impossible to directly achieve intelligent association.

[0003] The existing technology relies on manual matching of two types of identifiers, without establishing a unique intelligent mapping relationship, which can easily lead to data mismatch (such as matching the energy consumption data of room A to the BIM element of room B), resulting in distorted energy consumption simulation results. The energy consumption optimization strategy based on these results loses its specificity. At the same time, manual matching is inefficient, which hinders the progress of simulation optimization work.

[0004] Furthermore, adjustments to the spatial functions of buildings during operation (such as converting office areas into meeting rooms) are quite common and directly alter the building's energy consumption characteristics. However, the existing mapping relationships cannot be automatically updated in sync with these functional adjustments, requiring manual re-association. This not only incurs labor costs but also results in a loss of real-time data correlation, making it difficult for energy consumption simulation optimization to dynamically adapt to the actual building conditions and meet the needs of full life-cycle energy consumption management. Summary of the Invention

[0005] Based on this, the purpose of this invention is to provide a BIM-based building energy consumption simulation and optimization method and system to solve the problem that the mapping relationship created by the existing technology cannot be automatically updated synchronously with the function adjustment, and manual re-association is required, which leads to the energy consumption simulation optimization being unable to dynamically adapt to the actual building state.

[0006] The first aspect of the present invention proposes:

[0007] A BIM-based method for simulating and optimizing building energy consumption, wherein the method includes:

[0008] Based on the bidirectional mapping relationship between the physical characteristics of building components and their digital twin images, a semantic segmentation algorithm is introduced to perform component-level semantic annotation on the BIM model, and physical sensing data of the building entity components are collected simultaneously.

[0009] Based on the semantic annotation, a digital twin is used to perform virtual simulation and reverse inference based on the physical sensing data to generate a corresponding traceability report. Simultaneously, distributed federated learning nodes are constructed based on energy consumption data from the building design, construction, and operation and maintenance stages to mine cross-domain correlation patterns between component functions and energy consumption parameters at different stages through federated transfer learning algorithms.

[0010] Based on the aforementioned cross-domain correlation rules, a dynamic mapping relationship between building components and building energy consumption is created, and a corresponding dynamic energy consumption simulation and deduction framework is generated simultaneously based on the dynamic mapping relationship and the source tracing report.

[0011] The dynamic energy consumption simulation and deduction framework outputs the energy consumption evolution trend under different functional adjustment scenarios to generate full-time energy consumption simulation results, and simultaneously outputs the corresponding energy consumption optimization scheme based on the full-time energy consumption simulation results.

[0012] The beneficial effects of this invention are as follows: This technical solution precisely overcomes the core pain points of existing technologies where the mapping relationship cannot be automatically and synchronously updated with the functional adjustments of building components, requiring manual re-association, which leads to lagging energy consumption simulation optimization and difficulty in dynamically adapting to the real-time actual state of the building. By constructing a two-way mapping relationship between the physical characteristics of building components and their digital twin images, combined with semantic segmentation algorithms to achieve precise semantic annotation at the component level of the BIM model, synchronously and in real-time collecting physical sensor data of building entity components, and combining digital twin virtual simulation reverse inference and cross-stage federated transfer learning to mine cross-domain correlation patterns, this invention efficiently constructs an automatically updatable dynamic mapping relationship between building components and energy consumption, adapting to changes in building status in real time without manual intervention, and significantly improving the accuracy of energy consumption simulation inference.

[0013] Furthermore, the step of mining the cross-domain correlation between component functions and energy consumption parameters at different stages using the federated transfer learning algorithm includes:

[0014] Based on the semantic annotation and the physical sensing data, virtual sample sets for each simulation stage are generated through the digital twin. A cross-validation mechanism between federated nodes is introduced simultaneously. The virtual sample sets are dynamically calibrated based on real samples to output corresponding associated features.

[0015] Based on the aforementioned association features, a knowledge conflict detection algorithm is introduced to identify conflicts in the associated knowledge at each simulation stage, so as to output the corresponding initial association rules.

[0016] The engineering adaptation indicators of each simulation stage are detected, and the initial correlation rules are optimized in combination with the energy consumption control standards of different building types and climate regions to generate the corresponding cross-domain correlation rules.

[0017] Furthermore, the step of optimizing the initial association pattern to generate the corresponding cross-domain association pattern includes:

[0018] The core semantic features in the semantic annotation are extracted and converted into hard constraints. Simultaneously, the fuzzy clustering-Bayesian inference fusion algorithm is combined to cross-match the rules in the initial association pattern to eliminate false association rules and generate intermediate association patterns.

[0019] Based on the temporal characteristics of each simulation stage, a temporal attention mechanism is introduced to capture the dynamic correlation between energy consumption parameters and component functions at different stages.

[0020] Based on the dynamic association characteristics, the intermediate association rules are dynamically adjusted according to time-series parameters to generate the corresponding cross-domain association rules.

[0021] Furthermore, the step of generating a corresponding dynamic energy consumption simulation and extrapolation framework based on the dynamic mapping relationship and the source tracing report includes:

[0022] The transmission links of each simulation stage in the source tracing report are parsed out, and the annotation results based on the semantic annotation are synchronously introduced to generate a collaborative decision-making mechanism of federated learning nodes to generate a dedicated inference unit adapted to the transmission links.

[0023] The virtual sensing data corresponding to the digital twin is extracted from the source tracing report, and the data flow interaction logic between each dedicated simulation unit is dynamically adjusted according to the virtual sensing data to match the corresponding energy consumption simulation requirements.

[0024] An initial simulation framework is created based on the energy consumption simulation requirements, and the initial simulation framework is dynamically modified to generate the dynamic energy consumption simulation framework.

[0025] Furthermore, the step of dynamically correcting the initial simulation framework to generate the corresponding dynamic energy consumption simulation framework includes:

[0026] Based on heterogeneous data from the entire building lifecycle, a knowledge graph of energy consumption influencing factors is constructed, and an attention mechanism is introduced simultaneously to uncover potential influencing factors that are not explicitly associated in the knowledge graph of energy consumption influencing factors.

[0027] The potential influencing factors are converted into corresponding feature parameters, and a corresponding set of inference rules is generated simultaneously based on the feature parameters.

[0028] The initial simulation framework is integrated with the simulation rule set to perform extreme scenario pre-run, and the corresponding performance bottleneck parameters generated during the run are extracted synchronously. The performance bottleneck parameters are then iteratively optimized through an adaptive genetic algorithm to generate the dynamic energy consumption simulation framework.

[0029] Furthermore, the step of outputting the energy consumption evolution trend under different functional adjustment scenarios through the dynamic energy consumption simulation and inference framework to generate full-time series energy consumption simulation results includes:

[0030] The scene feature thresholds of the building design, construction and operation and maintenance stages are extracted, and different functional adjustment scenarios are dynamically matched with each scene feature threshold to generate a scene inference input benchmark that is adapted to the whole life cycle.

[0031] The energy consumption evolution patterns under different functional adjustment scenarios are distinguished by density clustering algorithm, and exclusive prediction sub-models adapted to each of the energy consumption evolution patterns are generated simultaneously through the dynamic energy consumption simulation and inference framework.

[0032] The core parameters of the dedicated prediction sub-model are dynamically adjusted based on the cross-domain correlation rules to output the energy consumption evolution trend. Simultaneously, the various energy consumption evolution trends are integrated and processed to generate the full-time series energy consumption simulation results.

[0033] Furthermore, the step of integrating the various energy consumption evolution trends to generate the full-time-series energy consumption simulation results includes:

[0034] A corresponding time-series correlation inference matrix is ​​created based on the changing pattern of the energy consumption evolution trend, and the corresponding key time nodes are calculated based on the time-series correlation inference matrix.

[0035] Based on the key time nodes, the energy consumption evolution trend is divided into several trend segments, and the temporal misalignment and connection deviation of each trend segment are corrected simultaneously through a graph neural network to generate initial integrated time series data.

[0036] Based on the source tracing report, the initial integrated time series data is subjected to iterative time series optimization processing to generate the corresponding full time series energy consumption simulation results.

[0037] The second aspect of the present invention proposes:

[0038] A BIM-based building energy consumption simulation and optimization system, wherein the system includes:

[0039] The acquisition module is used to perform component-level semantic annotation of the BIM model based on the two-way mapping relationship between the physical characteristics of building components and the digital twin mirror image, and to simultaneously acquire physical sensing data of building entity components.

[0040] The module is used to perform virtual simulation and reverse inference based on the semantic annotation and the physical sensing data through the digital twin to generate a corresponding traceability report. Simultaneously, it constructs distributed federated learning nodes based on the energy consumption data of the building design, construction and operation and maintenance stages, so as to mine the cross-domain correlation between component functions and energy consumption parameters at different stages through federated transfer learning algorithm.

[0041] The generation module is used to create a dynamic mapping relationship between building components and building energy consumption based on the cross-domain association rules, and simultaneously generate a corresponding dynamic energy consumption simulation and deduction framework based on the dynamic mapping relationship and the source tracing report.

[0042] The output module is used to output the energy consumption evolution trend under different functional adjustment scenarios through the dynamic energy consumption simulation and deduction framework, so as to generate full-time energy consumption simulation results and simultaneously output the corresponding energy consumption optimization scheme based on the full-time energy consumption simulation results.

[0043] Furthermore, the building module is specifically used for:

[0044] Based on the semantic annotation and the physical sensing data, virtual sample sets for each simulation stage are generated through the digital twin. A cross-validation mechanism between federated nodes is introduced simultaneously. The virtual sample sets are dynamically calibrated based on real samples to output corresponding associated features.

[0045] Based on the aforementioned association features, a knowledge conflict detection algorithm is introduced to identify conflicts in the associated knowledge at each simulation stage, so as to output the corresponding initial association rules.

[0046] The engineering adaptation indicators of each simulation stage are detected, and the initial correlation rules are optimized in combination with the energy consumption control standards of different building types and climate regions to generate the corresponding cross-domain correlation rules.

[0047] Furthermore, the building module is specifically used for:

[0048] The core semantic features in the semantic annotation are extracted and converted into hard constraints. Simultaneously, the fuzzy clustering-Bayesian inference fusion algorithm is combined to cross-match the rules in the initial association pattern to eliminate false association rules and generate intermediate association patterns.

[0049] Based on the temporal characteristics of each simulation stage, a temporal attention mechanism is introduced to capture the dynamic correlation between energy consumption parameters and component functions at different stages.

[0050] Based on the dynamic association characteristics, the intermediate association rules are dynamically adjusted according to time-series parameters to generate the corresponding cross-domain association rules.

[0051] Furthermore, the generation module is specifically used for:

[0052] The transmission links of each simulation stage in the source tracing report are parsed out, and the annotation results based on the semantic annotation are synchronously introduced to generate a collaborative decision-making mechanism of federated learning nodes to generate a dedicated inference unit adapted to the transmission links.

[0053] The virtual sensing data corresponding to the digital twin is extracted from the source tracing report, and the data flow interaction logic between each dedicated simulation unit is dynamically adjusted according to the virtual sensing data to match the corresponding energy consumption simulation requirements.

[0054] An initial simulation framework is created based on the energy consumption simulation requirements, and the initial simulation framework is dynamically modified to generate the dynamic energy consumption simulation framework.

[0055] Furthermore, the generation module is specifically used for:

[0056] Based on heterogeneous data from the entire building lifecycle, a knowledge graph of energy consumption influencing factors is constructed, and an attention mechanism is introduced simultaneously to uncover potential influencing factors that are not explicitly associated in the knowledge graph of energy consumption influencing factors.

[0057] The potential influencing factors are converted into corresponding feature parameters, and a corresponding set of inference rules is generated simultaneously based on the feature parameters.

[0058] The initial simulation framework is integrated with the simulation rule set to perform extreme scenario pre-run, and the corresponding performance bottleneck parameters generated during the run are extracted synchronously. The performance bottleneck parameters are then iteratively optimized through an adaptive genetic algorithm to generate the dynamic energy consumption simulation framework.

[0059] Furthermore, the output module is specifically used for:

[0060] The scene feature thresholds of the building design, construction and operation and maintenance stages are extracted, and different functional adjustment scenarios are dynamically matched with each scene feature threshold to generate a scene inference input benchmark that is adapted to the whole life cycle.

[0061] The density clustering algorithm is used to distinguish the energy consumption evolution patterns under different functional adjustment scenarios, and the dynamic energy consumption simulation and inference framework is used to generate exclusive prediction sub-models that are adapted to each of the energy consumption evolution patterns.

[0062] The core parameters of the dedicated prediction sub-model are dynamically adjusted based on the cross-domain correlation rules to output the energy consumption evolution trend. Simultaneously, the various energy consumption evolution trends are integrated and processed to generate the full-time series energy consumption simulation results.

[0063] Furthermore, the output module is specifically used for:

[0064] A corresponding time-series correlation inference matrix is ​​created based on the changing pattern of the energy consumption evolution trend, and the corresponding key time nodes are calculated based on the time-series correlation inference matrix.

[0065] Based on the key time nodes, the energy consumption evolution trend is divided into several trend segments, and the temporal misalignment and connection deviation of each trend segment are corrected simultaneously through a graph neural network to generate initial integrated time series data.

[0066] Based on the source tracing report, the initial integrated time series data is subjected to iterative time series optimization processing to generate the corresponding full time series energy consumption simulation results.

[0067] The third aspect of the present invention proposes:

[0068] A computer includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the BIM-based building energy consumption simulation and optimization method as described above.

[0069] The fourth aspect of the present invention proposes:

[0070] A readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the BIM-based building energy consumption simulation and optimization method as described above.

[0071] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0072] Figure 1 A flowchart of the BIM-based building energy consumption simulation and optimization method provided in the first embodiment of the present invention;

[0073] Figure 2 The structural block diagram of the BIM-based building energy consumption simulation and optimization system provided in the third embodiment of the present invention.

[0074] The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Detailed Implementation

[0075] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Several embodiments of the invention are illustrated in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0076] It should be noted that when a component is said to be "fixed to" another component, it can be directly on the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0077] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0078] Please see Figure 1 The image shows a BIM-based building energy consumption simulation and optimization method provided in the first embodiment of the present invention. The BIM-based building energy consumption simulation and optimization method provided in this embodiment can accurately construct the simulation mapping relationship, thereby improving the optimization efficiency.

[0079] Specifically, this embodiment provides:

[0080] A BIM-based method for simulating and optimizing building energy consumption, wherein the method includes:

[0081] Step S10: Based on the bidirectional mapping relationship between the physical characteristics of building components and the digital twin mirror, a semantic segmentation algorithm is introduced to perform component-level semantic annotation on the BIM model, and physical sensing data of building entity components are collected simultaneously.

[0082] It is important to note that, firstly, BIM models are the foundation of building energy consumption simulation. However, traditional BIM models lack component-level semantic associations, and the physical entity data is disconnected from the virtual model, resulting in insufficient accuracy in energy consumption simulation. Therefore, based on the two-way mapping relationship between the physical characteristics of building components (such as material thermal conductivity, dimensions, and installation location) and digital twin images (the physical component status is synchronized to the virtual image, and the virtual simulation results feed back into physical control), semantic segmentation algorithms (such as the U-Net model) are introduced to perform component-level semantic annotation on the BIM model (such as labeling "external wall insulation components," "window lighting components," and "air conditioning heat dissipation components," clarifying the component's function and energy consumption association attributes). Simultaneously, physical sensing data (such as component surface temperature, energy consumption value, and ambient temperature and humidity) is collected through sensors deployed on the building's physical components (such as temperature sensors, humidity sensors, and energy consumption metering sensors). Specifically, semantic annotation lays the foundation for the association between components and energy consumption, while physical sensing data provides real data support for virtual-physical linkage. Together, they constitute the core data input for subsequent simulation and deduction.

[0083] Step S20: Based on the semantic annotation, the digital twin performs virtual simulation and reverse inference based on the physical sensing data to generate a corresponding traceability report. Simultaneously, a distributed federated learning node is constructed based on the energy consumption data of the building design, construction and operation and maintenance stages to mine the cross-domain correlation between component functions and energy consumption parameters at different stages through federated transfer learning algorithm.

[0084] It should be noted that, secondly, to trace the root causes of abnormal energy consumption and uncover the correlation patterns of energy consumption throughout the entire life cycle, based on the aforementioned semantic annotation (clarifying the identity and function of components), a digital twin (a virtual model replicating the physical state and operation process of a building) is used to receive physical sensor data. Virtual simulation and reverse engineering are then performed (e.g., simulating the change in building energy consumption after the thermal insulation performance of a certain exterior wall component declines, tracing the core components that cause increased energy consumption), generating a source tracing report (including information such as component performance status, energy consumption anomaly transmission path, and root cause location). Simultaneously, energy consumption data at each stage of building design, construction, and operation and maintenance are scattered across different entities (design units, construction units, and operation and maintenance units), creating data silos and requiring privacy protection. Therefore, a [further details are needed]. Distributed federated learning nodes (with each stage's data treated as an independent node, without revealing the original data) utilize federated transfer learning algorithms (balancing the privacy protection features of federated learning with the cross-domain adaptability of transfer learning) to uncover cross-domain correlations between component functions at different stages (such as component selection in the design stage, component installation accuracy in the construction stage, and component aging status in the operation and maintenance stage) and energy consumption parameters (such as air conditioning energy consumption, heating energy consumption, and lighting energy consumption). Specifically, this correlation breaks down stage barriers, providing core theoretical support for the subsequent construction of dynamic mapping relationships and the establishment of simulation frameworks.

[0085] Step S30: Based on the cross-domain association rules, a dynamic mapping relationship between building components and building energy consumption is created, and a corresponding dynamic energy consumption simulation and deduction framework is generated simultaneously based on the dynamic mapping relationship and the source tracing report.

[0086] It should be noted that, next, to achieve a precise correlation between component status and energy consumption changes, a dynamic mapping relationship between building components and building energy consumption is created based on cross-domain correlation rules (such as quantifying the correlation relationship that "for every 1cm reduction in the thickness of the external wall insulation layer, winter energy consumption increases by 5%", and this relationship can be dynamically updated as components age and the environment changes); simultaneously, combined with the traceability report (which clarifies the current performance status of the component and the root cause of abnormal energy consumption), the correlation logic of the dynamic mapping relationship and the traceability dimension of the traceability report are integrated to generate a dynamic energy consumption simulation and deduction framework. Specifically, this framework differs from the traditional fixed framework and can dynamically adjust the deduction logic according to component status, environmental changes, and stage switching to ensure the adaptability and accuracy of the simulation.

[0087] Step S40: Output the energy consumption evolution trend under different functional adjustment scenarios through the dynamic energy consumption simulation and deduction framework to generate full-time energy consumption simulation results, and simultaneously output the corresponding energy consumption optimization scheme based on the full-time energy consumption simulation results.

[0088] It should be noted that, finally, through the dynamic energy consumption simulation and extrapolation framework, the energy consumption evolution trend under different functional adjustment scenarios (such as "replacing high-efficiency insulation components", "optimizing air conditioning operation strategies", "adjusting window lighting area") is simulated (such as energy consumption change curves for 1 year, 5 years, and 10 years after adjustment), generating full-time series energy consumption simulation results (covering energy consumption data at each time point of the building's entire life cycle); based on these results, the optimal energy consumption adjustment scheme is selected (such as the component replacement and operation strategy combination with the largest energy consumption reduction under the premise of meeting the building's functional requirements), ultimately completing the simulation and optimization of building energy consumption and achieving precise control of energy consumption throughout the entire life cycle.

[0089] Second Embodiment

[0090] Furthermore, the step of mining the cross-domain correlation between component functions and energy consumption parameters at different stages using the federated transfer learning algorithm includes:

[0091] Based on the semantic annotation and the physical sensing data, virtual sample sets for each simulation stage are generated through the digital twin. A cross-validation mechanism between federated nodes is introduced simultaneously. The virtual sample sets are dynamically calibrated based on real samples to output corresponding associated features.

[0092] Based on the aforementioned association features, a knowledge conflict detection algorithm is introduced to identify conflicts in the associated knowledge at each simulation stage, so as to output the corresponding initial association rules.

[0093] The engineering adaptation indicators of each simulation stage are detected, and the initial correlation rules are optimized in combination with the energy consumption control standards of different building types and climate regions to generate the corresponding cross-domain correlation rules.

[0094] It is important to note that, firstly, the sample data for federated learning nodes must balance the comprehensiveness of virtual simulation with the reliability of real data. Therefore, based on semantic annotation (clarifying component functions and attributes to ensure consistency of sample annotation) and physical sensor data (real data benchmark), different scenarios in each stage of design, construction, and operation and maintenance (such as component selection scenarios in the design stage, installation deviation scenarios in the construction stage, and aging scenarios in the operation and maintenance stage) are simulated using digital twins to generate virtual sample sets for each simulation stage (including component functional parameters, energy consumption parameters, environmental parameters, etc.). At the same time, a cross-validation mechanism between federated nodes is introduced (nodes at each stage mutually verify the rationality of samples). Using the real samples corresponding to the physical sensor data as a benchmark, the virtual sample sets are dynamically calibrated (such as correcting the deviation between the component insulation performance in the virtual simulation and the real data), and the associated features are output (such as core associated features such as component insulation performance-energy consumption, component installation accuracy-energy loss, etc.). Specifically, the calibrated virtual samples are fused with real samples to ensure the authenticity and comprehensiveness of the associated features.

[0095] Secondly, there may be conflicts in the related knowledge of each simulation stage (such as the knowledge of "component selection-energy consumption" in the design stage and the knowledge of "component aging-energy consumption" in the operation and maintenance stage). For example, the correlation logic between the energy consumption predicted in the design stage and the actual energy consumption in the operation and maintenance stage is inconsistent. Therefore, a knowledge conflict detection algorithm (such as a conflict detection algorithm based on rule reasoning) is introduced to identify conflicts in the related knowledge of each simulation stage (such as identifying the conflict between "the design stage believes that a certain component accounts for 10% of energy consumption" and "the actual proportion in the operation and maintenance stage is 20%). After eliminating conflicting knowledge, the initial correlation rules are output (such as the conflict-free rules such as "the thermal insulation performance of the component decreases → the energy consumption increases" and "the window is not airtight → the energy loss"). Specifically, the initial correlation rules lay the foundation for subsequent optimization and avoid the distortion of rules caused by knowledge conflicts.

[0096] Finally, the initial correlation patterns did not consider the actual adaptability of the project and industry control standards, and further optimization is needed. Therefore, the project adaptability indicators of each simulation stage were detected (such as the adaptability of component selection in the design stage, the adaptability of installation technology in the construction stage, and the adaptability of maintenance costs in the operation and maintenance stage). Combined with the energy consumption control standards of different building types (such as residential, commercial, and industrial buildings) and climate regions (such as frigid regions and hot-summer-cold-winter regions) (such as the national "General Specification for Building Energy Conservation and Renewable Energy Utilization"), the initial correlation patterns were optimized (for example, for frigid regions, the correlation weight of "external wall insulation component performance - heating energy consumption" was strengthened; for commercial buildings, the correlation logic of "air conditioning component operation - energy consumption" was optimized). Finally, a cross-domain correlation pattern was generated. Specifically, this pattern not only covers all stages of the entire life cycle, but also adapts to the actual project and control standards, providing a core basis for the construction of dynamic mapping relationships.

[0097] Furthermore, the step of optimizing the initial association pattern to generate the corresponding cross-domain association pattern includes:

[0098] The core semantic features in the semantic annotation are extracted and converted into hard constraints. Simultaneously, the fuzzy clustering-Bayesian inference fusion algorithm is combined to cross-match the rules in the initial association pattern to eliminate false association rules and generate intermediate association patterns.

[0099] Based on the temporal characteristics of each simulation stage, a temporal attention mechanism is introduced to capture the dynamic correlation between energy consumption parameters and component functions at different stages.

[0100] Based on the dynamic association features, the intermediate association rules are dynamically adjusted according to time-series parameters to generate the corresponding cross-domain association rules.

[0101] It should be noted that, firstly, to ensure that the association rules conform to the core functions and engineering constraints of the components, core semantic features (such as the core attributes of the components, such as "thermal insulation function," "load-bearing function," and "lighting function," which directly determine the energy consumption association logic) are extracted from the semantic annotations and converted into hard constraints (such as "the energy consumption association rules of thermal insulation components must meet the thermal insulation performance index constraints"). Simultaneously, there may be pseudo-association rules in the initial association rules (such as "a false association between component color and energy consumption," which has no actual causal relationship). Therefore, a fuzzy clustering-Bayesian inference fusion algorithm is used (fuzzy clustering classifies rules, and Bayesian inference judges the credibility of rules) to cross-match the rules in the initial association rules (such as matching the consistency of "component thermal insulation performance-energy consumption" rules at different stages), eliminating pseudo-association rules (such as rules with credibility below a preset threshold), and simultaneously generating intermediate association rules. Specifically, hard constraints ensure the compliance of the rules, and the elimination of pseudo-associations ensures the reliability of the rules.

[0102] The temporal characteristics of each stage of a building's entire life cycle differ significantly (e.g., the design stage is static planning, the construction stage is dynamic installation, and the operation and maintenance stage is long-term operation). The relationship between component function and energy consumption parameters changes dynamically over time (e.g., the impact of component installation accuracy on energy consumption is short-term during the construction stage, while the impact of component aging on energy consumption is long-term during the operation and maintenance stage). Therefore, based on the temporal characteristics of each simulation stage (e.g., the planning cycle in the design stage, the construction period in the construction stage, and the operating years in the operation and maintenance stage), a temporal attention mechanism is introduced (focusing on the core correlation characteristics of different temporal nodes) to capture the dynamic correlation characteristics between energy consumption parameters and component function at different stages (e.g., in the early stages of operation and maintenance, when component performance is good, the energy consumption correlation weight is low; in the later stages of operation and maintenance, when components age, the energy consumption correlation weight increases). Specifically, capturing dynamic correlation characteristics allows the patterns to adapt to temporal changes, avoiding the limitations of traditional static patterns.

[0103] Finally, based on the aforementioned dynamic association characteristics, the intermediate association rules are dynamically adjusted according to time-series parameters (such as adjusting the weight of association rules at different stages and correcting the association threshold of time-series nodes). For example, for the operation and maintenance stage, "component aging rate" is incorporated into the association rules as a time-series parameter, quantifying the dynamic association logic that "energy consumption increases by 8% for every 10% increase in aging rate". The adjusted rules cover all stages of the entire life cycle and adapt to the dynamic changes in time-series, ultimately generating cross-domain association rules. Specifically, these rules provide core support for the dynamic updating of dynamic mapping relationships, ensuring that the mapping relationships are synchronized with the temporal changes of the building.

[0104] Furthermore, the step of generating a corresponding dynamic energy consumption simulation and extrapolation framework based on the dynamic mapping relationship and the source tracing report includes:

[0105] The transmission links of each simulation stage in the source tracing report are parsed out, and the annotation results based on the semantic annotation are synchronously introduced to generate a collaborative decision-making mechanism of federated learning nodes to generate a dedicated inference unit adapted to the transmission links.

[0106] The virtual sensing data corresponding to the digital twin is extracted from the source tracing report, and the data flow interaction logic between each dedicated simulation unit is dynamically adjusted according to the virtual sensing data to match the corresponding energy consumption simulation requirements.

[0107] An initial simulation framework is created based on the energy consumption simulation requirements, and the initial simulation framework is dynamically modified to generate the dynamic energy consumption simulation framework.

[0108] It should be noted that, firstly, the source tracing report includes the energy consumption anomaly transmission links for each simulation stage (such as the transmission path of "component installation deviation → decreased insulation performance → increased energy consumption"). These links are the core logical basis for the simulation. Therefore, the transmission links for each simulation stage in the source tracing report are analyzed to clarify the core components, parameter relationships, and transmission directions in the links. Simultaneously, based on the semantic annotation results (clarifying the component identity and function), a collaborative decision-making mechanism of federated learning nodes is introduced (federated nodes at each stage jointly participate in the inference logic decision-making to ensure that the framework adapts to all stages), generating dedicated inference units adapted to the transmission links (such as "component installation deviation inference unit", "component aging energy consumption inference unit", and "air conditioning operation energy consumption inference unit"). Specifically, the dedicated inference units cover different transmission links in a targeted manner to improve the inference accuracy of the framework.

[0109] The dynamic energy consumption simulation framework needs to adapt to real-time data changes. Therefore, virtual sensing data corresponding to the digital twin (such as component temperature and energy consumption prediction values ​​simulated by the digital twin) is extracted from the source tracing report. This data can reflect the real-time status of the virtual simulation. The data flow interaction logic between each dedicated simulation unit is dynamically adjusted according to the virtual sensing data (e.g., when the virtual sensing data shows "abnormal external wall energy consumption", the data flow interaction between the "external wall component simulation unit" and the "overall energy consumption simulation unit" is strengthened, and the interaction between unrelated units is weakened), matching the corresponding energy consumption simulation requirements (e.g., the current focus is on simulating the impact of external wall components on overall energy consumption). Specifically, the dynamic adjustment of the data flow interaction logic allows the framework to adapt to real-time simulation requirements, avoiding the problem of rigid data flow in traditional frameworks.

[0110] Based on the matched energy consumption simulation requirements, an initial simulation framework is created (integrating dedicated simulation units, data flow interaction logic, and core simulation rules). The initial framework may have insufficient adaptability and simulation accuracy, so it is dynamically corrected (such as adjusting the parameter thresholds of the simulation units and optimizing the data flow interaction efficiency). Finally, a dynamic energy consumption simulation framework is generated. Specifically, this framework can be dynamically adjusted according to the transmission link, real-time data, and simulation requirements, providing core support for the output of energy consumption evolution trends and ensuring the dynamism and accuracy of the simulation.

[0111] Furthermore, the step of dynamically correcting the initial simulation framework to generate the corresponding dynamic energy consumption simulation framework includes:

[0112] Based on heterogeneous data from the entire building lifecycle, a knowledge graph of energy consumption influencing factors is constructed, and an attention mechanism is introduced simultaneously to uncover potential influencing factors that are not explicitly associated in the knowledge graph of energy consumption influencing factors.

[0113] The potential influencing factors are converted into corresponding feature parameters, and a corresponding set of inference rules is generated simultaneously based on the feature parameters.

[0114] The initial simulation framework is integrated with the simulation rule set to perform extreme scenario pre-run, and the corresponding performance bottleneck parameters generated during the run are extracted synchronously. The performance bottleneck parameters are then iteratively optimized through an adaptive genetic algorithm to generate the dynamic energy consumption simulation framework.

[0115] It is important to note that, firstly, a large amount of heterogeneous data exists throughout the building's entire lifecycle (such as design parameters, construction records, operation and maintenance data, climate data, and material data). The energy consumption influencing factors contained in this data (explicit factors such as component insulation performance, and implicit factors such as indirect climate effects) have not been fully explored. Therefore, based on this heterogeneous data, an energy consumption influencing factor knowledge graph is constructed (nodes represent influencing factors, and edges represent the relationships between factors, such as "climate temperature → component insulation performance → energy consumption"). Simultaneously, an attention mechanism is introduced to focus on highly correlated nodes in the knowledge graph and to uncover potential influencing factors that are not explicitly related (such as "material loss during construction → local performance degradation of components → implicit energy consumption loss," which is not directly reflected in the explicit data). Specifically, the discovery of potential influencing factors allows the framework to cover a more comprehensive range of energy consumption impact dimensions, improving the completeness of the simulation.

[0116] To integrate potential influencing factors into the framework's deduction logic, the identified potential influencing factors are transformed into corresponding characteristic parameters (such as transforming "material loss" into "loss rate parameter" and "hidden energy loss" into "energy loss coefficient"). Simultaneously, based on these characteristic parameters and combined with building energy consumption control standards and engineering realities, corresponding deduction rule sets are generated (such as the deduction rule that "for every 5% increase in the material loss rate, the energy loss coefficient increases by 3%"). Specifically, the deduction rule set provides core logical support for the revision of the initial framework, making up for the deficiencies in the initial framework's rules.

[0117] The initial simulation framework is integrated with the simulation rule set to conduct pre-runs in extreme scenarios (such as extreme cold / heat, severe component aging, and sudden failures, which easily expose the framework's performance bottlenecks). Performance bottleneck parameters generated during the pre-runs are extracted (such as insufficient simulation accuracy, excessive data processing latency, and excessive energy consumption prediction deviation under extreme scenarios). These performance bottleneck parameters are iteratively optimized using an adaptive genetic algorithm (with adaptive optimization capabilities and the ability to dynamically adjust optimization strategies) (such as optimizing the core weights of the simulation model, improving data processing efficiency, and correcting prediction deviation thresholds). After multiple iterations, the framework can adapt to both extreme and conventional scenarios, ultimately generating a dynamic energy consumption simulation framework. Specifically, this framework covers both explicit and potential influencing factors and adapts to extreme scenarios, ensuring the comprehensiveness and reliability of energy consumption simulation.

[0118] Furthermore, the step of outputting the energy consumption evolution trend under different functional adjustment scenarios through the dynamic energy consumption simulation and inference framework to generate full-time series energy consumption simulation results includes:

[0119] The scene feature thresholds of the building design, construction and operation and maintenance stages are extracted, and different functional adjustment scenarios are dynamically matched with each scene feature threshold to generate a scene inference input benchmark that is adapted to the whole life cycle.

[0120] The density clustering algorithm is used to distinguish the energy consumption evolution patterns under different functional adjustment scenarios, and the dynamic energy consumption simulation and inference framework is used to generate exclusive prediction sub-models that are adapted to each of the energy consumption evolution patterns.

[0121] The core parameters of the dedicated prediction sub-model are dynamically adjusted based on the cross-domain correlation rules to output the energy consumption evolution trend. Simultaneously, the various energy consumption evolution trends are integrated and processed to generate the full-time series energy consumption simulation results.

[0122] It should be noted that, firstly, the scenario characteristics of each stage of building design, construction, and operation and maintenance differ significantly (e.g., the scenario in the design stage is "component selection and layout", the scenario in the construction stage is "installation and commissioning", and the scenario in the operation and maintenance stage is "long-term operation and maintenance"). The energy consumption characteristic thresholds of different scenarios are different (e.g., the energy consumption budget threshold in the design stage, the temporary energy consumption threshold in the construction stage, and the normal energy consumption threshold in the operation and maintenance stage). Therefore, the scenario characteristic thresholds of each stage are extracted. Simultaneously, different functional adjustment scenarios (e.g., component replacement, operation strategy optimization, layout adjustment) are dynamically matched with the characteristic thresholds of each scenario (e.g., the scenario of "replacing high-efficiency insulation components" is matched with the normal energy consumption threshold of the operation and maintenance stage). This generates a scenario projection input benchmark that is adapted to the entire life cycle. Specifically, this benchmark ensures that the projection input of different functional adjustment scenarios conforms to the scenario characteristics of the corresponding stage, improving the adaptability of trend output.

[0123] The energy consumption evolution patterns differ under different functional adjustment scenarios (e.g., the energy consumption evolution pattern of the "component replacement" scenario is "short-term sharp drop followed by long-term stability", while the "operation strategy optimization" scenario is "continuous and slow decline"). Therefore, density clustering algorithms (such as the DBSCAN algorithm) are used to distinguish the energy consumption evolution patterns under different functional adjustment scenarios (scenarios with similar evolution trends are grouped into the same pattern). Simultaneously, through the generated dynamic energy consumption simulation and inference framework, a suitable exclusive prediction sub-model is built for each energy consumption evolution pattern (e.g., a "short drop-stability" prediction sub-model is built for the "short-term sharp drop followed by long-term stability" pattern, and a "linear decline" prediction sub-model is built for the "continuous and slow decline" pattern). Specifically, the exclusive prediction sub-model is adapted to different evolution patterns to improve the accuracy of trend prediction.

[0124] To ensure the reliability of energy consumption evolution trends, the core parameters of the dedicated prediction sub-model are dynamically adjusted based on the generated cross-domain correlation patterns (e.g., adjusting the energy consumption reduction parameters of the "component replacement" sub-model according to the correlation pattern of "component insulation performance-energy consumption"). Based on the adjusted dedicated prediction sub-model, the energy consumption evolution trends under various functional adjustment scenarios are output (e.g., the energy consumption change curve for the next 5 years under a certain scenario). Simultaneously, the various energy consumption evolution trends are integrated and processed (e.g., integrating trend curves from different scenarios and stages), ultimately generating full-time-series energy consumption simulation results (covering energy consumption data for all time nodes and functional adjustment scenarios throughout the building's entire life cycle). Specifically, this result provides core data support for the output of energy consumption optimization schemes.

[0125] Furthermore, the step of integrating the various energy consumption evolution trends to generate the full-time-series energy consumption simulation results includes:

[0126] A corresponding time-series correlation inference matrix is ​​created based on the changing pattern of the energy consumption evolution trend, and the corresponding key time nodes are calculated based on the time-series correlation inference matrix.

[0127] Based on the key time nodes, the energy consumption evolution trend is divided into several trend segments, and the temporal misalignment and connection deviation of each trend segment are corrected simultaneously through a graph neural network to generate initial integrated time series data.

[0128] Based on the source tracing report, the initial integrated time series data is subjected to iterative time series optimization processing to generate the corresponding full time series energy consumption simulation results.

[0129] It should be noted that, firstly, the changing patterns of various energy consumption evolution trends are temporally correlated (e.g., the end point of the energy consumption trend in the design phase is the start point of the construction phase, and the end point of the construction phase is the start point of the operation and maintenance phase). Therefore, based on the changing patterns of energy consumption evolution trends (e.g., the rate of increase / decrease of the trend, stable intervals, and abrupt change nodes), a corresponding temporal correlation inference matrix is ​​created (matrix elements quantify the temporal correlation strength and connection relationship of different trends). Based on this matrix, key time nodes are calculated (e.g., phase switching nodes, energy consumption abrupt change nodes, and trend stabilization nodes, such as the "construction completion → operation and maintenance start" node and the "component aging leading to energy consumption abrupt change" node). Specifically, key time nodes provide a temporal segmentation and connection basis for trend integration, avoiding temporal confusion.

[0130] Based on key time nodes, the energy consumption evolution trend is divided into several trend segments (e.g., the operation and maintenance phase trend is divided into "initial stable segment", "aging rising segment", and "post-maintenance declining segment"). There may be time sequence misalignment (e.g., timestamp deviation of a segment) and connection deviation (e.g., the energy consumption at the end of the previous segment is inconsistent with the energy consumption at the beginning of the next segment). Therefore, graph neural networks (e.g., GNN models) are used to correct these deviations (adjusting timestamps based on the time sequence correlation inference matrix and correcting energy consumption connection values) to generate initial integrated time series data. Specifically, the initial integrated time series data solves the connection problem of trend segments and lays the foundation for the generation of the final result.

[0131] The initial integrated time-series data may deviate from the actual energy consumption status (e.g., the root causes of energy consumption anomalies in the source tracing report are not considered). Therefore, based on the generated source tracing report (including information such as the root causes of energy consumption anomalies and component performance status), the initial integrated time-series data undergoes iterative time-series optimization processing (e.g., based on the information in the source tracing report that "energy consumption increases due to aging of a certain component", the energy consumption value of the corresponding time-series segment is corrected; based on the information of "energy consumption anomaly rectification", the trend segment after rectification is optimized). After multiple rounds of iterative optimization, it is ensured that the integrated time-series data not only conforms to the trend evolution law but also fits the actual energy consumption status, and finally generates a full-time-series energy consumption simulation result. Specifically, this result accurately covers the energy consumption changes throughout the entire life cycle of the building, providing accurate and reliable data support for the formulation of energy consumption optimization schemes, and ensuring the feasibility and effectiveness of the optimization schemes.

[0132] Please see Figure 2 The third embodiment of the present invention provides:

[0133] A BIM-based building energy consumption simulation and optimization system, wherein the system includes:

[0134] The acquisition module is used to perform component-level semantic annotation of the BIM model based on the two-way mapping relationship between the physical characteristics of building components and the digital twin mirror image, and to simultaneously acquire physical sensing data of building entity components.

[0135] The module is used to perform virtual simulation and reverse inference based on the semantic annotation and the physical sensing data through the digital twin to generate a corresponding traceability report. Simultaneously, it constructs distributed federated learning nodes based on the energy consumption data of the building design, construction and operation and maintenance stages, so as to mine the cross-domain correlation between component functions and energy consumption parameters at different stages through federated transfer learning algorithm.

[0136] The generation module is used to create a dynamic mapping relationship between building components and building energy consumption based on the cross-domain association rules, and simultaneously generate a corresponding dynamic energy consumption simulation and deduction framework based on the dynamic mapping relationship and the source tracing report.

[0137] The output module is used to output the energy consumption evolution trend under different functional adjustment scenarios through the dynamic energy consumption simulation and deduction framework, so as to generate full-time energy consumption simulation results and simultaneously output the corresponding energy consumption optimization scheme based on the full-time energy consumption simulation results.

[0138] Furthermore, the building module is specifically used for:

[0139] Based on the semantic annotation and the physical sensing data, virtual sample sets for each simulation stage are generated through the digital twin. A cross-validation mechanism between federated nodes is introduced simultaneously. The virtual sample sets are dynamically calibrated based on real samples to output corresponding associated features.

[0140] Based on the aforementioned association features, a knowledge conflict detection algorithm is introduced to identify conflicts in the associated knowledge at each simulation stage, so as to output the corresponding initial association rules.

[0141] The engineering adaptation indicators of each simulation stage are detected, and the initial correlation rules are optimized in combination with the energy consumption control standards of different building types and climate regions to generate the corresponding cross-domain correlation rules.

[0142] Furthermore, the building module is specifically used for:

[0143] The core semantic features in the semantic annotation are extracted and converted into hard constraints. Simultaneously, the fuzzy clustering-Bayesian inference fusion algorithm is combined to cross-match the rules in the initial association pattern to eliminate false association rules and generate intermediate association patterns.

[0144] Based on the temporal characteristics of each simulation stage, a temporal attention mechanism is introduced to capture the dynamic correlation between energy consumption parameters and component functions at different stages.

[0145] Based on the dynamic association features, the intermediate association rules are dynamically adjusted according to time-series parameters to generate the corresponding cross-domain association rules.

[0146] Furthermore, the generation module is specifically used for:

[0147] The transmission links of each simulation stage in the source tracing report are parsed out, and the annotation results based on the semantic annotation are synchronously introduced to generate a collaborative decision-making mechanism of federated learning nodes to generate a dedicated inference unit adapted to the transmission links.

[0148] The virtual sensing data corresponding to the digital twin is extracted from the source tracing report, and the data flow interaction logic between each dedicated simulation unit is dynamically adjusted according to the virtual sensing data to match the corresponding energy consumption simulation requirements.

[0149] An initial simulation framework is created based on the energy consumption simulation requirements, and the initial simulation framework is dynamically modified to generate the dynamic energy consumption simulation framework.

[0150] Furthermore, the generation module is specifically used for:

[0151] Based on heterogeneous data from the entire building lifecycle, a knowledge graph of energy consumption influencing factors is constructed, and an attention mechanism is introduced simultaneously to uncover potential influencing factors that are not explicitly associated in the knowledge graph of energy consumption influencing factors.

[0152] The potential influencing factors are converted into corresponding feature parameters, and a corresponding set of inference rules is generated simultaneously based on the feature parameters.

[0153] The initial simulation framework is integrated with the simulation rule set to perform extreme scenario pre-run, and the corresponding performance bottleneck parameters generated during the run are extracted synchronously. The performance bottleneck parameters are then iteratively optimized through an adaptive genetic algorithm to generate the dynamic energy consumption simulation framework.

[0154] Furthermore, the output module is specifically used for:

[0155] The scene feature thresholds of the building design, construction and operation and maintenance stages are extracted, and different functional adjustment scenarios are dynamically matched with each scene feature threshold to generate a scene inference input benchmark that is adapted to the whole life cycle.

[0156] The energy consumption evolution patterns under different functional adjustment scenarios are distinguished by density clustering algorithm, and exclusive prediction sub-models adapted to each of the energy consumption evolution patterns are generated simultaneously through the dynamic energy consumption simulation and inference framework.

[0157] The core parameters of the dedicated prediction sub-model are dynamically adjusted based on the cross-domain correlation rules to output the energy consumption evolution trend. Simultaneously, the various energy consumption evolution trends are integrated and processed to generate the full-time series energy consumption simulation results.

[0158] Furthermore, the output module is specifically used for:

[0159] A corresponding time-series correlation inference matrix is ​​created based on the changing pattern of the energy consumption evolution trend, and the corresponding key time nodes are calculated based on the time-series correlation inference matrix.

[0160] Based on the key time nodes, the energy consumption evolution trend is divided into several trend segments, and the temporal misalignment and connection deviation of each trend segment are corrected simultaneously through a graph neural network to generate initial integrated time series data.

[0161] Based on the source tracing report, the initial integrated time series data is subjected to iterative time series optimization processing to generate the corresponding full time series energy consumption simulation results.

[0162] The fourth embodiment of the present invention provides a computer, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the BIM-based building energy consumption simulation and optimization method as described above.

[0163] The fifth embodiment of the present invention provides a readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the BIM-based building energy consumption simulation and optimization method as described above.

[0164] In summary, the BIM-based building energy consumption simulation and optimization method and system provided in the above embodiments of the present invention can accurately create simulation mapping relationships, thereby improving optimization efficiency.

[0165] It should be noted that the above modules can be functional modules or program modules, and can be implemented through software or hardware. For modules implemented through hardware, the above modules can reside in the same processor; or the above modules can be located in different processors in any combination.

[0166] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0167] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0168] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0169] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0170] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A BIM-based method for simulating and optimizing building energy consumption, characterized in that, The method includes: Based on the bidirectional mapping relationship between the physical characteristics of building components and their digital twin images, a semantic segmentation algorithm is introduced to perform component-level semantic annotation on the BIM model, and physical sensing data of the building entity components are collected simultaneously. Based on the semantic annotation, a digital twin is used to perform virtual simulation and reverse inference based on the physical sensing data to generate a corresponding traceability report. Simultaneously, distributed federated learning nodes are constructed based on energy consumption data from the building design, construction, and operation and maintenance stages to mine cross-domain correlation patterns between component functions and energy consumption parameters at different stages through federated transfer learning algorithms. Based on the aforementioned cross-domain correlation rules, a dynamic mapping relationship between building components and building energy consumption is created, and a corresponding dynamic energy consumption simulation and deduction framework is generated simultaneously based on the dynamic mapping relationship and the source tracing report. The dynamic energy consumption simulation and deduction framework outputs the energy consumption evolution trend under different functional adjustment scenarios to generate full-time energy consumption simulation results, and simultaneously outputs the corresponding energy consumption optimization scheme based on the full-time energy consumption simulation results. The steps for mining the cross-domain correlation between component functions and energy consumption parameters at different stages using the federated transfer learning algorithm include: Based on the semantic annotation and the physical sensing data, virtual sample sets for each simulation stage are generated through the digital twin. A cross-validation mechanism between federated nodes is introduced simultaneously. The virtual sample sets are dynamically calibrated based on real samples to output corresponding associated features. Based on the aforementioned association features, a knowledge conflict detection algorithm is introduced to identify conflicts in the associated knowledge at each simulation stage, so as to output the corresponding initial association rules. The engineering adaptation indicators of each simulation stage are detected, and the initial correlation rules are optimized in combination with the energy consumption control standards of different building types and climate regions to generate the corresponding cross-domain correlation rules. The step of optimizing the initial association pattern to generate the corresponding cross-domain association pattern includes: The core semantic features in the semantic annotation are extracted and converted into hard constraints. Simultaneously, the fuzzy clustering-Bayesian inference fusion algorithm is combined to cross-match the rules in the initial association pattern to eliminate false association rules and generate intermediate association patterns. Based on the temporal characteristics of each simulation stage, a temporal attention mechanism is introduced to capture the dynamic correlation between energy consumption parameters and component functions at different stages. Based on the dynamic association features, the intermediate association rules are dynamically adjusted according to time-series parameters to generate the corresponding cross-domain association rules.

2. The BIM-based building energy consumption simulation and optimization method according to claim 1, characterized in that, The step of generating a corresponding dynamic energy consumption simulation and extrapolation framework based on the dynamic mapping relationship and the source tracing report includes: The transmission links of each simulation stage in the source tracing report are parsed out, and the annotation results based on the semantic annotation are synchronously introduced to generate a collaborative decision-making mechanism of federated learning nodes to generate a dedicated inference unit adapted to the transmission links. The virtual sensing data corresponding to the digital twin is extracted from the source tracing report, and the data flow interaction logic between each dedicated simulation unit is dynamically adjusted according to the virtual sensing data to match the corresponding energy consumption simulation requirements. An initial simulation framework is created based on the energy consumption simulation requirements, and the initial simulation framework is dynamically modified to generate the dynamic energy consumption simulation framework.

3. The BIM-based building energy consumption simulation and optimization method according to claim 2, characterized in that, The step of dynamically correcting the initial simulation framework to generate the corresponding dynamic energy consumption simulation framework includes: Based on heterogeneous data from the entire building lifecycle, a knowledge graph of energy consumption influencing factors is constructed, and an attention mechanism is introduced simultaneously to uncover potential influencing factors that are not explicitly associated in the knowledge graph of energy consumption influencing factors. The potential influencing factors are converted into corresponding feature parameters, and a corresponding set of inference rules is generated simultaneously based on the feature parameters. The initial simulation framework is integrated with the simulation rule set to perform extreme scenario pre-run, and the corresponding performance bottleneck parameters generated during the run are extracted synchronously. The performance bottleneck parameters are then iteratively optimized through an adaptive genetic algorithm to generate the dynamic energy consumption simulation framework.

4. The BIM-based building energy consumption simulation and optimization method according to claim 1, characterized in that, The step of outputting the energy consumption evolution trend under different functional adjustment scenarios through the dynamic energy consumption simulation and inference framework to generate full-time series energy consumption simulation results includes: The scene feature thresholds of the building design, construction and operation and maintenance stages are extracted, and different functional adjustment scenarios are dynamically matched with each scene feature threshold to generate a scene inference input benchmark that is adapted to the whole life cycle. The energy consumption evolution patterns under different functional adjustment scenarios are distinguished by density clustering algorithm, and exclusive prediction sub-models adapted to each of the energy consumption evolution patterns are generated simultaneously through the dynamic energy consumption simulation and inference framework. The core parameters of the dedicated prediction sub-model are dynamically adjusted based on the cross-domain correlation rules to output the energy consumption evolution trend. Simultaneously, the various energy consumption evolution trends are integrated and processed to generate the full-time series energy consumption simulation results.

5. The BIM-based building energy consumption simulation and optimization method according to claim 4, characterized in that, The step of integrating and processing the various energy consumption evolution trends to generate the full-time-series energy consumption simulation results includes: A corresponding time-series correlation inference matrix is ​​created based on the changing pattern of the energy consumption evolution trend, and the corresponding key time nodes are calculated based on the time-series correlation inference matrix. Based on the key time nodes, the energy consumption evolution trend is divided into several trend segments, and the temporal misalignment and connection deviation of each trend segment are corrected simultaneously through a graph neural network to generate initial integrated time series data. Based on the source tracing report, the initial integrated time series data is subjected to iterative time series optimization processing to generate the corresponding full time series energy consumption simulation results.

6. A BIM-based building energy consumption simulation and optimization system, characterized in that, The system is used to implement the BIM-based building energy consumption simulation and optimization method as described in any one of claims 1 to 5, the system comprising: The acquisition module is used to perform component-level semantic annotation of the BIM model based on the two-way mapping relationship between the physical characteristics of building components and the digital twin mirror image, and to simultaneously acquire physical sensing data of building entity components. The module is used to perform virtual simulation and reverse inference based on the semantic annotation and the physical sensing data through the digital twin to generate a corresponding traceability report. Simultaneously, it constructs distributed federated learning nodes based on the energy consumption data of the building design, construction and operation and maintenance stages, so as to mine the cross-domain correlation between component functions and energy consumption parameters at different stages through federated transfer learning algorithm. The generation module is used to create a dynamic mapping relationship between building components and building energy consumption based on the cross-domain association rules, and simultaneously generate a corresponding dynamic energy consumption simulation and deduction framework based on the dynamic mapping relationship and the source tracing report. The output module is used to output the energy consumption evolution trend under different functional adjustment scenarios through the dynamic energy consumption simulation and deduction framework, so as to generate full-time energy consumption simulation results and simultaneously output the corresponding energy consumption optimization scheme based on the full-time energy consumption simulation results.

7. A computer comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that, When the processor executes the computer program, it implements the BIM-based building energy consumption simulation and optimization method as described in any one of claims 1 to 5.

8. A readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the BIM-based building energy consumption simulation and optimization method as described in any one of claims 1 to 5.

Images