A low-carbon building monitoring system and method based on digital twins

Abstract

This invention discloses a low-carbon building monitoring system and method based on digital twins, belonging to the field of digital twin technology. The system includes a data acquisition unit, a computing power collaboration unit, a control execution unit, and a user interaction unit. The data acquisition unit performs targeted, timed monitoring of specific areas, establishes a unified semantic standard and dynamic knowledge graph, and improves the comprehensiveness and efficiency of data correlation analysis. The computing power collaboration unit constructs a three-level computing power collaboration system, integrates building monitoring information, and performs complex simulations of digital twin modules to accurately track energy transmission paths and equipment failure risks. Then, it performs predictive optimization and long-term control management, thereby establishing an integrated management framework for the entire life cycle of building clusters, encompassing the macro-geographical environment, meso-energy links, and micro-equipment internals. This enables cross-domain collaborative analysis, meets real-time requirements, and achieves multi-objective optimization design that balances energy costs, energy efficiency, and comfort, thus helping the building industry achieve its carbon neutrality goals.

Description

Technical Field

[0001] This invention relates to the field of digital twin technology, and in particular to a low-carbon building monitoring system and method based on digital twins. Background Technology

[0002] Digital twins are an innovative concept and method that integrates multiple cutting-edge technologies. They create a virtual mapping of a physical entity, process, or system in a digital way. This virtual mapping can synchronize the status, operation data, and environmental parameters of the physical object in real time, and predict and simulate the future development trend of the physical object through algorithmic models. Low-carbon building monitoring systems and methods based on digital twins have shown significant potential in promoting building energy efficiency and carbon emission reduction, but their practical application still faces multi-dimensional technical bottlenecks and systemic challenges.

[0003] Traditional systems suffer from data silos due to inconsistent semantic standards for multi-source data such as building structure, energy pipelines, and environmental monitoring. They lack an integrated management framework covering the entire lifecycle of building clusters, from the macro-geographical environment to the meso-energy links and the micro-equipment internals. This makes it difficult to accurately track energy transmission paths and equipment failure risks, hinders cross-domain collaborative analysis, and fails to meet real-time requirements. Furthermore, building energy consumption is affected by dynamic factors such as weather, user behavior, and equipment aging, leading to delayed responses to dynamic changes in the environment and behavior.

[0004] To address the aforementioned technical shortcomings, a solution is proposed. Summary of the Invention

[0005] The purpose of this invention is to solve the problem of "data silos" in existing technologies due to inconsistent semantic standards of multi-source data. It also addresses the lack of an integrated management framework covering the entire lifecycle of building clusters, including the macro-geographical environment, meso-energy links, and micro-equipment internals. This makes it difficult to accurately track energy transmission paths and equipment failure risks, achieve cross-domain collaborative analysis, and meet real-time requirements. Furthermore, building energy consumption is affected by dynamic factors such as weather, user behavior, and equipment aging, leading to a lag in response to dynamic changes in the environment and behavior.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A low-carbon building monitoring method based on digital twins includes the following steps:

[0008] Step 1: Collect building monitoring information: By setting the information collection cycle and marking key building nodes, the building monitoring information is monitored in a targeted and timed manner in a specific area. The building monitoring information includes building accessory monitoring parameters and environmental behavior monitoring parameters.

[0009] Step 2, integrate building monitoring information: construct a unified data model from multiple sources through semantic mapping and knowledge graphs, obtain standardized data from multiple sources, and share the data.

[0010] Step 3: Construct a digital twin model: Build a three-dimensional dynamic digital twin model using the monitoring parameters of building accessories, and update the data dynamically in real time;

[0011] Step 4: Establish a hybrid prediction model: Dynamically predict building energy consumption and equipment operating status by integrating standardized data from multiple sources, and carry out coordinated control and optimization of the building energy system to obtain low-carbon building regulation and optimization schemes and corresponding optimization control instructions.

[0012] Step 5: Receive optimization control commands: Real-time control of the building's energy equipment and environmental control equipment is achieved through optimization control commands, thereby enabling precise regional adjustment of the building's operating status.

[0013] Step Six: Visualized Query and Behavioral Interaction: Users can query building monitoring information and low-carbon building regulation and optimization schemes in real time through a visual interface, and input personalized behavioral preferences to dynamically adjust user behavior data.

[0014] Furthermore, the specific process for collecting building monitoring information is as follows:

[0015] Building accessory monitoring parameters include building structure data, energy pipeline network data, and equipment operation data;

[0016] Environmental behavior monitoring parameters include environmental weather data and user behavior data;

[0017] Building monitoring information is collected through multi-source sensors;

[0018] Building structure data includes basic geometric indicators, spatial relationship indicators, and complex shape indicators; energy network data includes network indicators and energy consumption indicators; equipment operation data includes operation indicators and performance indicators.

[0019] Environmental weather data includes environmental and meteorological indicators; user behavior data includes indicators of human activity and energy consumption preferences.

[0020] Furthermore, the specific process for integrating building monitoring information is as follows:

[0021] S2-1, The specific process for preprocessing building monitoring information is as follows:

[0022] Preprocessing building structure data and equipment operation data: missing values ​​are handled by linear interpolation of data nearest neighbors, and outliers are identified by the 3σ principle;

[0023] Preprocessing energy pipeline network data and environmental weather data: By using historical data and combining it with time series data, an LSTM neural network is built to obtain the predicted values ​​of the data. These predicted values ​​are then combined with the collected data values ​​to assess the data deviation. A threshold for the data deviation is then set and compared with the threshold to identify outliers.

[0024] Preprocessing user behavior data: By using historical data and combining it with time series data, an LSTM neural network is built to obtain and mark the reasonable range of user behavior data. When the detected value of the collected user behavior data exceeds the reasonable range, it is judged as an outlier.

[0025] For data outliers identified during preprocessing, manual verification is performed to re-detect and correct the outliers. Normalization is then performed to remove the dimensions of the parameter indicators.

[0026] S2-2, Establishing a unified semantic standard through analysis of building monitoring information:

[0027] By combining building monitoring information with building industry standards and internationally accepted data standards, a unified semantic vocabulary covering building accessory monitoring parameters and environmental behavior monitoring parameters will be developed.

[0028] S2-3, establish semantic mapping rules by analyzing the semantic differences of multi-source sensors:

[0029] Establish mapping rules from the semantics of raw sensor data to a unified semantic standard, thereby converting the data into standardized units.

[0030] S2-4, Set and label entities and relationships between entities in the knowledge graph, construct the knowledge graph and update it dynamically:

[0031] Knowledge graph entities include building structures, energy networks, equipment, environment, and users; relationships between entities include connection relationships, usage relationships, and influence relationships.

[0032] By using natural language processing (NLP) technology and machine learning algorithms, entities and relationships are extracted from the collected building monitoring information to construct a knowledge graph;

[0033] Establish a dynamic update mechanism for the knowledge graph and regularly perform consistency checks and optimizations on the knowledge graph;

[0034] S2-5, then spatiotemporal alignment is performed by aligning timestamps with unified geographic coordinates to obtain multi-source fusion standardized data.

[0035] Furthermore, the specific process of constructing a digital twin model is as follows:

[0036] Establish a multi-scale, full life-cycle management mechanism for building clusters, including macro, meso, and micro scales;

[0037] The macro-scale is constructed by integrating 3D geographic information system and building information model. Environmental weather data is input through 3D-GIS to build a macro-spatial framework of the building's geographical location and surrounding environment. Building structural data is input through BIM to mark the building structure zones.

[0038] At the mesoscale, digital links between equipment, pipelines, and energy consumption are constructed through energy pipeline network data. This connects building floor equipment with the energy pipeline network and marks all equipment and pipelines, thereby tracking energy transmission paths and calculating losses in real time.

[0039] At the microscale, the internal structure of the equipment is modeled and the equipment operation data is input to predict the probability of equipment failure, then the degree of equipment risk is determined and corresponding risk warning signals are generated.

[0040] Furthermore, the specific process for establishing a hybrid prediction model is as follows:

[0041] Dynamically predict building energy consumption and equipment operating status by integrating standardized data from multiple sources;

[0042] By analyzing building structure data, energy pipeline network data, and equipment operation data, dynamic space-energy coupling feature vectors can be obtained. ;

[0043] By analyzing environmental weather data and user behavior data, multi-scale time-series-behavioral feature fusion vectors can be obtained. ;

[0044] Dynamic space-energy coupling feature vector Multi-scale temporal-behavioral feature fusion vector Perform concatenation to obtain the concatenated feature vector. And time series prediction is performed using an LSTM deep learning model to obtain the predicted output vector. .

[0045] Furthermore, obtain the dynamic space-energy coupling feature vector. The specific process is as follows:

[0046] Dynamically predict building energy consumption and equipment operating status by integrating standardized data from multiple sources;

[0047] Building geometric features G are extracted from building structure data, and dynamic topological features T(t) are extracted from energy pipeline data and equipment operation data.

[0048] By performing linear regression fitting through historical data analysis, the weighting factors of the dynamic topological feature T(t) and the building geometric feature G are obtained and labeled as follows: , ;

[0049] Then, by weighting and synthesizing the dynamic topological feature T(t) with the building geometric feature G and its corresponding weighting factors, a dynamic space-energy coupling feature vector is obtained. .

[0050] Furthermore, obtain multi-scale temporal-behavioral feature fusion vectors. The specific process is as follows:

[0051] Environmental time-series features E(t) are extracted from environmental weather data, and user behavior features B(t) are extracted from user behavior data.

[0052] The behavior clustering center Ck is constructed through semantic mapping, the number of behavior patterns is labeled as K, any behavior pattern is labeled as k, and the behavior alignment function is obtained through knowledge graph and labeled as Align;

[0053] By calculating the variance of energy consumption fluctuations under different building behavior patterns, weights for each behavior pattern can be set. ;

[0054] The environmental temporal features E(t) are decomposed into multi-scale temporal features using the wavelet transform function WT, and combined with user behavior features B(t), behavior cluster centers Ck, behavior alignment function Align, and behavior pattern weights. This allows for the acquisition of multi-scale temporal-behavioral feature fusion vectors. .

[0055] Furthermore, the specific process for obtaining low-carbon building regulation and optimization schemes and corresponding optimization control instructions is as follows:

[0056] Set the equipment control command U as the action space to regulate the building's equipment;

[0057] The energy cost coefficient C(U) is obtained by combining the energy consumption of the equipment with the electricity unit price, and by combining the pipeline flow with the energy unit price; the energy efficiency index coefficient E(U) is obtained by combining the user comfort contribution per unit of energy consumption; and the comfort deviation coefficient D(U) is obtained by combining the deviation of the actual comfort data value from the comfort range.

[0058] Then, weighting coefficients are set for the energy cost coefficient C(U), the energy efficiency index coefficient E(U), and the comfort deviation coefficient D(U), thereby outputting a multi-objective optimization function;

[0059] The minimum value obtained by the multi-objective optimization function is the corresponding equipment control command U, which is the optimization control command of the low-carbon building regulation and optimization scheme.

[0060] A low-carbon building monitoring system based on digital twins includes a data acquisition unit, a computing power collaboration unit, a control execution unit, and a user interaction unit. The data acquisition unit, computing power collaboration unit, control execution unit, and user interaction unit are connected in communication. The system executes the aforementioned low-carbon building monitoring method based on digital twins.

[0061] The data acquisition unit is used to collect building monitoring information: by setting the information collection cycle and marking key building nodes, the building monitoring information can be monitored in a targeted and timed manner in a specific area; the building monitoring information includes building accessory monitoring parameters and environmental behavior monitoring parameters;

[0062] The computing power collaboration unit includes edge nodes, regional servers, and the cloud: the edge nodes are equipped with a multimodal fusion module, the regional servers are equipped with a digital twin module, and the cloud is equipped with a prediction and optimization module;

[0063] The multimodal fusion module is used to integrate building monitoring information: it constructs a unified data model from multiple sources through semantic mapping and knowledge graphs, obtains standardized multi-source fusion data, and performs data transmission and sharing.

[0064] The digital twin module is used to build digital twin models: it constructs a three-dimensional dynamic digital twin model by monitoring parameters of building accessories and updates the data dynamically in real time;

[0065] The prediction and optimization module is used to establish a hybrid prediction model: it dynamically predicts building energy consumption and equipment operating status through multi-source fusion standardized data, and performs coordinated control and optimization of building energy systems, thereby obtaining low-carbon building regulation and optimization schemes and corresponding optimization control instructions;

[0066] The control and execution unit is used to receive optimization control commands: through optimization control commands, the energy equipment and environmental control equipment of the building are controlled in real time, thereby making precise regional adjustments to the building's operating status;

[0067] The user interaction unit is used for visual query and behavioral interaction: users can query real-time building monitoring information and low-carbon building regulation and optimization schemes through a visual interface, and input personalized behavioral preferences to dynamically adjust user behavior data.

[0068] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0069] This invention enables targeted, timed monitoring of parameters such as building structure, energy pipelines, and user behavior in a specific region through multi-source devices in the data acquisition unit. By using unified semantic standards and dynamic knowledge graphs, it achieves deep integration of cross-domain data such as building structure stability analysis and wind speed data, thereby improving the comprehensiveness and efficiency of data correlation analysis.

[0070] This invention constructs a three-level computing power collaboration system through computing power collaboration units. The edge nodes are equipped with multimodal fusion modules to integrate building monitoring information, the regional servers are equipped with digital twin modules to perform complex simulations, accurately track energy transmission paths and equipment failure risks, and the cloud is equipped with a prediction and optimization module for long-term regulation and management. This establishes an integrated management framework for the entire life cycle of building clusters, encompassing the macro-geographical environment, meso-energy links, and micro-equipment internals. This enables cross-domain collaborative analysis, meets real-time requirements, and achieves multi-objective optimization design that balances energy costs, energy efficiency, and comfort.

[0071] This invention achieves precise regulation and low-carbon implementation by optimizing control commands, and realizes human-computer interaction and dynamic adaptation through a visual interface. It enables timely response to dynamic changes in the building environment and user behavior, and provides quantitative support for carbon footprint tracking and green building certification through full-cycle data-driven operation from building design to operation and maintenance, thus helping the construction industry achieve its carbon neutrality goals. Attached Figure Description

[0072] Figure 1 A schematic diagram of the steps in the method flow of the present invention is shown;

[0073] Figure 2 A schematic diagram of the system modules of the present invention is shown. Detailed Implementation

[0074] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0075] Example 1:

[0076] like Figures 1-2 As shown, a low-carbon building monitoring system based on digital twins includes a data acquisition unit, a computing power collaboration unit, a control and execution unit, and a user interaction unit. The data acquisition unit, computing power collaboration unit, control and execution unit, and user interaction unit are connected in communication. The computing power collaboration unit includes edge nodes, regional servers, and the cloud: the edge nodes are equipped with a multimodal fusion module, the regional servers are equipped with a digital twin module, and the cloud is equipped with a prediction and optimization module.

[0077] The work steps are as follows:

[0078] S1, Data acquisition unit collects building monitoring information: By setting the information collection cycle and marking key building nodes, the building monitoring information is monitored in a targeted and timed manner in a specific area;

[0079] Building monitoring information includes building accessory monitoring parameters and environmental behavior monitoring parameters;

[0080] Among them, the monitoring parameters for building accessories include building structure data, energy pipeline network data, and equipment operation data; the monitoring parameters for environmental behavior include environmental weather data and user behavior data.

[0081] Building monitoring information is collected through multi-source sensors. The specific process is as follows:

[0082] S1-1, the building structure data includes basic geometric indicators, spatial relationship indicators, and complex shape indicators; among them, basic geometric indicators include the dimensions, coordinate positions, orientation angles, and cross-sectional shapes of building components; spatial relationship indicators include the spacing, verticality, parallelism, and spatial volume of building components; complex shape indicators include the radius of curvature and arc length of curved surfaces; building structure data are collected using a total station, GPS monitoring instrument, and laser rangefinder;

[0083] S1-2, Energy pipeline network data includes pipeline network indicators and energy consumption indicators; among which, pipeline network indicators include pipeline flow and grid power; energy consumption indicators include electricity consumption, gas consumption and water consumption; energy pipeline network data are collected through flow meters, pressure gauges, electricity meters, water meters and gas meters;

[0084] S1-3, Equipment operation data includes operation indicators and performance indicators; among them, operation indicators include equipment start / stop status, running time, and fault alarm frequency; performance indicators include equipment motor speed and equipment power factor; equipment power factor includes equipment voltage, current, and power; equipment operation data are collected through limit switches and power monitoring sensors;

[0085] S1-4, Environmental weather data includes environmental indicators and meteorological indicators; among them, environmental indicators include light intensity and sunshine duration; meteorological indicators include temperature, humidity, wind speed, air pressure, and rainfall; environmental weather data are collected through the API interface of the weather station;

[0086] S1-5, User behavior data includes personnel activity indicators and energy consumption preference indicators; among them, personnel activity indicators include pedestrian flow, area stay time, and entrance and exit passage frequency; energy consumption preference indicators include lighting switch-on time, air conditioner set temperature and usage period; user behavior data is collected through infrared sensors, smart sockets, air conditioner thermostats and user interaction units.

[0087] S2, the multimodal fusion module integrates building monitoring information: it constructs a unified data model from multiple sources through semantic mapping and knowledge graphs, obtains standardized multi-source fusion data, and performs data transmission and sharing;

[0088] S2-1, The specific process for preprocessing building monitoring information is as follows:

[0089] Preprocessing building structure data and equipment operation data: missing values ​​are handled by linear interpolation of data nearest neighbors, and outliers are identified by the 3σ principle;

[0090] Preprocessing energy pipeline network data and environmental weather data: By using historical data and combining it with time series data, an LSTM neural network is built to obtain the predicted values ​​of the data. These predicted values ​​are then combined with the collected data values ​​to assess the data deviation. A threshold for the data deviation is then set and compared with the threshold to identify outliers.

[0091] Preprocessing user behavior data: By using historical data and combining it with time series data, an LSTM neural network is built to obtain and mark the reasonable range of user behavior data. When the detected value of the collected user behavior data exceeds the reasonable range, it is judged as an outlier.

[0092] For data outliers identified during preprocessing, manual verification is performed to re-detect and correct the outliers. Normalization is then performed to remove the dimensions of the parameter indicators.

[0093] The normalization methods include Z-Score standardization and maximum-minimum standardization.

[0094] S2-2, Establishing a unified semantic standard through analysis of building monitoring information:

[0095] By combining building monitoring information with building industry standards and internationally accepted data standards, a unified semantic vocabulary covering building accessory monitoring parameters and environmental behavior monitoring parameters will be developed.

[0096] Among them, building industry standards include BIM standards and energy management standards; internationally accepted data standards include JSON-LD and RDF data formats;

[0097] For example, for the "basic geometric indicators - dimensions" in building structure data, the data format is uniformly specified as "length unit / meter + value", and its unique identifier in the semantic system is clearly defined.

[0098] S2-3, establish semantic mapping rules by analyzing the semantic differences of multi-source sensors:

[0099] Establish mapping rules from the semantics of raw sensor data to a unified semantic standard, thereby converting the data into standardized units.

[0100] Taking "pipeline flow" in energy pipeline network data as an example, the data collected by different flow meters may have different units, such as cubic meters per hour or liters per minute. Through mapping rules, these units are uniformly converted into standard units and associated with the corresponding concepts in a unified semantic vocabulary.

[0101] S2-4, Set and label entities and relationships between entities in the knowledge graph, construct the knowledge graph and update it dynamically:

[0102] Knowledge graph entities include building structures, energy networks, equipment, environment, and users;

[0103] Relationships between entities include connection relationships, usage relationships, and influence relationships;

[0104] For example, there is a connection between the equipment and the energy network, a usage relationship between the user and the equipment, and an impact relationship between the environment and the building structure;

[0105] By using natural language processing (NLP) technology and machine learning algorithms, entities and relationships are extracted from the collected building monitoring information to construct a knowledge graph;

[0106] For structured data, such as equipment operation data, entities and relationships are extracted directly through data parsing; for unstructured data, such as user behavior data, NLP techniques such as named entity recognition and relationship extraction are used for processing, and the extracted information is added to the knowledge graph.

[0107] Establish a dynamic update mechanism for the knowledge graph. As building monitoring information is collected in real time and new data is added, the entities and relationships in the knowledge graph are automatically updated. The knowledge graph is regularly checked for consistency and optimized to ensure its accuracy and completeness.

[0108] S2-5, then spatiotemporal alignment is performed by aligning timestamps with unified geographic coordinates to obtain multi-source fusion standardized data;

[0109] It should be noted that by integrating semantics from multiple domains, including building structure, energy management, environmental monitoring, and user behavior, and by constructing unified semantic standards and mapping rules, the semantic barriers between different domains are broken down, and deep integration of multi-source data at the semantic level is achieved.

[0110] For example, combining the stability analysis of building structures with wind speed and rainfall data from environmental weather data, and using semantic fusion, can more accurately assess the impact of extreme weather on building structures and provide more comprehensive data support for building safety early warning.

[0111] By introducing a dynamic knowledge graph as the core framework for data integration, the complex relationships and dynamic changes between various elements in the building system are reflected in real time. Through knowledge extraction and graph update mechanisms, the knowledge graph is updated quickly when building equipment malfunctions, user behavior patterns change, or environmental parameters fluctuate, so that the data model always maintains an accurate description of the building's operating status.

[0112] This dynamism provides a more flexible and intelligent data foundation for intelligent building management. For example, real-time relationship analysis based on knowledge graphs can promptly detect the chain reaction of equipment failures on energy consumption and user experience, thereby enabling more accurate fault diagnosis and optimization decisions.

[0113] The multi-source unified data model built on knowledge graph has powerful multi-dimensional data correlation analysis capabilities. It can not only perform vertical analysis of the same type of data, such as historical trend analysis of equipment operation data, but also perform horizontal correlation analysis of different types of data, such as correlation analysis between user behavior data and energy consumption data.

[0114] By uncovering potential causal relationships and correlation patterns between data, we can provide deeper insights for building energy efficiency optimization, user service improvement, and facility maintenance management. For example, by analyzing the relationship between users' energy consumption preferences and environmental weather data at different times, we can optimize the operation strategy of air conditioning systems, thereby reducing energy consumption while meeting user comfort requirements.

[0115] S3, the digital twin module, is used to build digital twin models: it builds a three-dimensional dynamic digital twin model by monitoring parameters of building accessories and updates the data in real time.

[0116] A multi-scale, full life-cycle management mechanism is established for building clusters, including macro, meso, and micro scales. The specific process is as follows:

[0117] S3-1, the macro scale is constructed by integrating three-dimensional geographic information system (3D-GIS) and building information model (BIM). Environmental weather data is input through 3D-GIS to build a macro spatial framework of the building's geographical location and surrounding environment, and building structural data is input through BIM to mark the building structure zones.

[0118] S3-2, at the mesoscale, constructs a digital link between equipment, pipelines and energy consumption through energy pipeline data, connects building floor equipment with energy pipelines, and marks all equipment and pipelines, thereby tracking energy transmission paths and calculating losses in real time;

[0119] S3-3, at the micro scale, models the internal structure of the equipment, inputs equipment operating data to predict the probability of equipment failure, determines the degree of equipment risk, and generates corresponding risk warning signals.

[0120] S4, the prediction and optimization module establishes a hybrid prediction model: it dynamically predicts building energy consumption and equipment operating status through multi-source fusion of standardized data, and performs coordinated control and optimization of the building energy system, thereby obtaining low-carbon building regulation and optimization schemes and corresponding optimization control instructions;

[0121] S4-1 dynamically predicts building energy consumption and equipment operating status through multi-source fusion of standardized data;

[0122] Building geometric features G are extracted from building structure data. Building geometric features G includes basic geometric indicators and spatial relationship indicators. Building geometric features are used to analyze the long-term impact on energy distribution.

[0123] Dynamic topology features T(t) are extracted from energy pipeline network data and equipment operation data. Dynamic topology features T(t) include pipeline network indicators, energy consumption indicators, operation indicators and performance indicators. Dynamic topology features T(t) are used to reflect the energy transmission path in real time.

[0124] By performing linear regression fitting through historical data analysis, the weighting factors of the dynamic topological feature T(t) and the building geometric feature G are obtained and labeled as follows: , ;

[0125] Then, by weighting and synthesizing the dynamic topological feature T(t) with the building geometric feature G and its corresponding weighting factors, a dynamic space-energy coupling feature vector is obtained. : .

[0126] S4-2, extract environmental time series features E(t) from environmental weather data. Environmental time series features E(t) include environmental indicators and meteorological indicators. Environmental time series features E(t) are real-time fluctuation data.

[0127] User behavior features B(t) are extracted from user behavior data. User behavior features B(t) include energy preference indicators and are periodic pattern data.

[0128] Historical behaviors within a building are classified using semantic mapping, and a behavior clustering center Ck is constructed to distinguish the building's behavior patterns, including office mode and residential mode. The number of behavior patterns is labeled K, and any behavior pattern is labeled k.

[0129] The behavior alignment function is obtained by using a knowledge graph and marked as Align. This function is used to associate and match environmental fluctuations with similar behavior patterns. For example, it can link temperature fluctuations in office mode with air conditioning equipment.

[0130] By calculating the variance of energy consumption fluctuations under different building behavior patterns, weights for each behavior pattern can be set. And assign values: the higher the variance of energy consumption fluctuation, the higher the weight assigned to this behavior pattern;

[0131] The environmental temporal features E(t) are decomposed into multi-scale temporal features using the wavelet transform function WT, and combined with user behavior features B(t), behavior cluster centers Ck, behavior alignment function Align, and behavior pattern weights. This allows for the acquisition of multi-scale temporal-behavioral feature fusion vectors. :

[0132] .

[0133] S4-3, dynamic space-energy coupling feature vector Multi-scale temporal-behavioral feature fusion vector Perform concatenation to obtain the concatenated feature vector. And time series prediction is performed using an LSTM deep learning model to obtain the predicted output vector. :

[0134] ;

[0135] By combining static physical properties and dynamic environmental behavior, the energy consumption of buildings can be dynamically predicted, enabling the assessment of building energy consumption levels and equipment operating status.

[0136] S4-4, then conduct coordinated control and optimization of the building energy system to obtain low-carbon building regulation and optimization schemes and corresponding optimization control instructions;

[0137] Set the equipment control command U as the action space to regulate the building's equipment, such as air conditioning equipment and lighting equipment;

[0138] The energy cost coefficient C(U) is obtained by combining the energy consumption of the equipment with the electricity unit price, and by combining the pipeline flow with the energy unit price.

[0139] The energy efficiency index coefficient E(U) is obtained by measuring the user comfort contribution per unit of energy consumption.

[0140] Among them, user comfort refers to the duration of use of air conditioning and lighting within the comfort range. The center value of the comfort range is set by the average value of temperature, humidity and brightness perceived by the user, and then the floating range is preset to obtain the comfort range of each device.

[0141] The comfort deviation coefficient D(U) is obtained by measuring the deviation between the actual comfort data value and the comfort range. For example, the temperature deviation is obtained by calculating the deviation difference between the actual temperature value and the temperature comfort range. Similarly, the deviation of humidity and brightness is obtained, and the comfort deviation coefficient D(U) is obtained by weighted summation.

[0142] Then, weighting coefficients are set for the energy cost coefficient C(U), the energy efficiency index coefficient E(U), and the comfort deviation coefficient D(U), thereby outputting a multi-objective optimization function:

[0143] ;

[0144] Wherein, λ1, λ2, and λ3 are the weighting coefficients of energy cost coefficient C(U), energy efficiency index coefficient E(U), and comfort deviation coefficient D(U), respectively. When the energy cost coefficient C(U) is lower, the energy efficiency index coefficient E(U) is higher, and the comfort deviation coefficient D(U) is lower, the multi-objective optimization function is lower.

[0145] S4-5, when the multi-objective optimization function takes the minimum value, the corresponding equipment control command U is the optimization control command of the low-carbon building regulation and optimization scheme;

[0146] In addition, to ensure the feasibility of the equipment control command U, collaborative control constraints are set, including upper and lower limits of equipment physical parameters, minimum comfort constraints, and grid power limits. This enables multi-objective performance optimization under the conditions of normal equipment operation, normal grid operation, and user comfort, thereby balancing comfort, energy efficiency, and cost.

[0147] By using digital twin models to divide the data into macro, meso, and micro scales, multi-objective optimization can be performed at different scales. Corresponding data indicators can be selected for analysis to generate low-carbon building regulation and optimization schemes for each scale.

[0148] S5, the control and execution unit receives optimization control commands: through optimization control commands, the energy equipment and environmental control equipment of the building are controlled in real time, thereby making precise regional adjustments to the building's operating status and achieving the target of low-carbon building.

[0149] S6, the user interaction unit performs visual queries and behavioral interactions: users can query real-time building monitoring information and low-carbon building regulation and optimization schemes through a visual interface, and input personalized behavioral preferences to dynamically adjust user behavior data.

[0150] In summary, this invention enables targeted and timely monitoring of parameters such as building structure, energy pipeline, and user behavior in a specific region through multi-source devices in the data acquisition unit. By using unified semantic standards and dynamic knowledge graphs, it achieves deep integration of cross-domain data such as building structure stability analysis and wind speed data, thereby improving the comprehensiveness and efficiency of data correlation analysis.

[0151] This invention constructs a three-level computing power collaboration system through computing power collaboration units. The edge nodes are equipped with multimodal fusion modules to integrate building monitoring information, the regional servers are equipped with digital twin modules to perform complex simulations, accurately track energy transmission paths and equipment failure risks, and the cloud is equipped with a prediction and optimization module for long-term regulation and management. This establishes an integrated management framework for the entire life cycle of building clusters, encompassing the macro-geographical environment, meso-energy links, and micro-equipment internals. This enables cross-domain collaborative analysis, meets real-time requirements, and achieves multi-objective optimization design that balances energy costs, energy efficiency, and comfort.

[0152] This invention achieves precise regulation and low-carbon implementation by optimizing control commands, and realizes human-computer interaction and dynamic adaptation through a visual interface. It enables timely response to dynamic changes in the building environment and user behavior, and provides quantitative support for carbon footprint tracking and green building certification through full-cycle data-driven operation from building design to operation and maintenance, thus helping the construction industry achieve its carbon neutrality goals.

[0153] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0154] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0155] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A low-carbon building monitoring method based on digital twins, characterized in that: Includes the following steps: Step 1: Collect building monitoring information: By setting the information collection cycle and marking key building nodes, the building monitoring information is monitored in a targeted and timed manner in a specific area. The building monitoring information includes building accessory monitoring parameters and environmental behavior monitoring parameters. Step 2, integrate building monitoring information: construct a unified data model from multiple sources through semantic mapping and knowledge graphs, obtain standardized data from multiple sources, and share the data. Step 3: Construct a digital twin model: Build a three-dimensional dynamic digital twin model using the monitoring parameters of building accessories, and update the data dynamically in real time; Step 4: Establish a hybrid prediction model: Dynamically predict building energy consumption and equipment operating status by integrating standardized data from multiple sources, and carry out coordinated control and optimization of the building energy system to obtain low-carbon building regulation and optimization schemes and corresponding optimization control instructions. The specific process of establishing a hybrid prediction model is as follows: Dynamically predict building energy consumption and equipment operating status by integrating standardized data from multiple sources; By analyzing building structure data, energy pipeline network data, and equipment operation data, dynamic space-energy coupling feature vectors can be obtained. Obtain the dynamic space-energy coupling feature vector The specific process is as follows: Dynamically predict building energy consumption and equipment operating status by integrating standardized data from multiple sources; Building geometric features G are extracted from building structure data, and dynamic topological features T(t) are extracted from energy pipeline data and equipment operation data. By performing linear regression fitting through historical data analysis, the weighting factors of the dynamic topological feature T(t) and the building geometric feature G are obtained and labeled as follows: , ; Then, by weighting and synthesizing the dynamic topological feature T(t) with the building geometric feature G and its corresponding weighting factors, a dynamic space-energy coupling feature vector is obtained. : ; By analyzing environmental weather data and user behavior data, multi-scale time-series-behavioral feature fusion vectors can be obtained. Obtain multi-scale temporal-behavioral feature fusion vectors The specific process is as follows: Environmental time-series features E(t) are extracted from environmental weather data, and user behavior features B(t) are extracted from user behavior data. The behavior clustering center Ck is constructed through semantic mapping, the number of behavior patterns is labeled as K, any behavior pattern is labeled as k, and the behavior alignment function is obtained through knowledge graph and labeled as Align; By calculating the variance of energy consumption fluctuations under different building behavior patterns, weights for each behavior pattern can be set. ; The environmental temporal features E(t) are decomposed into multi-scale temporal features using the wavelet transform function WT, and combined with user behavior features B(t), behavior cluster centers Ck, behavior alignment function Align, and behavior pattern weights. This allows for the acquisition of multi-scale temporal-behavioral feature fusion vectors. : ； Then, the dynamic space-energy coupling feature vector Multi-scale temporal-behavioral feature fusion vector Perform concatenation to obtain the concatenated feature vector. And time series prediction is performed using an LSTM deep learning model to obtain the predicted output vector. : ； Then, low-carbon building regulation and optimization schemes and corresponding optimization control instructions are obtained. The specific process is as follows: Set the equipment control command U as the action space to regulate the building's equipment; The energy cost coefficient C(U) is obtained by combining the energy consumption of the equipment with the electricity unit price, and by combining the pipeline flow with the energy unit price; the energy efficiency index coefficient E(U) is obtained by combining the user comfort contribution per unit of energy consumption; and the comfort deviation coefficient D(U) is obtained by combining the deviation of the actual comfort data value from the comfort range. By setting weighting coefficients for the energy cost coefficient C(U), energy efficiency coefficient E(U), and comfort deviation coefficient D(U), a multi-objective optimization function is output: ; Wherein, λ1, λ2, and λ3 are the weighting coefficients of the energy cost coefficient C(U), the energy efficiency index coefficient E(U), and the comfort deviation coefficient D(U), respectively; The minimum value is obtained by optimizing the multi-objective function, and the corresponding equipment control command U is the optimized control command of the low-carbon building regulation and optimization scheme. Step 5: Receive optimization control commands: Real-time control of the building's energy equipment and environmental control equipment is achieved through optimization control commands, thereby enabling precise regional adjustment of the building's operating status. Step Six: Visualized Query and Behavioral Interaction: Users can query building monitoring information and low-carbon building regulation and optimization schemes in real time through a visual interface, and input personalized behavioral preferences to dynamically adjust user behavior data.

2. The low-carbon building monitoring method based on digital twins according to claim 1, characterized in that: The specific process for collecting building monitoring information is as follows: Building accessory monitoring parameters include building structure data, energy pipeline network data, and equipment operation data; Environmental behavior monitoring parameters include environmental weather data and user behavior data; Building monitoring information is collected through multi-source sensors; Building structure data includes basic geometric indicators, spatial relationship indicators, and complex shape indicators; energy network data includes network indicators and energy consumption indicators; equipment operation data includes operation indicators and performance indicators. Environmental weather data includes environmental and meteorological indicators; user behavior data includes indicators of human activity and energy consumption preferences.

3. The low-carbon building monitoring method based on digital twins according to claim 2, characterized in that: The specific process of integrating building monitoring information is as follows: S2-1, The specific process for preprocessing building monitoring information is as follows: Preprocessing building structure data and equipment operation data: missing values ​​are handled by linear interpolation of data nearest neighbors, and outliers are identified by the 3σ principle; Preprocessing energy pipeline network data and environmental weather data: By using historical data and combining it with time series data, an LSTM neural network is built to obtain the predicted values ​​of the data. These predicted values ​​are then combined with the collected data values ​​to assess the data deviation. A threshold for the data deviation is then set and compared with the threshold to identify outliers. Preprocessing user behavior data: By using historical data and combining it with time series data, an LSTM neural network is built to obtain and mark the reasonable range of user behavior data. When the detected value of the collected user behavior data exceeds the reasonable range, it is judged as an outlier. For data outliers identified during preprocessing, manual verification is performed to re-detect and correct the outliers. Normalization is then performed to remove the dimensions of the parameter indicators. S2-2, Establishing a unified semantic standard through analysis of building monitoring information: By combining building monitoring information with building industry standards and internationally accepted data standards, a unified semantic vocabulary covering building accessory monitoring parameters and environmental behavior monitoring parameters will be developed. S2-3, establish semantic mapping rules by analyzing the semantic differences of multi-source sensors: Establish mapping rules from the semantics of raw sensor data to a unified semantic standard, thereby converting the data into standardized units. S2-4, Set and label entities and relationships between entities in the knowledge graph, construct the knowledge graph and update it dynamically: Knowledge graph entities include building structures, energy networks, equipment, environment, and users; relationships between entities include connection relationships, usage relationships, and influence relationships. By using natural language processing (NLP) technology and machine learning algorithms, entities and relationships are extracted from the collected building monitoring information to construct a knowledge graph; Establish a dynamic update mechanism for the knowledge graph and regularly perform consistency checks and optimizations on the knowledge graph; S2-5, then spatiotemporal alignment is performed by aligning timestamps with unified geographic coordinates to obtain multi-source fusion standardized data.

4. The low-carbon building monitoring method based on digital twins according to claim 3, characterized in that: The specific process of constructing a digital twin model is as follows: Establish a multi-scale, full life-cycle management mechanism for building clusters, including macro, meso, and micro scales; The macro-scale is constructed by integrating 3D geographic information system and building information model. Environmental weather data is input through 3D-GIS to build a macro-spatial framework of the building's geographical location and surrounding environment. Building structural data is input through BIM to mark the building structure zones. At the mesoscale, digital links between equipment, pipelines, and energy consumption are constructed through energy pipeline network data. This connects building floor equipment with the energy pipeline network and marks all equipment and pipelines, thereby tracking energy transmission paths and calculating losses in real time. At the microscale, the internal structure of the equipment is modeled and the equipment operation data is input to predict the probability of equipment failure, then the degree of equipment risk is determined and corresponding risk warning signals are generated.

5. A low-carbon building monitoring system based on digital twins, characterized in that: The system includes a data acquisition unit, a computing power coordination unit, a control and execution unit, and a user interaction unit. The data acquisition unit, the computing power coordination unit, the control and execution unit, and the user interaction unit are connected in communication. The system executes the low-carbon building monitoring method based on digital twin as described in any one of claims 1-4. The data acquisition unit is used to collect building monitoring information: by setting the information collection cycle and marking key building nodes, the building monitoring information can be monitored in a targeted and timed manner in a specific area. Building monitoring information includes building accessory monitoring parameters and environmental behavior monitoring parameters; The computing power collaboration unit includes edge nodes, regional servers, and the cloud: the edge nodes are equipped with a multimodal fusion module, the regional servers are equipped with a digital twin module, and the cloud is equipped with a prediction and optimization module; The multimodal fusion module is used to integrate building monitoring information: it constructs a unified data model from multiple sources through semantic mapping and knowledge graphs, obtains standardized multi-source fusion data, and performs data transmission and sharing. The digital twin module is used to build digital twin models: it constructs a three-dimensional dynamic digital twin model by monitoring parameters of building accessories and updates the data dynamically in real time; The prediction and optimization module is used to establish a hybrid prediction model: it dynamically predicts building energy consumption and equipment operating status through multi-source fusion standardized data, and performs coordinated control and optimization of building energy systems, thereby obtaining low-carbon building regulation and optimization schemes and corresponding optimization control instructions; The control and execution unit is used to receive optimization control commands: through optimization control commands, the energy equipment and environmental control equipment of the building are controlled in real time, thereby making precise regional adjustments to the building's operating status; The user interaction unit is used for visual query and behavioral interaction: users can query real-time building monitoring information and low-carbon building regulation and optimization schemes through a visual interface, and input personalized behavioral preferences to dynamically adjust user behavior data.

Images