Dynamic fusion system and method for digital twin energy management BIM modeling

Abstract

The invention provides a dynamic fusion system for digital twin energy management BIM modeling and a method thereof, which are used for improving the intelligent level of energy management and optimizing energy scheduling. The system comprises a data sensing node used for collecting heterogeneous operation data of energy facilities; the local twin processor is used for carrying out real-time digital twin mapping on the data and generating a dynamic feature vector; region twinborn nodes perform local optimization based on edge calculation, and realize decentralized collaboration through a block chain or a knowledge graph; the cloud BIM fusion engine constructs a digital twin BIM model, generates a self-optimization energy scheduling scheme and predicts a long-term trend; and the operation control terminal is used for presenting a real-time view and providing a regulation and control instruction. According to the method, technologies such as multi-source data fusion, streaming aggregation optimization and adaptive twin mapping are combined, and the accuracy and dynamic response capability of energy management are improved.

Description

Technical Field

[0001] The present invention relates to the fields of digital twin technology, building information modeling (BIM) technology, and energy management. Specifically, it relates to a dynamic fusion system and method for digital twin-based energy management BIM modeling, aiming to achieve efficient management, optimized scheduling, and intelligent decision-making of energy facilities through multi-source data fusion, edge computing, decentralized collaborative mechanisms, and intelligent optimization algorithms, improve energy utilization efficiency, reduce operating costs, and enhance the real-time perception and control capabilities of the energy system. A dynamic fusion system and method for digital twin energy management BIM modeling Background Art

[0002] With the development of intelligent energy management and building information modeling BIM (Building Information Modeling) technology, the application of digital twin (Digital Twin) technology in the field of energy management has gradually attracted attention. Traditional energy management systems mainly rely on static BIM models and preset optimization rules, making it difficult to perceive the operating status of energy facilities in real time and unable to dynamically adjust optimization strategies, resulting in low energy utilization efficiency and high operation and maintenance costs. In addition, existing energy management methods often rely on centralized computing architectures, and there are delays in the data transmission and processing processes, making it difficult to meet the real-time optimization requirements of complex energy systems.

[0003] In recent years, the rapid development of multi-source data acquisition technology, edge computing, blockchain, and artificial intelligence optimization algorithms based on the Internet of Things (IoT) has provided new technical means for energy management. However, the existing technologies still have the following problems: (1) The data acquisition and fusion methods are single, making it difficult to comprehensively and accurately perceive the dynamic operating status of energy facilities; (2) There is a lack of efficient local computing and collaborative optimization mechanisms, resulting in a lag in energy scheduling response; (3) Centralized architectures have problems such as high computing pressure and low data security, making it difficult to adapt to the distributed management requirements of large-scale energy systems.

[0004] To solve the above problems, the present invention proposes a dynamic fusion system for digital twin-based energy management BIM modeling, which realizes real-time perception, dynamic optimization, and intelligent decision-making of energy management through data perception, edge computing, decentralized collaborative optimization, and cloud intelligent analysis, improves energy utilization rate, reduces energy consumption costs, and enhances the robustness and adaptability of the system. Summary of the Invention

[0005] In summary, the present invention provides a dynamic fusion system and method for digital twin energy management BIM modeling, aiming to solve problems such as incomplete data acquisition, lagging optimization response, and single computing architecture in existing energy management systems, and realize the efficient and intelligent management of energy facilities.

[0006] According to the data perception nodes provided by the present inventor, they are deployed around energy facilities and configured with a multi-source sensor array for real-time collection of heterogeneous operation data of the energy facilities and output; a local twin processor, coupled with the data perception nodes, for receiving the heterogeneous operation data and performing real-time digital twin mapping to generate dynamic feature vectors; a regional twin node, interconnected with multiple local twin processors, for performing local energy optimization based on edge computing and collaborating with other regional twin nodes through a decentralized collaboration mechanism, where the decentralized collaboration mechanism uses blockchain or knowledge graph to achieve distributed decision-making; a cloud BIM fusion engine, interconnected with the regional twin node, for receiving the data transmitted by the regional twin node, constructing a digital twin BIM model, generating a self-optimizing energy scheduling plan and long-term trend prediction, and transmitting the results back to the regional twin node; an operation control terminal, interconnected with the cloud BIM fusion engine, for presenting a real-time view of the digital twin BIM model, energy anomaly alerts and scheduling suggestions, and supporting users to issue control commands.

[0007] Optionally, the data perception node performs primary feature extraction on the collected heterogeneous operation data to generate an energy state vector, where the energy state vector is calculated by the following formula:

[0008] where (Q1, Q2, …, Q m ) are the components of the heterogeneous operation data collected by the multi-source sensor array, and (k1, k2, …, k m ) are the corresponding dynamic calibration coefficients, and the dynamic calibration coefficients are adaptively updated based on the operation mode of the energy facilities and satisfy the normalization condition.

[0009] Optionally, the local twin processor is built-in with an adaptive twin mapping algorithm for generating the dynamic feature vectors and adopts a self-organizing hierarchical strategy to determine data classification according to the dynamic feature vectors, where key data is transmitted through a priority channel and regular data is transmitted through a secondary channel after compression.

[0010] Optionally, the system further includes a streaming aggregation relay, connected to multiple local twin processors through a distributed network, for receiving the compressed regular data transmitted by the local twin processors through the secondary channel, performing streaming clustering on the compressed regular data and forwarding it to the regional twin node, while the key data is directly transmitted by the local twin processor to the regional twin node through the priority channel.

[0011] Optionally, the streaming aggregation relay performs semantic layering and transmission priority evaluation on the compressed conventional data received, immediately forwards the high-correlation data to the regional twin node, and locally buffers the low-correlation data simultaneously to reduce network load.

[0012] Optionally, the streaming aggregation relay constructs an adaptive feedback network with multiple local twin processors according to the local optimization results generated by the regional twin node, and supports hierarchical pushing of the local optimization results to the corresponding local twin processors according to the real-time requirements of the energy facilities.

[0013] Optionally, the cloud BIM fusion engine calculates the energy dynamic index D based on the fusion data and generates the long-term trend prediction accordingly, where the energy dynamic index D is calculated by the following formula:

[0014] wherein, R is the real-time response value of the heterogeneous operation data, R0 is the reference response value, T is the current model deduction value, T0 is the benchmark deduction value, α is the sensitivity factor, and the sensitivity factor α is adjusted in the range of [0.8, 2.5] according to the type of the energy facilities.

[0015] Optionally, the system further includes a real-time tuning unit, which is coupled with the local twin processor, and dynamically adjusts the operation parameters of the energy facilities according to the local optimization results of the regional twin node to achieve the optimal balance of energy utilization rate.

[0016] Optionally, the cloud BIM fusion engine sends parameter calibration instructions or model update instructions to the data perception node through the regional twin node to support the online performance improvement of the energy facilities and reduce operation losses.

[0017] According to a dynamic fusion method for digital twin energy management BIM modeling provided by the present invention, the method includes the following steps: (1) Through data perception nodes deployed around the energy facilities, a multi-source sensor array is used to collect the heterogeneous operation data of the energy facilities in real time and output it; (2) The local twin processor is used to receive the heterogeneous operation data, and an adaptive twin mapping algorithm is adopted for real-time digital twin mapping to generate a dynamic feature vector; (3) The regional twin node is used to receive the data transmitted by the local twin processor, perform local energy optimization based on edge computing, and cooperate with other regional twin nodes through a decentralized cooperation mechanism, where the decentralized cooperation mechanism adopts a blockchain or a knowledge graph to implement distributed decision-making; (4) Utilize the cloud BIM fusion engine to receive the data transmitted by the regional twin node, construct a digital twin BIM model using a multi-scale dynamic fusion model, and conduct a global energy status deduction to generate a self-optimizing energy scheduling plan and long-term trend prediction; (5) Transmit the self-optimizing energy scheduling plan and long-term trend prediction generated by the cloud BIM fusion engine back to the regional twin node; (6) Receive the self-optimizing energy scheduling plan and long-term trend prediction transmitted back by the cloud BIM fusion engine through an operation control terminal, present a real-time view of the digital twin BIM model, energy anomaly alerts, and scheduling suggestions, and support the user in issuing control commands to the cloud BIM fusion engine.

[0018] Through digital twin technology, edge computing, and decentralized collaborative optimization, the present invention improves the intelligence level of energy management, realizes real-time data fusion, dynamic optimization scheduling, and adaptive feedback control, is applicable to the refined management of large energy systems, and has high application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a schematic diagram of the dynamic fusion system architecture for BIM modeling of a digital twin energy management system provided by the present invention.

[0020] Figure 2 It is a flowchart of the dynamic fusion method for BIM modeling of a digital twin energy management system provided by the present invention.

[0021] Figure 3 It is a schematic diagram of a streaming aggregation relay provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] To make the objectives, technical solutions, and advantages of the present application clearer, the technical solutions in the present application will be clearly and completely described below with reference to the accompanying drawings in the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments in the present application belong to the scope of protection of the present application.

[0023] The terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features; in the description of the present application, unless otherwise stated, "a plurality" means two or more.

[0024] To achieve the above objectives, please refer to Figure 1 andFigure 2 As shown, the present invention provides a dynamic fusion system for digital twin energy management BIM modeling, which can be applied to various scenarios. For example, in the intelligent building energy management scenario, data perception nodes are deployed in the heating, ventilation, air conditioning (HVAC) system, lighting system and power distribution system of the building. A multi-source sensor array is configured, including temperature sensors, current sensors and light sensors, which collect data once per second to monitor the energy consumption and environmental status of the building in real time. The local twin processor is installed in the machine room on each floor and is coupled with the data perception nodes. It receives heterogeneous operation data and generates dynamic feature vectors through real-time digital twin mapping. For example, a feature vector reflecting the energy use efficiency of the floor is generated based on the current temperature and electricity consumption. The regional twin nodes are deployed in the central control room of the building and are interconnected with the local twin processors on each floor. They optimize the HVAC operation parameters of each floor based on edge computing and record the energy distribution decisions through the blockchain to ensure data transparency and immutability. The cloud BIM fusion engine runs on the cloud server, receives the data from the regional twin nodes, constructs the digital twin BIM model of the building, predicts the energy consumption trend during the summer peak period based on historical data, and generates a scheduling plan for automatically adjusting the air conditioner power. The operation control terminal presents a real-time view through a tablet computer, displays the energy consumption anomalies on each floor (such as the lighting on a certain floor is not turned off) and optimization suggestions (such as reducing the air conditioner power during non-working hours), and supports the building administrator to issue control commands.

[0025] Furthermore, the system is extended to an industrial park containing a solar power station, wind power equipment and an energy storage system. Sensor arrays are installed on the solar panels, wind turbines and energy storage batteries by the data perception nodes to collect heterogeneous operation data such as light intensity, wind speed, battery voltage, etc., which are updated once per minute. The local twin processors are deployed beside each type of energy equipment, receive the data and generate dynamic feature vectors. For example, an operation efficiency vector of the wind power equipment is generated based on the wind speed and power generation. Multiple regional twin nodes are set up in the industrial park, which are respectively responsible for the data processing of the solar energy area, wind energy area and energy storage area. Local optimization is carried out through edge computing, and knowledge graphs are used to achieve cooperation between nodes to dynamically adjust the power distribution (such as storing excess solar energy in the battery). The cloud BIM fusion engine receives the data from each region, constructs the digital twin BIM model of the entire industrial park, predicts the energy supply and demand balance in the next 24 hours, and generates a scheduling plan (such as preferentially using the energy storage power at night). The operation control terminal provides a real-time view for the industrial park energy management team through desktop software, discovers anomalies (such as the efficiency of a certain wind power equipment decreases), and supports adjusting the operation parameters of the equipment according to the suggestions.

[0026] Furthermore, the system is applied to the optimization of the smart grid in an urban area. Data sensing nodes deploy sensors at substations, distribution lines, and residential electricity consumption terminals to collect data such as voltage, current, and load, and update it every 10 seconds. Local twin processors are installed at each substation to process heterogeneous data within their jurisdiction and generate dynamic feature vectors reflecting the grid load distribution. Regional twin nodes cover different urban areas (such as residential areas and industrial areas), optimize the local grid load based on edge computing, and achieve distributed decision-making for inter-regional power trading through blockchain (for example, the industrial area sells excess power to the residential area). The cloud BIM fusion engine integrates the city-wide data, constructs a digital twin BIM model of the power grid, predicts the power demand during peak hours, and generates a self-optimizing scheduling plan (such as adjusting the output power of the substation). The operation control terminal displays the real-time grid status on the large screen of the operation center, prompts anomalies (such as overload of a certain line), and supports engineers to issue control commands to balance the load.

[0027] It can be understood that through digital twin technology, edge computing, and decentralized collaborative optimization, the present invention improves the intelligence level of energy management, realizes real-time data fusion, dynamic optimization scheduling, and adaptive feedback control, is applicable to the refined management of large energy systems, and has high application value. Specifically, the system realizes the real-time collection and mapping of data through a multi-source sensor array and local twin processors, significantly improving the accuracy and timeliness of energy status monitoring; the edge computing and decentralized collaborative mechanism (such as blockchain or knowledge graph) of regional twin nodes optimize local resource allocation, enhancing the adaptability and collaboration efficiency of the system in complex scenarios; the cloud BIM fusion engine provides a global perspective and long-term trend prediction, providing a scientific basis for energy scheduling. In addition, the intuitive interaction design of the operation control terminal further improves the flexibility of user control, enabling the system not only to automatically optimize operation but also to meet the needs of manual intervention. This multi-level and multi-dimensional technology integration effectively reduces energy consumption and improves resource utilization efficiency, is particularly applicable to diverse scenarios such as buildings, industrial parks, and urban power grids, and promotes the development of energy management towards intelligence and refinement.

[0028] It can be understood that the dynamic fusion system for digital twin energy management BIM modeling can be applied to various scenarios. Here, the intelligent building energy management system is taken as an example, but the present invention is not limited thereto.

[0029] In some embodiments, data-aware nodes are deployed in the HVAC system, lighting system, and power distribution system of a building. A multi-source sensor array is configured, including temperature sensors, current sensors, and light sensors, which collect heterogeneous operation data once per second, such as indoor temperature, lighting current, and power distribution power on each floor. The data-aware nodes perform primary feature extraction on the collected heterogeneous operation data to generate an energy status vector. For example, for the HVAC system on a certain floor, the sensors collect a temperature Q1 = 25 ∘ °C and a current Q2 = 10A. Combining the dynamic calibration coefficients k1 = 0.6 (based on the current cooling mode) and k2 = 0.4 (based on the electricity consumption weight), through the formula: = = 15.52; The energy status vector V = 15.52 of this floor is calculated, reflecting its energy consumption status. The dynamic calibration coefficients are adaptively updated according to the building operation mode (such as cooling or heating). For example, in the heating mode, k1 may be adjusted to 0.7, and all coefficients satisfy the normalization condition (i.e., ∑k i = 1).

[0030] Furthermore, in the lighting system, the data-aware node collects a light intensity Q1 = 300 lux and a current Q2 = 5A. The dynamic calibration coefficients are k1 = 0.3 (based on the influence of natural light) and k2 = 0.7 (based on the electricity consumption priority) respectively. Calculate the energy status vector: = = 190.14; Generate V = 90.14, reflecting the energy status of the lighting system. The dynamic calibration coefficients are adaptively adjusted according to the day or night mode and optimized under the normalization constraint.

[0031] Furthermore, in the power distribution system, the data-aware node collects a voltage Q1 = 220V and a load current Q2 = 20A. The dynamic calibration coefficients are k1 = 0.5 and k2 = 0.5 (based on the equalization weight). Calculate the energy status vector and generate V = 110.45V, reflecting the operation status of the power distribution system. The coefficients are adaptively updated according to the peak or valley load periods to ensure data accuracy.

[0032] It can be understood that the above further improves the refinement and intelligence level of energy management. Specifically, the data-aware nodes collect heterogeneous operation data in real time through the multi-source sensor array and use the formula Generating an energy state vector enables the precise quantification of the operating state of energy facilities. The adaptive update mechanism of the dynamic calibration coefficient can flexibly adjust the weights according to different operating modes (such as cooling, heating, or peak hours), ensuring that the energy state reflected by the vector is highly consistent with the actual demand. This method not only improves the real-time performance and accuracy of data processing but also provides reliable basic data for subsequent digital twin mapping and optimal scheduling, reduces energy waste caused by data deviation, enhances the applicability of the system in complex scenarios, and thus provides efficient support for the refined management of large-scale energy systems.

[0033] In some embodiments, the local twin processor is installed in the machine room on each floor and is coupled with the data sensing nodes of the building to receive heterogeneous operating data from the heating, ventilation, and air conditioning (HVAC) system, lighting system, and power distribution system. The data sensing nodes collect data once per second, including temperature (e.g., 25°C), current (e.g., 10A), light intensity (e.g., 300 lux), and voltage (e.g., 220V). The local twin processor incorporates an adaptive twin mapping algorithm to process this data and generate a dynamic feature vector to reflect the energy operating state of each floor. For example, for the HVAC system on the 5th floor, the input data is temperature Q1 = 25°C and current Q2 = 10A. The algorithm assigns weights according to the current cooling mode (e.g., w1 = 0.6, w2 = 0.4) and calculates the components of the dynamic feature vector: V5 = [w1⋅Q1, w 2⋅ Q2] = [0.6⋅25, 0.4⋅10] = [15, 4], generating the dynamic feature vector V5 = [15, 4]. Similarly, for the lighting system on the 10th floor, the input light intensity is Q1 = 300 lux and current Q2 = 5A, and the weights are adjusted to w1 = 0.3 and w2 = 0.7 (due to the influence of natural light during the day), generating V 10 = [90, 3.5].

[0034] Specifically, the local twin processor adopts a self-organizing hierarchical strategy to process the data according to the characteristics of the dynamic feature vector. For example, during the peak working hours (9:00 - 11:00 am), the modulus of the dynamic feature vector V5 = [15, 4] of the HVAC on the 5th floor is relatively large ( = 15.52), and the temperature exceeds the comfort range (set at 22°C - 24°C), which is determined as critical data; while for the lighting system on the 10th floor, V10 = [90, 3.5] ( = 90.06), since the energy consumption is normal, it is classified as regular data. The key data is directly transmitted to the area twin node in the central control room of the building through the priority channel, and the transmission delay is controlled within 50 milliseconds to ensure real-time response; the regular data is compressed by the local twin processor (for example, compressed to 60% of the original size using the Lempel-Ziv algorithm) and transmitted through the secondary channel, with a delay of about 200 milliseconds.

[0035] Furthermore, the adaptive twin mapping algorithm is dynamically adjusted according to the real-time state of the building. For example, when the external temperature rises to 35 °C, the weight w1 of the HVAC system increases from 0.6 to 0.8, reflecting the greater impact of temperature on energy consumption, and a new V5 = [20, 4] is regenerated. At the same time, the self-organizing hierarchical strategy adjusts the threshold according to the peak electricity consumption of the building (for example, the total load reaches 500 kW at 10:00 am), and the dynamic feature vectors with a modulus exceeding 20 (such as the new modulus of V5 = 20.4) are preferentially classified as key data to ensure efficient processing during peak periods. In addition, the local twin processor also records the hourly data classification statistics. For example, the 5th floor generates 1000 groups of dynamic feature vectors from 9:00 to 10:00 am, of which 20% are key data and 80% are regular data, providing a basis for subsequent optimization.

[0036] It can be understood that the local twin processor described in this embodiment, its built-in adaptive twin mapping algorithm and self-organizing hierarchical strategy significantly improve the intelligence and efficiency of energy management. Specifically, the local twin processor generates dynamic feature vectors in real time through the adaptive twin mapping algorithm, and can dynamically adjust the mapping weight according to the operating state of energy facilities (such as temperature changes or peak electricity consumption), ensuring that the feature vectors accurately reflect the current energy consumption characteristics and providing accurate data support for subsequent optimization. The self-organizing hierarchical strategy intelligently classifies the dynamic feature vectors, quickly transmits the key data (such as abnormal energy consumption or peak load) through the priority channel, ensuring real-time performance; while the regular data is transmitted through the secondary channel after compression, effectively reducing the network bandwidth occupancy and improving the data processing efficiency. This hierarchical transmission mechanism not only optimizes resource allocation, but also reduces system latency, enabling the building energy management to quickly respond to anomalies and implement regulation in a complex and changing environment.

[0037] In some embodiments, such as Figure 3As shown, the streaming aggregation relay is connected to the local twin processors on all floors through the building's distributed network and is responsible for receiving the compressed regular data transmitted through the secondary channel. For example, during the peak period from 9:00 to 10:00 in the morning, the streaming aggregation relay receives approximately 800 groups of regular data per minute (such as the compressed data of the lighting system on the 10th floor), and the size of each group of data is approximately 0.6KB. It uses a streaming clustering algorithm (based on online K-means) to cluster these data in real time into categories such as "normal lighting energy consumption" or "low-load power distribution". Taking the lighting data on the 10th and 12th floors as an example, the streaming aggregation relay clusters two groups of similar data into one category, generates a clustering center, and batch forwards it to the regional twin node every 5 seconds. At the same time, the key data of the HVAC system on the 5th floor is directly transmitted to the regional twin node through the priority channel without passing through the streaming aggregation relay to ensure real-time performance.

[0038] Specifically, when the external environment changes (such as the light intensity increasing at 10:00 in the morning), the streaming aggregation relay dynamically adjusts the clustering parameters, re-integrates the updated regular data, and forwards it. If the network is interrupted for 10 seconds, the streaming aggregation relay can cache up to 500 groups of data and send them in order after recovery; if abnormal regular data is detected (such as a sudden increase in current on a certain floor), it is converted into key data and directly transmitted through the priority channel. This mechanism ensures the efficient processing and reliable transmission of the data stream.

[0039] Furthermore, the streaming aggregation relay receives the compressed regular data transmitted through the secondary channel by the local twin processors on each floor through the building's distributed network. During the peak period from 9:00 to 10:00 in the morning, it receives approximately 800 groups of regular data per minute. It performs semantic layering on these data and divides them into semantic layers such as "lighting energy consumption", "HVAC low load", and "power distribution stability" according to the energy type and operating status. For example, the regular data of the lighting system on the 10th floor (reflecting the light and current status after compression) is classified into the "lighting energy consumption" layer, while the HVAC on the 3rd floor is classified into the "HVAC low load" layer due to low-power operation. Then, the streaming aggregation relay conducts a transmission priority assessment: if the data on the 10th floor indicates normal light but slightly higher current (for example, exceeding the average value by 10%), its priority is rated as "high correlation" and it is immediately forwarded to the regional twin node to optimize lighting adjustment; the HVAC data on the 3rd floor has a small change and is rated as "low correlation", and it is buffered locally for 5 seconds and then batch sent to reduce network load.

[0040] Specifically, when the building operation status changes (e.g., external light intensity increases at 10:00 am), the streaming aggregation relay dynamically updates the semantic stratification. The data on the 10th floor, due to significant changes in light, still maintains high correlation and is transmitted immediately, while the low-load data on other floors continues to be buffered. If the normal data of a certain floor is abnormal (such as the lighting current on the 8th floor suddenly increases by 50%), the streaming aggregation relay raises its priority and forwards it immediately. This stratification and evaluation mechanism optimizes the data transmission efficiency.

[0041] It can be understood that this embodiment further improves the efficiency and intelligence of the energy management system. The streaming aggregation relay performs semantic stratification on the normal data, clearly differentiating different energy types and operation states, providing more targeted data support for the regional twin nodes; the transmission priority evaluation ensures that high-correlation data (such as anomalies or critical changes) is transmitted immediately, and low-correlation data is buffered locally, effectively optimizing the utilization of network resources, reducing the bandwidth pressure, and ensuring data fluency especially in dynamic environments or peak hours.

[0042] In some embodiments, the regional twin nodes generate local optimization results based on the high-energy consumption data of the HVAC system on the 5th floor (e.g., the temperature of 28°C exceeds the comfortable range), suggesting an increase in air-conditioning power by 20%; at the same time, the lighting system on the 10th floor is recommended to reduce its brightness by 10% due to sufficient light. These optimization results are transmitted back from the regional twin nodes to the streaming aggregation relay. The streaming aggregation relay constructs an adaptive feedback network with the local twin processors on each floor, and by analyzing the real-time requirements of the building (such as the current floor occupancy rate and the external temperature of 35°C), divides the optimization results into two levels of "urgent" and "routine" for pushing. The air-conditioning adjustment suggestion on the 5th floor, which involves comfort, is marked as "urgent" and pushed to the local twin processor on the 5th floor through a high-speed feedback channel (with a delay less than 50 milliseconds); the lighting adjustment suggestion on the 10th floor, which has less impact, is marked as "routine" and pushed through a secondary channel (with a delay of about 200 milliseconds).

[0043] Specifically, when the building status changes (e.g., the external temperature drops to 30°C at 10:30), the regional twin nodes update the optimization results, reducing the increase in air-conditioning power on the 5th floor to 10%. The streaming aggregation relay dynamically adjusts the feedback network and re-evaluates the pushing priority to ensure that the 5th floor still receives the update instruction with priority. This hierarchical pushing mechanism ensures the pertinence and timeliness of the feedback.

[0044] In some embodiments, the cloud BIM fusion engine runs on a cloud server and receives the fusion data transmitted by the regional twin nodes, including the operation states of the HVAC, lighting, and power distribution systems on each floor. At 10:00 am, the engine is based on the real-time response value R = 28 of the HVAC on the 5th floor ∘ C (current temperature), reference response value R0 = 24 ∘Calculate the energy dynamic index D based on C (comfortable temperature), the current model deduction value T = 30 kW (current energy consumption), and the baseline deduction value T0 = 25 kW (historical average energy consumption). The building HVAC is a high-sensitivity device with a sensitivity factor α = 2.0 (adjustable within the range of [0.8, 2.5]). Through the formula: = = 9.79; Generate D = 9.79, indicating that the HVAC energy consumption on the 5th floor is on the high side. Similarly, for the lighting system on the 10th floor, with R = 5 A, R0 = 4 A, T = 2 kW, T0 = 1.8 kW, and α = 1.0 (lower lighting sensitivity), the calculated D = 1.25 ⋅ 1.246 = 1.56, showing normal operation.

[0045] Based on these indices, the cloud BIM fusion engine combines historical data (energy consumption records for the past 3 months) and external variables (such as the weather forecast for the next week) to predict the long-term trends of the building. For example, it is predicted that the HVAC energy consumption on the 5th floor will increase by 15% during the summer peak period (July - August), while the lighting system will only increase by 5% due to the enhanced natural light. These prediction results are sent back to the regional twin nodes for adjusting the long-term scheduling strategy.

[0046] It can be understood that the function of the cloud BIM fusion engine to calculate the energy dynamic index and generate long-term trend predictions further enhances the scientific nature and forward-looking nature of energy management. The cloud BIM fusion engine quantifies the energy dynamic index D to accurately reflect the real-time operating status of the facilities and adjusts the sensitivity factor according to the device type to ensure the applicability of the index. This quantification method provides a reliable basis for long-term trend prediction, enabling the system to identify the energy consumption change trends in advance based on historical data and external conditions, such as predicting the peak demand growth. The feedback of the prediction results supports the regional nodes to optimize the scheduling strategy, reduces the risk of energy waste and equipment overload, enhances the system's planning ability in complex environments, and provides important support for the refined management and sustainable development of large-scale energy systems.

[0047] In some embodiments, the real-time optimization unit is coupled to the local twin processors in the machine rooms on each floor and receives the local optimization results generated by the regional twin nodes. At 10:00 am, the HVAC system on the 5th floor triggered an optimization recommendation from the regional twin node due to the temperature rising to 28°C (exceeding the comfortable range of 24°C): increase the air-conditioning power by 20% (from 50 kW to 60 kW). Based on this recommendation and combined with the real-time status of the building (such as the current floor occupancy rate of 80% and the external temperature of 35°C), the real-time optimization unit dynamically adjusts the operating parameters of the HVAC. It lowers the cooling temperature setting from 25°C to 23°C and increases the fan speed by 10% (from 1000 rpm to 1100 rpm) to quickly restore comfort while avoiding excessive energy consumption. After the adjustment, the temperature on the 5th floor drops to 24°C within 15 minutes, and the energy consumption stabilizes at 58 kW, achieving an optimal balance of utilization.

[0048] Specifically, when the external temperature drops to 30°C at 10:30, the regional twin node updates the optimization result and recommends reducing the power increase to 10% (55 kW). The real-time optimization unit responds quickly, adjusts the fan speed back to 1050 rpm, and slightly adjusts the cooling temperature to 24°C to ensure the dynamic balance between energy efficiency and comfort. Throughout the process, the real-time optimization unit records the adjustment effects and feeds them back to the local twin processors to provide data support for subsequent optimizations.

[0049] In some embodiments, based on the analysis of long-term operation data, the cloud BIM fusion engine discovers that the temperature sensor of the data perception node of the HVAC system on the 5th floor is continuously 2°C higher in high-temperature environments (external temperature of 35°C), which affects the accuracy of the dynamic feature vectors. At 10:00 am, the engine generates a parameter calibration instruction and sends it to the data perception node on the 5th floor through the regional twin node. The instruction adjusts the calibration coefficient of the temperature sensor from 1.0 to 0.95, making the collected value closer to the real temperature (for example, calibrating from 28°C to 26.6°C). After calibration, the operating efficiency of the HVAC is improved, reducing excessive cooling caused by misjudgment, and the power drops from 60 kW to 55 kW.

[0050] At the same time, based on the energy consumption trend in the past week (such as the efficiency decline of the lighting system during low load at night), the cloud BIM fusion engine optimizes the digital twin model and updates the prediction algorithm for the lighting system on the 10th floor (upgrading from linear regression to an exponential model with time decay). At 10:30 am, the engine sends a model update instruction to the data perception node on the 10th floor through the regional twin node, requiring the sensor to increase the sampling frequency of light changes (from once per second to twice per second) to capture the subtle requirements of night-time lighting adjustment. After the update, the lighting system reduces 10% of the ineffective energy consumption during low load, and the operation loss is significantly reduced.

[0051] In some embodiments, such asFigure 2 As shown, a dynamic fusion method for base digital twin energy management BIM modeling is also provided, and the specific steps are as follows: (1) The data perception nodes collect heterogeneous operation data in real time and output it. At 10:00 am, the data perception nodes of the 5th floor HVAC system collect data of 28°C and 12A through temperature and current sensors and output it once per second.

[0052] (2) The local twin processor receives the data and generates dynamic feature vectors. The 5th floor local twin processor adopts an adaptive twin mapping algorithm to map the temperature and current data into dynamic feature vectors V5 = [16.8, 4.8] (based on weight adjustment), reflecting the current energy consumption status.

[0053] (3) The regional twin nodes perform local energy optimization and cooperation. The regional twin nodes receive the 5th floor data, analyze the temperature over - standard based on edge computing, generate optimization suggestions (increase the air - conditioner power by 15%), and cooperate with other floor nodes through the blockchain to confirm the feasibility of power distribution.

[0054] (4) The cloud BIM fusion engine constructs a model and deduces the global state. The cloud receives the regional data, constructs a digital twin BIM model of the building using a multi - scale dynamic fusion model, deduces the global energy consumption, discovers that the 5th floor needs to be adjusted first, generates a self - optimization scheduling plan (increase the power to 57.5kW) and a long - term prediction (the summer energy consumption increases by 10%).

[0055] (5) The optimization plan and prediction are transmitted back to the regional twin nodes. The cloud transmits the scheduling plan and prediction results at 10:05, and the regional twin nodes are ready to execute and distribute them to the local processors.

[0056] (6) The operation control terminal presents a view and supports regulation. The administrator views the real - time BIM model through a tablet computer, sees the temperature anomaly prompt and scheduling suggestions (increase the power to 57.5kW) on the 5th floor, manually confirms the instruction, and the cloud then coordinates the execution. At 10:15, the temperature drops to 24°C.

[0057] It can be understood that through the real - time collection of data perception nodes, the dynamic mapping of local twin processors, the edge optimization and cooperation of regional twin nodes, the global deduction of the cloud BIM fusion engine, and the interactive regulation of the operation control terminal, this method realizes the full - process closed - loop management from data collection to optimization execution. Each step is closely connected, ensuring the effective implementation of real - time data fusion, dynamic optimization scheduling, and adaptive feedback. For example, quickly responding to temperature anomalies and optimizing the air - conditioner power not only ensures comfort but also avoids waste. In addition, this method supports user participation in decision - making, enhances flexibility, is suitable for the refined management of large - scale energy systems, and has the practical value of high - efficiency energy conservation and wide application.

[0058] The above are only exemplary embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structural transformation made under the technical concept of the present invention by using the content of the specification and drawings of the present invention, or any direct / indirect application in other related technical fields, is included in the patent protection scope of the present invention.

Claims

1. A dynamic fusion system for digital twin energy management BIM modeling, characterized in that, Including: Data-aware nodes, deployed around energy facilities, configured with multi-source sensor arrays, for real-time collection of heterogeneous operation data of the energy facilities and output; Local twin processors, coupled with the data-aware nodes, for receiving the heterogeneous operation data and performing real-time digital twin mapping to generate dynamic feature vectors; Regional twin nodes, interconnected with multiple local twin processors, for local energy optimization based on edge computing and collaborating with other regional twin nodes through a decentralized cooperation mechanism, where the decentralized cooperation mechanism uses blockchain or knowledge graph to achieve distributed decision-making; Cloud BIM fusion engine, interconnected with the regional twin nodes, for receiving the data transmitted by the regional twin nodes, constructing a digital twin BIM model, generating a self-optimizing energy scheduling plan and long-term trend prediction, and transmitting the results back to the regional twin nodes; Operation control terminal, interconnected with the cloud BIM fusion engine, for presenting the real-time view of the digital twin BIM model, energy anomaly alerts and scheduling suggestions, and supporting users to issue control commands.

2. The dynamic fusion system for digital twin energy management BIM modeling according to claim 1, characterized in that, The data-aware nodes perform primary feature extraction on the collected heterogeneous operation data to generate an energy state vector, where the energy state vector is calculated by the following formula: ； Among them, (Q1, Q2, …, Q m ) is the heterogeneous operation data component collected by the multi-source sensor array, (k1, k2, …, k m ) is the corresponding dynamic calibration coefficient, which is adaptively updated based on the operation mode of the energy facility and satisfies the normalization condition.

3. The dynamic fusion system for digital twin energy management BIM modeling according to claim 1, characterized in that, The local twin processors are built-in with an adaptive twin mapping algorithm for generating the dynamic feature vectors and adopt a self-organizing hierarchical strategy to determine data classification according to the dynamic feature vectors, where critical data is transmitted through a priority channel and regular data is transmitted through a secondary channel after compression.

4. The dynamic fusion system for digital twin energy management BIM modeling according to claim 3, characterized in that, The system further includes a streaming aggregation relay, connected to multiple local twin processors through a distributed network, for receiving the compressed regular data transmitted by the local twin processors through the secondary channel, performing streaming clustering on the compressed regular data and forwarding it to the regional twin nodes, while the critical data is directly transmitted by the local twin processors to the regional twin nodes through the priority channel.

5. The dynamic fusion system for digital twin energy management BIM modeling according to claim 4, wherein The streaming aggregation relay performs semantic layering and transmission priority evaluation on the received compressed regular data, immediately forwards the highly correlated data to the regional twin nodes, and locally buffers the low-correlated data to reduce network load.

6. The dynamic fusion system for digital twin energy management BIM modeling according to claim 5, characterized in that, The streaming aggregation relay constructs an adaptive feedback network with multiple local twin processors according to the local optimization results generated by the regional twin nodes, supporting hierarchical pushing of the local optimization results to the corresponding local twin processors according to the real-time requirements of the energy facilities.

7. The dynamic fusion system for digital twin energy management BIM modeling according to claim 1, characterized in that, The cloud BIM fusion engine calculates the energy dynamic index D based on the fusion data and generates the long-term trend prediction accordingly, where the energy dynamic index D is calculated by the following formula: ； Wherein, R is the real-time response value of the heterogeneous operation data, R0 is the reference response value, T is the current model deduction value, T0 is the benchmark deduction value, α is the sensitivity factor, and the sensitivity factor α is adjusted in the range of [0.8, 2.5] according to the type of the energy facility.

8. The dynamic fusion system for digital twin energy management BIM modeling according to claim 1, characterized in that, The system further includes a real-time tuning unit, coupled to the local twin processor, which dynamically adjusts the operating parameters of the energy facilities according to the local optimization results of the regional twin nodes to achieve an optimal balance of energy utilization efficiency.

9. The dynamic fusion system for digital twin energy management BIM modeling according to claim 1, characterized in that, The cloud BIM fusion engine sends parameter calibration instructions or model update instructions to the data sensing nodes through the regional twin nodes to support the online performance improvement of the energy facilities and reduce operation losses.

10. A dynamic fusion method for digital twin energy management BIM modeling based on the system described in claim 1, characterized in that, It includes the following steps: (1) Through the data sensing nodes deployed around the energy facilities, a multi-source sensor array is used to collect the heterogeneous operation data of the energy facilities in real time and output it; (2) The local twin processor is used to receive the heterogeneous operation data, and an adaptive twin mapping algorithm is used for real-time digital twin mapping to generate dynamic feature vectors; (3) The regional twin nodes are used to receive the data transmitted by the local twin processor, perform local energy optimization based on edge computing, and cooperate with other regional twin nodes through a decentralized cooperation mechanism, where the decentralized cooperation mechanism uses blockchain or knowledge graph to achieve distributed decision-making; (4) The cloud BIM fusion engine is used to receive the data transmitted by the regional twin nodes, construct a digital twin BIM model using a multi-scale dynamic fusion model, and perform global energy state deduction to generate a self-optimizing energy scheduling plan and long-term trend prediction; (5) The self-optimizing energy scheduling plan and long-term trend prediction generated by the cloud BIM fusion engine are transmitted back to the regional twin nodes; (6) Through the operation control terminal, the self-optimizing energy scheduling plan and long-term trend prediction transmitted back by the cloud BIM fusion engine are received, the real-time view of the digital twin BIM model, energy anomaly prompts, and scheduling suggestions are presented, and the user is supported to issue control commands to the cloud BIM fusion engine.

Images