Building energy consumption real-time monitoring and carbon emission evaluation method based on digital twinning

Abstract

The invention provides a building energy consumption real-time monitoring and carbon emission evaluation method based on digital twinning, and the method comprises the steps: obtaining regional energy supply real-time data and building group load demand data, predicting an energy supply fluctuation trend and a load change trend, and obtaining a short-term supply and demand dynamic prediction result; adjusting the operation mode of the building group combined cooling heating and power system in real time according to the mode switching optimization scheme, dynamically matching the output change of the distributed energy, and feeding back the output change to the digital twinborn model; carbon emission data is extracted from the optimized load distribution scheme, a digital twinborn model is used for predicting the total carbon emission amount and peak value change trend, the threshold value and constraint of carbon emission control are adjusted, and a final energy scheduling scheme is generated; and obtaining operation data according to the final energy scheduling scheme, calculating the energy utilization efficiency and the carbon emission reduction amount of the building group through an energy efficiency evaluation method preset in the digital twin model, and obtaining an execution effect evaluation result of the scheduling scheme.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of information technology, and in particular to a building energy consumption real-time monitoring and carbon emission evaluation method based on digital twinning. BACKGROUND

[0002] Building group energy management is a key area for achieving low-carbon urban development, as it plays a crucial role in optimizing energy use and reducing carbon emissions, thereby promoting sustainable urbanization. With the expansion of urban building scale and the growth of energy demand, energy collaborative scheduling and carbon emission management of building groups have become the focus of research. However, existing methods are often limited to energy optimization of a single building, ignoring the complementary characteristics of internal loads and the complex effects of regional energy supply fluctuations. These methods struggle to achieve efficient scheduling adjustments when faced with dynamic changes in energy flow networks or fluctuations in distributed energy output, resulting in low energy utilization efficiency and poor carbon emission control. In this field, the core challenge lies in the dynamic reconstruction of building group energy flow networks and the rapid switching of combined cooling heating and power (CCHP) system operation modes. When regional energy supply structures fluctuate, such as unstable renewable energy output, energy flow networks need to be reconfigured to adapt to new supply and demand balances. This topological reconstruction directly affects the operation efficiency of CCHP systems, as the system needs to switch between different modes to respond to load changes. However, the flexibility of mode switching is limited by the insufficient matching between distributed energy and demand-side response, making it difficult to optimize scheduling strategies in real time. This chain effect makes it difficult for building groups to achieve energy-efficient utilization and effective carbon emission control in dynamic environments. Therefore, how to dynamically adjust building group energy scheduling strategies under regional energy supply fluctuations, utilize internal load complementary characteristics, and simultaneously achieve coordinated optimization of energy flow network topology reconstruction, CCHP system mode switching, and distributed energy output fluctuations, has become a key problem in achieving regional total carbon emission and peak control. SUMMARY

[0003] The present application provides a building energy consumption real-time monitoring and carbon emission evaluation method based on digital twinning, mainly including:

[0004] Obtain real-time data of regional energy supply and load demand data of building groups, predict energy supply fluctuation trend and load change trend, and obtain short-term supply and demand dynamic prediction results;

[0005] According to the short-term supply and demand dynamic prediction results, use graph theory algorithm to analyze the connectivity and transmission efficiency of building group energy flow network topology, determine the node connection strength and energy transmission efficiency of the current network, combine with the spatial distribution characteristics of building group energy consumption, judge the overall supply-demand matching degree of the network, and obtain the reconstruction demand evaluation results;

[0006] If topology reconstruction is needed, adjust the connection structure of the energy flow network through the minimum spanning tree algorithm to obtain the reconstructed network topology dynamic configuration, and if the reconstruction demand evaluation result indicates that the current network can effectively match the supply and demand, maintain the original network topology unchanged;

[0007] A virtual mapping model of the building group energy system is constructed, energy supply and demand data of key nodes are extracted from the reconstructed network topology configuration, energy efficiency and carbon emissions under different load distribution ratios are calculated in combination with operation parameters of the combined cooling heating and power system, and an optimization scheme of mode switching is formed;

[0008] According to the optimization scheme of mode switching, the operation mode of the combined cooling heating and power system of the building group is adjusted in real time, the output change of the distributed energy is dynamically matched, and feedback is fed back to the digital twin model;

[0009] The adjusted system operation parameters and real-time output data of the distributed energy are obtained, the load distribution inside the building group is cooperatively scheduled, the total energy consumption and carbon emission of the building group are minimized as the objective function, and an optimized load distribution scheme is generated;

[0010] Carbon emission data is extracted from the optimized load distribution scheme, the total carbon emission and peak change trend are predicted by using the digital twin model, the threshold and constraint of carbon emission control are adjusted, and a final energy scheduling scheme is generated;

[0011] According to the final energy scheduling scheme, operation data is obtained, the energy utilization efficiency and carbon emission reduction amount of the building group are calculated by using the preset energy efficiency evaluation method in the digital twin model, and the execution effect evaluation result of the scheduling scheme is obtained.

[0012] The technical scheme provided by the embodiment of the present application can include the following beneficial effects:

[0013] The present application discloses a kind of based on digital twin's building energy consumption real-time monitoring and carbon emission evaluation method, by obtaining regional energy supply and building load data, using time series analysis to predict supply and demand trend, energy network topology structure is analyzed using graph theory algorithm, according to demand evaluation result dynamically reconstructs network topology.Based on digital twin technology, a virtual model is constructed, key node data is extracted to optimize the operation mode of the combined cooling heating and power system, and source-load coordination is realized.Minimize total energy consumption and carbon emission as the goal by using particle swarm optimization algorithm to cooperatively schedule building load.Combined with carbon emission prediction, dynamically adjust energy scheduling instruction, preferentially use renewable energy, and start demand response if necessary.Finally, the optimization effect is evaluated by digital twin model, and the changes of energy consumption and carbon emission are visualized.The present application realizes the building group energy utilization efficiency improvement and carbon emission reduction, and provides an intelligent solution for regional energy management. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1A flow chart of a building energy consumption real-time monitoring and carbon emission evaluation method based on digital twinning.

[0015] Figure 2 A schematic diagram of a building energy consumption real-time monitoring and carbon emission evaluation method based on digital twinning.

[0016] Figure 3 Another schematic diagram of a building energy consumption real-time monitoring and carbon emission evaluation method based on digital twinning. DETAILED DESCRIPTION

[0017] For a further understanding of the present application, reference will be made to the following description taken in conjunction with the accompanying drawings. The present application will be further described with reference to the drawings, in which embodiments of the application are illustrated. It is to be understood that the specific embodiments described herein are merely illustrative of the application and are not to be used as limitations. It is also to be understood that the drawings are only schematic and that the specific embodiments are not to be interpreted as limiting the scope of the application. In addition, it is to be understood that the specific embodiments described herein are merely illustrative of the application and are not to be used as limitations.

[0018] As Figure 1 -3, the building energy consumption real-time monitoring and carbon emission evaluation method based on digital twinning can specifically include:

[0019] S101, obtain real-time data of regional energy supply and load demand data of building groups, predict energy supply fluctuation trend and load change trend, and obtain short-term supply and demand dynamic prediction results.

[0020] The real-time data of regional energy supply includes power supply, heat supply and cold supply. The load demand data of the building groups in each period is collected. The load data is classified and marked according to the building function type. The deviation value of the supply and load in each period is calculated to identify the time period distribution characteristics of the supply and demand imbalance, and the supply and demand deviation time series data is obtained. For the supply and demand deviation time series data, the periodic characteristic parameters are extracted, including daily cycle fluctuation rate, weekly cycle fluctuation rate and seasonal fluctuation rate. The ARIMA time series model is used to fit the periodic characteristic parameters and historical deviation data. According to the occurrence rule of historical peak value and valley value, the supply prediction value and load prediction value in each period within the next 24 hours are calculated, and the supply and demand prediction time series is obtained. According to the prediction values of adjacent periods in the supply and demand prediction time series, the supply prediction value change rate and the load prediction value change rate are calculated. If the change rate exceeds the preset threshold, the period is marked as a supply and demand fluctuation key period. By comparing the deviation degree of the prediction value and the historical actual value of the same period, the prediction parameter weight of the next period is adjusted, and the short-term supply and demand dynamic prediction result is obtained.

[0021] For example, the dynamic balance between district energy supply and building load demand is the core issue of smart energy management. In practical application scenarios, a building complex usually includes office buildings, commercial complexes, residential areas, and other buildings with different functional types, each with significantly different energy consumption characteristics. Office buildings have high energy consumption during the morning of weekdays from 9 am to 6 pm, while residential areas have high energy consumption during the morning and evening. This difference makes accurate prediction and matching of supply and demand extremely important.

[0022] In one possible implementation, the system collects real-time power load data through smart meters deployed in each building power distribution room, obtains heat supply and demand data through flow meters and temperature sensors at heat stations, and collects cold-related parameters through monitoring equipment in the refrigeration room. These data are collected at 5-minute intervals to form high-precision time series data streams. The calculation of supply and demand deviation not only considers the absolute numerical difference, but also considers the conversion efficiency and transmission loss of different energy types. When the power supply is 1000 kW and the actual load demand is 1200 kW, a 200 kW deviation value will be recorded and marked as a supply and demand imbalance feature for that period.

[0023] It should be noted that the periodic feature extraction of energy consumption is the key to prediction accuracy. The daily cycle fluctuation rate reflects the energy consumption change law within 24 hours, which is obtained by calculating the deviation of each period energy consumption value from the daily average; the weekly cycle fluctuation rate reflects the difference pattern between weekdays and weekends; the seasonal fluctuation rate is mainly affected by climate factors, and the summer cooling demand and winter heating demand will cause the overall shift of the energy consumption baseline. The application of the ARIMA model requires first performing stationarity test on the time series data, eliminating the trend through difference operation, and then determining the model order according to the autocorrelation function and partial autocorrelation function. This model can capture the autoregressive and moving average characteristics of time series data, thereby achieving accurate prediction of future supply and demand status.

[0024] Specifically, when the system detects that the supply prediction value change rate of a certain period exceeds the preset threshold of 15%, it means that there may be a large fluctuation on the energy supply side, and the system will immediately start the dynamic adjustment mechanism. Comparative analysis of historical same period data can help identify abnormal situations, such as the load demand of the same period last year being 800 kW, while this year's predicted value reaching 1000 kW, which may be due to the use of new equipment or changes in building functions. By giving higher weight to recent data and gradually reducing the influence of historical data, the prediction parameters are adjusted adaptively.

[0025] In one embodiment, the calculation of supply-demand matching degree adopts the ratio method. When the supply prediction value is 1000 kW and the load prediction value is 1200 kW, the matching degree is 0.83. If the preset matching threshold is 0.9, the period is identified as a supply-demand imbalance risk period. This quantitative evaluation method can help energy management personnel to identify potential supply shortages or surpluses in advance. The distribution rule analysis of the risk period reveals the spatio-temporal characteristics of the energy supply-demand contradiction. For example, the occurrence of imbalance risk for three consecutive periods indicates a persistent supply pressure, while scattered single-point imbalance may be caused by incidental factors.

[0026] Preferably, the generation of the supply-demand dynamic prediction report comprehensively considers multi-dimensional information. The accurate identification of peak time helps to develop peak clipping and valley filling strategies, the determination of valley time provides the best opportunity for energy storage system charging, and the quantitative evaluation of fluctuation amplitude provides the basis for the flexibility requirements of energy dispatching. Compared with traditional experience judgment, this data-driven prediction method can reduce the prediction deviation by more than 30%, significantly improving the operation efficiency and supply reliability of the regional energy system.

[0027] S102, according to the short-term supply-demand dynamic prediction result, the connectivity and transmission efficiency of the building group energy flow network are analyzed by using the graph theory algorithm, the node connection strength and energy transmission efficiency of the current network are determined, the supply-demand matching degree of the whole network is judged combined with the spatial distribution characteristics of the building group energy consumption, and the reconstruction demand evaluation result is obtained.

[0028] According to the supply and demand time series data in the short-term supply and demand dynamic prediction result, a building group energy network topology graph is constructed, wherein each building is a network node, and the energy transmission pipeline between nodes is an edge. By obtaining the real-time energy supply and load demand of each node, the energy flow direction and transmission flow between adjacent nodes are calculated to obtain the energy network topology structure containing node attributes and edge weights. For the energy network topology structure, the Dijkstra algorithm is used to calculate the shortest transmission path between any two nodes. According to the path length and actual transmission flow, the transmission loss rate of each edge is calculated. The node connection strength is determined by counting the number of supply edges and demand edges connected to each node. If the node connection strength is lower than the preset threshold, the node is marked as a weak node to obtain network connectivity data containing a weak node set. According to the position of the weak node in the network connectivity data, combined with the energy consumption intensity value of each building node, the transmission path loss rate between the weak node and the high energy consumption node is calculated. By comparing the ratio of the current transmission flow to the rated capacity of the pipeline, the position of the transmission bottleneck edge and the bottleneck degree coefficient are determined to obtain the transmission efficiency coefficient of each edge and the bottleneck distribution diagram. The transmission efficiency coefficient of each edge and the bottleneck distribution diagram are used to obtain the network overall transmission efficiency by averaging the efficiency coefficients of all edges. The node matching degree is obtained by dividing the absolute value of the difference between the supply and demand of each node by the demand. The network supply and demand matching degree is obtained by averaging the matching degrees of all nodes. If the matching degree is lower than the reconstruction threshold, a reconstruction demand evaluation report containing the weak node position, bottleneck edge distribution and matching degree value is generated.

[0029] For example, the topology structure of the building group energy network is the basis for realizing energy optimization scheduling. In actual regional energy management, a building group often contains dozens of buildings with different functions, which are connected to each other through cables, heat pipes and cold water pipes in underground pipe corridors, forming a complex energy transmission network. Each building can be both a consumer and a supplier of energy at a certain time, such as an office building equipped with a photovoltaic power generation system that can supply power to surrounding buildings on weekends.

[0030] Specifically, in the construction process of the energy network topology graph, the system abstracts each building as a node, and the node attributes include the maximum supply capacity, average load demand and energy type of the building. The energy transmission pipeline between buildings is abstracted as an edge with weight, and the weight of the edge reflects the transmission capacity and transmission loss characteristics of the pipeline. When A office building supplies power to B shopping mall through a cable with a rated power of 500kW, the connection in the topology graph is represented as a directed edge from node A to node B. The edge attributes record information such as transmission capacity, line impedance and real-time transmission flow.

[0031] It should be noted that the application of Dijkstra algorithm in energy network is different from the traditional shortest path problem. Here, "shortest" not only refers to physical distance, but more importantly, the minimum transmission loss. The algorithm considers the influence of factors such as line length, pipe diameter, insulation condition, etc. on transmission efficiency when calculating the path. When 1000kW of power needs to be supplied from the energy center to a building, the algorithm will find the path with the minimum total transmission loss among all feasible paths, even if this path may be longer in physical distance.

[0032] In one possible implementation, the evaluation of node connection strength uses the degree centrality index. The number of supply edges of a node reflects its importance as an energy supply point, and the number of demand edges reflects its dependence on external energy. When a node has only one supply edge and no backup path, this node is identified as a weak node. This weakness means that once the supply line fails, the node will face the risk of energy interruption. By setting the connection strength threshold to 2, the system can filter out all high-risk nodes with single-point connection. The identification of transmission bottlenecks is crucial for network optimization. The system monitors the transmission flow of each pipeline in real time and compares it with the rated capacity to calculate the capacity utilization. When the actual flow of a heat pipe reaches 90% of the rated capacity, this pipe becomes a potential transmission bottleneck. The bottleneck degree coefficient is quantified by the ratio of actual flow to rated capacity. The closer the coefficient is to 1, the heavier the load of the pipeline, and the more critical it becomes to the transmission capacity of the entire network.

[0033] Preferably, the evaluation of the overall transmission efficiency of the network needs to consider the efficiency of all transmission paths. The transmission efficiency coefficient of each edge is determined by the ratio of actual transmission power to input power, which reflects the energy loss during transmission. By weighted averaging the efficiency coefficients of all active edges, with the weight being the proportion of the transmission flow of each edge to the total flow, the average transmission efficiency of the entire network can be obtained. This index intuitively reflects the operating state of the energy network. The calculation of supply-demand matching degree evaluates the network performance from another dimension. For each node, the system calculates the deviation between the actual obtained energy supply and demand, and divides the deviation value by the demand to obtain the relative deviation rate. When a building needs 100kW but only gets 80kW supply, its matching degree is 0.8. By averaging the matching degrees of all nodes, the supply-demand matching degree of the entire network is obtained. When this value is lower than the preset threshold of 0.85, it indicates that there is a serious imbalance between supply and demand in the network, and topology reconstruction is needed. The reconstruction requirement evaluation report will list the location of weak nodes, the distribution of bottleneck pipelines, and the specific values of various evaluation indexes in detail, providing data support for subsequent network optimization.

[0034] S103, if topology reconstruction is needed, the connection structure of the energy flow network is adjusted by the minimum spanning tree algorithm to obtain the reconstructed network topology dynamic configuration. If the reconstruction demand evaluation result indicates that the current network can effectively match supply and demand, the original network topology is maintained unchanged.

[0035] According to the supply-demand matching degree value in the reconstruction demand evaluation report, if the matching degree is lower than the preset reconstruction threshold, the high supply node set whose supply exceeds its demand and the high demand node set whose demand exceeds the obtained supply are obtained, the expected loss value of the candidate connection line is determined by calculating the product of the physical distance between each node pair and the transmission loss rate per unit distance, and the candidate line set containing all feasible connections and their loss values is obtained. For the candidate line set, the minimum spanning tree algorithm is used to select the basic connection scheme that minimizes the total transmission loss of the network, and the node pairs whose physical distance is less than the preset distance threshold but are not directly connected between the supply nodes and the demand nodes are identified on the basic connection scheme. By adding the direct connection lines of these node pairs and deleting the lines whose transmission flow is lower than the flow threshold, a topology adjustment scheme containing the new line list and the deleted line list is obtained. According to the new lines and the retained lines in the topology adjustment scheme, combined with the daily highest load period and the daily lowest load period recorded in the historical load data, the expected transmission flow of each line in different load periods after adjustment is calculated, the supply guarantee degree of each node under the new topology is determined by accumulating all the inbound and outbound flow of each node, and the adjusted network connection configuration is obtained. Using the adjusted network connection configuration, the actual obtained supply and demand of each node are compared to calculate the new supply-demand matching degree, if the difference between the new matching degree and the original matching degree is greater than the preset improvement threshold, the network topology dynamic configuration containing the coordinates of the new lines, the identification of the deleted lines and the updated node connection relationship is generated, otherwise the original network topology in the evaluation report is maintained.

[0036] For example, the topology reconstruction of the energy network is a key means to solve the regional supply-demand imbalance. When the reconstruction demand evaluation report shows that the supply-demand matching degree is lower than the preset threshold of 0.85, it means that the existing network structure cannot meet the demand of efficient energy distribution, and the situation must be improved by adjusting the network connection relationship. The core of reconstruction is to identify high supply nodes and high demand nodes and establish more direct transmission paths between them.

[0037] Specifically, high-supply nodes refer to buildings that produce more energy than they consume, such as industrial plants equipped with large-area photovoltaic panels or energy centers with combined heat and power plants. These nodes can generate more than 500 kW of surplus power during certain periods. Correspondingly, high-demand nodes are buildings that consume much more energy than they can supply, such as large shopping malls or data centers, which can have an energy gap of up to 800 kW. The determination of the candidate line set needs to consider the physical distance and transmission loss between nodes. When two buildings are 200 meters apart and use a specific type of cable, the transmission loss per 100 meters is about 2%, so the expected loss value of this line is 4%.

[0038] It should be noted that the application of the minimum spanning tree algorithm in energy network optimization has special significance. This algorithm gradually selects the edge with the smallest loss through the greedy strategy, ensuring that all nodes are connected while minimizing the total transmission loss of the entire network. In practical applications, the algorithm will preferentially select lines that connect high-supply nodes with high-demand nodes and have smaller transmission losses. However, pure tree structures have single-point failure risks, so key direct connection lines need to be added to the tree structure to form a network structure with certain redundancy.

[0039] In one possible implementation, the addition of direct connection lines follows the principle of proximity and the principle of maximizing benefits. When a photovoltaic plant with a daily surplus of 300 kW is found to be only 150 meters away from an office building with a daily gap of 400 kW, but needs to pass through three intermediate nodes to be connected in the current topology, the system will suggest adding a direct connection line between the two. This new line not only shortens the energy transmission path, but also reduces the loss of intermediate links. At the same time, for lines with a monthly transmission flow of less than 50 kW, the system will mark them as inefficient lines and include them in the deletion candidate list. Analysis of historical load data is crucial for evaluating the effectiveness of topology adjustments. By extracting the load curve over the past year, it can be identified that the daily electricity peak period usually occurs at 10 am and 3 pm, while the trough period is from 3 am to 5 am. The expected transmission flow of the new line during the peak period may reach 85% of the rated capacity, while during the trough period it is only 20%. This load fluctuation characteristic requires the network topology to have sufficient flexibility and adaptability.

[0040] Preferably, the calculation of supply security reflects the ability of the node to obtain stable energy supply under the new topology. For each demand node, the system will count the total supply capacity of all its incoming edges and compare it with the peak demand of the node. When the peak demand of an office building is 600kW, and the total supply capacity of its three incoming edges under the new topology reaches 750kW, the supply security of the node is 1.25, indicating that it can meet its energy demand even in extreme cases. The generation of the dynamic configuration of the network topology is the final output of the entire reconstruction process. The system compares the supply and demand matching degree before and after adjustment, and only when the new matching degree is improved by more than 0.1 compared with the original value, is the reconstruction scheme considered to have implementation value. The configuration file records in detail the starting and ending node coordinates, transmission capacity specifications, expected operation time of each newly added line, as well as the specific identification and removal plan of each deleted line. This data-driven reconstruction method can accurately locate the weak links of the network, achieve the maximum improvement of supply and demand matching degree with the minimum reconstruction cost, compared with the traditional experience-based reconstruction, it can reduce 30% of unnecessary line construction, and at the same time reduce the overall transmission loss by more than 15%.

[0041] S104, constructing a virtual mapping model of the building group energy system, extracting energy supply and demand data of key nodes from the reconstructed network topology configuration, combining with operation parameters of the combined cooling heating and power system, calculating energy efficiency and carbon emissions under different load distribution ratios, and forming an optimization scheme of mode switching.

[0042] According to the dynamic configuration of the reconstructed network topology, a virtual mapping relationship containing all building nodes and connection lines is constructed, real-time energy data and physical attribute parameters of each node are imported to establish a digital twin model of the building group energy network, the connection degree value and betweenness centrality index of each node are calculated in the twin model to obtain the importance quantitative score of each node. According to the importance quantitative score of each node, the nodes with connection degree exceeding the preset connection threshold and betweenness centrality score in the top 20% are selected as key nodes, the real-time supply, demand and transmission flow data of the key nodes are extracted from the digital twin model, and the refrigeration power, heating power and power generation power parameters of the combined cooling, heating and power equipment connected with the key nodes are obtained to obtain the key node operation state data. According to the supply and demand values and equipment power parameters in the key node operation state data, a linear programming method is used to construct an objective function, in which the cold load ratio, heat load ratio and electric load ratio are used as optimization variables, and by solving the energy utilization rate and carbon dioxide emission under different proportion combinations, a load distribution scheme set and corresponding energy efficiency value and carbon emission value are obtained. The load distribution scheme set and corresponding energy efficiency value and carbon emission value are used to compare whether the energy efficiency value of each scheme is higher than the historical average energy efficiency and the carbon emission value is lower than the regional carbon emission standard, and the optimal scheme meeting the conditions is selected, and according to the load demand change law in different time periods within a day, the corresponding optimal scheme is matched to generate an operation mode switching scheme containing switching time points, various load distribution proportions and expected energy efficiency and carbon emission values.

[0043] For example, the application of digital twin technology in building group energy management is becoming an important way to improve energy efficiency. By constructing a virtual mapping of the physical world, energy managers can simulate and verify various optimization strategies without affecting actual operation. This virtual mapping not only contains static network topology, but also reflects the dynamic energy flow state of each node in real time.

[0044] In one possible implementation, the construction of the digital twin model starts with an accurate modeling of the physical network. Each building has a corresponding digital entity in the virtual space, containing its geographic location, building area, list of energy-using equipment, historical energy consumption curve, and other multi-dimensional information. The energy pipelines connecting the buildings are also digitized, recording attributes such as pipe diameter, material, laying path, and insulation condition. When the electricity load of a commercial complex in the actual network suddenly increases from 300 kW to 500 kW, the digital twin model will update the state of the node synchronously and calculate the impact of this change on the entire network. The quantitative evaluation of node importance adopts the classic indicators in graph theory. The connectivity reflects the number of direct connections between a node and other nodes, and a connectivity of 5 means that the node has direct energy transmission relationships with 5 other nodes. The betweenness centrality measures the intermediary role of a node in the entire network, i.e., how many shortest paths need to pass through the node. When the betweenness centrality score of an energy station reaches 0.35, it means that 35% of the energy transmission paths in the network pass through this node, making it a critical energy gathering and distribution hub.

[0045] It should be noted that the identification of critical nodes is crucial for the stable operation of the entire energy network. These nodes are often regional energy centers, large heat exchangers, or power distribution hubs, and their operating status directly affects the energy supply of multiple downstream buildings. By setting the connectivity threshold to 4 and the betweenness centrality percentile to the top 20%, the system can accurately filter out those nodes that have both more connections and play a key role in the network. The real-time data collection frequency of these nodes is usually set to once per minute to ensure that the digital twin model can timely reflect their state changes. Combined heat and power (CHP) systems are an important energy supply method for modern building groups, and the optimization of their operating parameters is directly related to the overall energy efficiency. A typical combined heat and power system includes a gas turbine, a waste heat boiler, and an absorption chiller. The gas turbine generates high-temperature flue gas while generating electricity, the waste heat boiler recovers heat to produce steam or hot water, and the absorption chiller uses heat to produce cooling. The key to the system is how to allocate the proportions of electricity, heat, and cooling loads to achieve the best energy utilization efficiency.

[0046] Specifically, the linear programming method plays a core role in load distribution optimization. The objective function is usually set as the weighted combination of maximum energy utilization rate and minimum carbon emissions. The decision variables include the proportion of electrical load, the proportion of heat load, and the proportion of cooling load, and the sum of these three proportions is 1. The constraint conditions include the capacity limits of various devices, the satisfaction degree of load demand, the upper limit requirement of carbon emissions, etc. When the cooling load demand reaches the peak value in the afternoon of summer, the optimization result may suggest that 60% of the energy is used for cooling, 25% for power supply, and 15% for hot water supply.

[0047] Preferably, the switching of operation modes needs to take into account the time-varying characteristics of the load. Office buildings mainly require cooling and lighting power from 9 am to 11 am on weekdays, while residential areas mainly require heating and domestic power from 18 pm to 22 pm. By analyzing historical data, typical load patterns can be identified, such as "weekday daytime mode", "weekday nighttime mode", "weekend mode", etc. Each mode corresponds to an optimized load distribution scheme. When the actual time enters the corresponding time period, the system automatically switches to the corresponding operation mode. Compared with traditional experience scheduling, this optimization method based on digital twinning can improve the overall energy efficiency by more than 15% and reduce carbon emissions by about 20%. Through continuous data collection and model updating, the system can also continuously learn and adapt to changes in building group energy consumption characteristics, achieving truly intelligent energy management.

[0048] S105, according to the optimization scheme of mode switching, real-time adjustment of the operation mode of the building group combined cooling, heating and power system, dynamic matching of the output change of distributed energy, and feedback to the digital twinning model.

[0049] According to the target load distribution ratio in the operation mode switching scheme, the real-time power generation of the photovoltaic array, the real-time output of the wind turbine generator set and the current state of charge of the energy storage battery are obtained. By calculating the difference between the total output of distributed energy and the total load demand of the building group, the specific value of the energy supply gap or surplus is determined, and the real-time data of supply and demand deviation is obtained. For the real-time data of supply and demand deviation, a decision model is constructed using reinforcement learning algorithm, taking the distributed energy output value, load demand value and energy storage state as environmental state input, and taking the power adjustment range of the combined cooling, heating and power system as action output. The optimal adjustment strategy is obtained by maximizing the long-term cumulative reward, and the corresponding power increase and decrease instructions are generated according to the positive and negative nature of the deviation. According to the adjustment range in the power increase and decrease instructions, if the instruction is to increase the output, the operation power of the corresponding equipment in the combined cooling, heating and power system is increased, and if the instruction is to reduce the output, the equipment power is reduced and a charging instruction is sent to the energy storage system. By real-time monitoring of the total supply power and total demand power after adjustment, the source-load matching degree is determined, and the operation state data containing the real-time power values of various equipment are obtained. Using the equipment power, energy output and energy storage charging and discharging parameters in the operation state data, these real-time parameters are transmitted to the digital twinning platform through a pre-set communication protocol, the attribute values of the corresponding equipment nodes in the virtual model are updated synchronously, the state consistency between the physical system and the digital system is realized, and the synchronous update log with time mark and parameter change amount is generated.

[0050] For example, the coordinated operation of distributed energy and centralized energy is a core challenge in modern building group energy management. In actual operation, photovoltaic power generation is significantly volatile due to sunlight intensity, and wind power generation is random due to wind speed changes, which brings great challenges to energy supply and demand balance. Through real-time monitoring and intelligent scheduling, the advantages of various energy forms can be complementary.

[0051] Specifically, real-time calculation of supply and demand deviation is the basis of the entire scheduling system. On a sunny afternoon, the power generation of the rooftop photovoltaic array reaches 800 kW, while the total load demand of the building group is only 600 kW, and the system calculates that there is a surplus of 200 kW of energy. Conversely, on a rainy day or at night, the photovoltaic output drops to zero, while the load demand remains at 500 kW, and the system identifies a supply gap of 500 kW. This real-time deviation data provides a quantitative basis for subsequent scheduling decisions. The application of reinforcement learning algorithms in energy scheduling reflects the advantages of artificial intelligence technology. The algorithm learns the optimal strategy through continuous interaction with the environment, and the core is to establish the mapping relationship between state, action and reward. The state space includes the current photovoltaic output, wind power output, energy storage capacity, load demand and other multi-dimensional information; the action space defines the power range that can be adjusted by the combined cooling, heating and power system; and the reward function considers factors such as energy utilization efficiency, supply and demand matching degree and operating cost.

[0052] In one possible implementation, the algorithm is trained offline based on historical operation data. By simulating the operation under different weather conditions and different load scenarios, the algorithm gradually learns to make optimal decisions under various operating conditions. When a sudden drop in photovoltaic output is detected, the algorithm will quickly increase the output power of the gas turbine to make up for the gap; when the wind power output unexpectedly increases, the algorithm will reduce the output of the combined supply system and instruct the energy storage system to absorb the excess power.

[0053] It should be noted that the adjustment of the combined cooling, heating and power system has certain inertia and constraints. It usually takes 15-20 minutes for the gas turbine to start up and run at full load, and the charging and discharging power of the energy storage battery is limited by the rated capacity. Therefore, the scheduling strategy must be forward-looking, anticipating possible supply and demand imbalances and preparing in advance. This predictive ability is obtained by the reinforcement learning algorithm through extensive training. The energy storage system plays a key buffering role in source-load coordination. When the distributed energy output exceeds the demand, the excess power is stored in the battery pack; when there is a supply gap, the energy storage system quickly discharges to meet the load demand. The response time of modern energy storage systems can reach milliseconds, much faster than the adjustment speed of traditional power generation equipment. By reasonably configuring the energy storage capacity and charging and discharging strategy, the volatility of distributed energy can be effectively smoothed.

[0054] Preferably, the real-time synchronization of the digital twin model ensures the consistency between the virtual world and the physical world. Whenever the state of a device in the physical system changes, the corresponding parameters are transmitted to the digital twin platform through an industrial Ethernet or wireless communication network. These parameters include the real-time power of various power generation devices, the state of charge of energy storage systems, the voltage and frequency of key nodes, and so on. After receiving these data, the digital twin model updates the operating state of the virtual devices and recalculates the energy flow of the entire network. The standardization of communication protocols is the guarantee of reliable data transmission. Common protocols include Modbus, IEC61850, and other industrial standards, which define the format, transmission method, and error handling mechanism of data. Each update message contains information such as timestamp, device identification, parameter type, and value, ensuring data integrity and traceability. This intelligent scheduling method based on reinforcement learning can better adapt to complex and variable operating environments compared to traditional rule-based scheduling. Through continuous learning and optimization, the system's supply-demand matching degree can be improved to more than 95%, and energy utilization efficiency can be improved by about 20%. At the same time, due to the reduction of frequent start-stop and large-scale adjustment of devices, the service life of the devices is also prolonged.

[0055] S106, the adjusted system operation parameters and real-time output data of distributed energy are obtained, and the load distribution inside the building group is cooperatively scheduled to minimize the total energy consumption and carbon emissions of the building group as the objective function, to generate an optimized load distribution scheme.

[0056] The adjusted combined cooling heating and power system operation parameters and real-time output data of distributed energy are obtained, including the current total power, total heat and total cold force values available for distribution, while collecting the energy consumption demand types, demand levels and load adjustable ranges of each building in the building group, taking the proportion of adjustable load of each building in the total load as the flexibility coefficient to obtain the building load feature set containing demand and flexibility coefficient. According to the demand difference and flexibility coefficient of each building in the building load feature set, the electric, heat and cold load distribution of each building is mapped to the particle position vector in the particle swarm optimization algorithm, the particle velocity update formula and position update formula are set by using the particle swarm optimization algorithm, the total energy consumption and carbon emissions corresponding to each particle are calculated as the fitness value, and the particle position set with the minimum fitness is obtained by iterative optimization. For the optimal solution in the particle position set, the corresponding power, heat and cold distribution values of each building are analyzed, the distribution values are adjusted by constraining the upper and lower limits of the preset demand, and the total energy consumption value and total carbon emission value of the building group are accumulated by multiplying the adjusted distribution values by the corresponding energy consumption coefficient and carbon emission factor. The total energy consumption value and total carbon emission value of the building group are used to generate a scheduling instruction table containing building number, time period identifier, power distribution, heat distribution and cold distribution according to the time sequence, forming an executable load distribution optimization scheme.

[0057] For example, the coordinated scheduling of loads within a building complex is a key step in achieving overall energy efficiency optimization. Different types of buildings have differentiated energy consumption characteristics and load flexibility, which provides space for optimization scheduling. The air conditioning load of an office building has high adjustability, while the computing load of a data center is relatively rigid. By reasonably allocating limited energy resources, overall efficiency maximization can be achieved while meeting the basic needs of each building.

[0058] Specifically, the calculation of the load flexibility coefficient reflects the ability of each building to participate in demand response. For an office building equipped with an intelligent air conditioning system, the refrigeration load can be adjusted by ±20% while ensuring that the indoor temperature is within the range of 24-26 degrees. Therefore, the cold load flexibility coefficient of this building is 0.2. In contrast, the temperature and humidity requirements of a hospital operating room are strict, and its air conditioning load cannot be adjusted, with a flexibility coefficient close to 0. This differentiated flexibility feature provides scheduling space for optimization algorithms.

[0059] In one possible implementation, the construction of the building load feature set needs to consider multiple dimensions. In addition to load flexibility, it also includes the baseline demand, peak demand, and energy consumption priority of each building. The lighting and air conditioning demand of a shopping mall during business hours belongs to high priority, while the ventilation load of an underground garage can be appropriately reduced. By integrating these multi-dimensional information, a complete portrait of each building is formed, laying the foundation for subsequent optimization allocation. The application of the particle swarm optimization algorithm in load allocation problems reflects the advantages of swarm intelligence. The algorithm regards each possible allocation scheme as a particle, and the particle moves in the multi-dimensional solution space to find the optimal position. The position vector of each particle contains the load allocation values of all buildings, such as the position of a particle may be represented as: office building A allocates 300kW of power, 150kW of heat, and 200kW of cold; shopping mall B allocates 500kW of power, 100kW of heat, and 400kW of cold... and so on.

[0060] It's important to note that particle movement follows specific rules. Each particle records its own historical best position while also monitoring the overall group's best position, updating its velocity and position by combining individual and group experience. Velocity updates consider inertia weights, individual learning factors, and social learning factors, ensuring particles maintain a degree of exploratory activity while converging towards desirable areas. This mechanism allows the algorithm to strike a balance between global search and refined local search. The design of the fitness function directly impacts the optimization performance. In this scenario, the fitness value is a weighted average of total energy consumption and total carbon emissions. Energy consumption calculations need to consider the conversion efficiency of different energy forms; electricity is highly efficient when used directly, while heat and cold energy suffer some losses during transmission. Carbon emissions are closely related to the energy source; using clean energy sources like solar and wind power results in zero carbon emissions, while gas-fired power generation produces corresponding carbon emissions. By setting appropriate weighting coefficients, a balance can be found between energy efficiency and environmental protection.

[0061] Preferably, the handling of constraints ensures the feasibility of the allocation scheme. Each building has its upper and lower limits for operation, which are derived from equipment capacity, comfort standards, or production process requirements. When the allocation value given by the optimization algorithm exceeds these limits, the system will automatically correct it. For example, if the minimum power demand of a teaching building is 100kW to maintain basic lighting and equipment operation, even if the algorithm suggests allocating 80kW to reduce total energy consumption, the system will adjust it to 100kW. The generation of the scheduling instruction table is the transformation of the optimization result into actual control. This table records in detail the energy supply that each building should receive in each time period, forming an executable operation guide. The table not only contains numerical information but also indicates the execution priority and timing relationship. When the total supply capacity suddenly decreases in a certain time period, the system will prioritize the supply to high-priority buildings and reduce the load of other buildings in order of flexibility coefficient. This collaborative scheduling method based on particle swarm optimization can significantly improve energy utilization efficiency compared to traditional on-demand allocation or average allocation. By fully exploring the load flexibility potential of different buildings, total energy consumption can be reduced by 10-15% and carbon emissions by 15-20% without affecting user comfort. At the same time, by taking into account the differentiated needs of each building, the supply and demand contradictions caused by a "one-size-fits-all" approach are avoided, thus improving user satisfaction.

[0062] S107. Extract carbon emission data from the optimized load allocation scheme, use a digital twin model to predict the total carbon emission and peak change trend, adjust the threshold and constraints of carbon emission control, and generate the final energy dispatch scheme.

[0063] The energy consumption of each building and the corresponding carbon emission coefficient are extracted from the load distribution optimization scheme. By calculating the carbon dioxide emissions generated by each building using electricity, heat, and cold, and combining real-time fluctuation data of regional energy supply, including renewable energy output changes and traditional energy supply fluctuations, dynamic carbon emission data containing time series are obtained. According to the dynamic carbon emission data, the current carbon emission value and the historical carbon emission sequence are input into the digital twin model. The moving average method is used to predict the trend of total carbon emissions and the time of peak value occurrence in the future period. If the predicted peak value exceeds the preset upper limit of carbon emissions, the carbon emission threshold of each building is lowered to obtain a more stringent carbon control index set. Using the more stringent carbon control index set, the energy supply structure of each building is redistributed, and the proportion of renewable energy supply is increased to a higher level. At the same time, according to the importance level of the load of each building, the reducible load is identified and marked. By sending reduction instructions to these loads, the power demand is reduced, and the adjusted energy demand data of each building is obtained. According to the adjusted energy demand data of each building and the real-time supply capacity of renewable energy, detailed scheduling instructions containing building number, time period, energy type, and supply amount are generated according to the principle of renewable energy first and low-carbon energy second. The instructions are sent to the control equipment of each building through the communication network, and the final energy scheduling scheme containing the execution time sequence and parameter setting is output.

[0064] For example, accurate calculation and dynamic control of carbon emissions are the core of achieving the goal of building clusters. The carbon emissions of each building not only need to be monitored in real time, but also need to predict future trends and take control measures in advance. The carbon emission coefficients of different energy types differ significantly, which provides operational space for optimizing energy structure.

[0065] Specifically, the differentiated characteristics of carbon emission coefficients determine the priority of energy selection. When using grid power, if the local power grid is mainly supplied by thermal power plants, each consumption of 1 kWh of electricity will generate about 0.8 kg of carbon dioxide emissions. Each cubic meter of natural gas produces about 2.2 kg of carbon dioxide. In contrast, the carbon emission coefficient of renewable energy such as photovoltaic and wind power is zero. This difference allows significant changes in total carbon emissions by adjusting the energy supply structure under the same energy demand.

[0066] In one possible implementation, the fluctuation of regional energy supply has a direct impact on carbon emissions. When the sun is shining during the day, photovoltaic power generation increases, and the proportion of clean energy in the entire region increases. At this time, increasing energy consumption will not bring additional carbon emissions; while at night or on rainy days, renewable energy output decreases, and the proportion of traditional energy increases. The same energy consumption behavior will produce more carbon emissions. This time-varying characteristic requires carbon emission control to have dynamic adaptability. The digital twin model plays a key role in carbon emission prediction. By inputting historical carbon emission data, current operating status, and future load prediction into the model, the carbon emission change curve within the next 24 hours can be simulated. Moving average method, as a simple and effective time series prediction method, predicts future values by weighted average of historical data. When the prediction shows that the carbon emission peak in a certain period will exceed the regional limit, the system needs to adjust the control strategy in advance.

[0067] It should be noted that the dynamic adjustment of carbon control indicators reflects the preventive management thinking. Traditional ex post control often leads to measures taken after carbon emissions exceed the standard, while dynamic adjustment based on prediction can tighten control thresholds in advance. When the prediction shows that there will be a carbon emission peak at 3 pm and it may exceed the standard, the system will start to gradually reduce the carbon emission permission value of each building in the morning, avoiding sharp changes that affect users through gradual adjustment. The implementation of the renewable energy first principle needs to consider multiple factors. In addition to the environmental advantage of zero carbon emissions, the instability of its supply also needs to be considered. The system will assess the availability of renewable energy in real time, and when photovoltaic output is sufficient, it will be allocated to buildings with high load demand first; when renewable energy is insufficient, it needs to reasonably allocate limited clean energy to ensure the low-carbon operation of key buildings.

[0068] Preferably, the implementation of load reduction follows a hierarchical response mechanism. The loads in the building can be classified into three levels according to their importance: the first level is the critical load such as medical equipment and data center servers, which cannot be reduced; the second level is the comfort load such as part of the air conditioning and elevators, which can be adjusted moderately; and the third level is the auxiliary load such as landscape lighting and advertising display screens, which can be greatly reduced or turned off. When it is necessary to reduce carbon emissions, the system first reduces the third level load, and if it is still insufficient, it gradually involves the second level load. The generation of the dispatching instruction needs to consider the technical feasibility and the execution effect comprehensively. Each instruction contains specific execution objects, execution time, adjustment amplitude and expected effect. The dispatching instruction received by an office building at 2 pm may include: increasing the air conditioning set temperature by 2 degrees, turning off the lighting in non-office areas, and enabling the full capacity of the roof photovoltaic system. The combination of these specific measures can reduce carbon emissions by 30% in this period on the premise of ensuring basic comfort. Compared with fixed threshold control, this carbon emission control method based on prediction and dynamic adjustment can more flexibly respond to the fluctuations of energy supply and demand. Through early prediction and gradual adjustment, sudden over-standard of carbon emissions is avoided, and renewable energy is maximized. Practice shows that using this method can reduce annual carbon emissions by more than 25% without affecting the normal use of the building.

[0069] S108, obtain running data according to the final energy dispatching scheme, calculate the energy utilization efficiency and carbon emission reduction amount of the building group through the preset energy efficiency evaluation method in the digital twin model, and obtain the execution effect evaluation result of the dispatching scheme.

[0070] From the execution record of the final energy dispatching scheme, the actual consumed electric power, heat power and cold power of each building and the supply source proportion of each type of energy are extracted, the device running parameters and energy transmission loss data during the dispatching execution are obtained, the energy consumption and supply amount of each period are summarized through the building group running data, and the building group running data containing time stamp is obtained. According to the energy input and output values in the building group running data, the ratio of total effective energy output to total energy input is calculated as the comprehensive energy utilization efficiency in the digital twin model, and the historical carbon emission benchmark value before optimization is extracted. The difference between the current total carbon emission and the benchmark value is calculated to obtain the energy efficiency percentage and the carbon emission reduction amount. The energy efficiency percentage and the carbon emission reduction amount are combined with the energy consumption values and carbon emission distribution of each building at different time periods, and the energy consumption comparison column chart and the carbon emission time sequence curve are created through the graphic generation module. The pre-optimization value, post-optimization value and change amplitude are marked in the chart to obtain the visual chart containing multi-dimensional comparison information. According to the data comparison result shown in the visual chart, the comprehensive energy utilization efficiency value, the total carbon emission reduction amount and the energy consumption change percentage of each building are summarized to generate the energy management effect evaluation result containing quantitative evaluation indexes and chart display.

[0071] The effect evaluation of the energy scheduling scheme is a key link in the whole energy management closed loop. It not only verifies the effectiveness of the optimization strategy, but also provides data support for subsequent continuous improvement. Through in-depth analysis of the massive operation data generated during the execution, the energy use status and optimization effectiveness of the building group can be comprehensively mastered.

[0072] In a possible implementation, the extraction of operation data needs to cover multiple dimensions. The execution record contains detailed energy consumption information of each building at each time period, such as that an office building consumes 150kW of electricity at 9am, of which 60% comes from photovoltaic power generation, 30% comes from the power grid, and 10% comes from energy storage discharge. This detailed data record provides a reliable basis for subsequent efficiency calculation. At the same time, device operation parameters such as real-time power of combined cooling heating and power system, gas consumption, waste heat recovery rate, and pipeline transmission loss data need to be collected synchronously and have accurate timestamps. The calculation of comprehensive energy utilization efficiency reflects the overall level of energy conversion and utilization. This index considers the overall efficiency from primary energy input to final useful energy output. When the gas turbine consumes 100 cubic meters of natural gas, generates 300kWh of electricity, and the waste heat boiler recovers 200kWh of heat energy from flue gas, and the absorption chiller uses part of the heat energy to produce 150kWh of cooling capacity, the total useful energy output is 650kWh. Considering the calorific value of natural gas and the conversion efficiency of each link, the comprehensive energy utilization rate can reach more than 75%, which is much higher than the 45% of the traditional distributed supply system.

[0073] It should be noted that the comparison before and after optimization needs to be based on the same baseline conditions. The historical carbon emission baseline value is usually selected as the average value of the month before optimization or the data of the same period last year to eliminate the influence of seasonal factors. By comparing the carbon emissions under the same weather conditions and load demand, the emission reduction effect brought by the optimization measures can be accurately evaluated. After the implementation of optimization, the monthly carbon emission of a building group decreased from 1200 tons to 900 tons, with a reduction of 300 tons and a reduction rate of 25%. Data visualization plays an irreplaceable role in effect presentation. The column chart can intuitively show the energy consumption comparison of each building. By displaying the data before and after optimization side by side, the manager can see the improvement at a glance. The time series curve shows the dynamic change process of carbon emission. The peaks and troughs of the curve reflect the emission characteristics of the peak and valley periods of energy use. In the chart design, different colors are used to distinguish the data before and after optimization, and a label box is used to highlight the key improvement points, such as "midday photovoltaic utilization rate increased by 35%" and "nighttime energy storage discharge reduced power grid dependence by 20%".

[0074] Specifically, the production of visual charts needs to take into account the audience's understanding ability. For technical personnel, detailed breakdown energy consumption analysis charts can be provided to show the consumption composition of various types of energy such as electricity, heat, and cold; for management, the focus is on presenting the overall energy efficiency improvement and cost savings. Through the combination of multi-level and multi-angle charts, a three-dimensional evaluation perspective is formed.

[0075] Preferably, the presentation of the evaluation results adopts a structured report form. The report begins with an executive summary, which summarizes the optimization effectiveness with a few key numbers: the comprehensive energy utilization rate has increased from 60% to 75%, monthly carbon emissions have been reduced by 300 tons, and energy costs have been reduced by 18%. The main part then elaborates on the energy efficiency improvement, carbon emission reduction effectiveness, and building performance, respectively, with the support of corresponding charts. Each evaluation dimension includes three elements: current value, improvement rate, and horizontal comparison. This data-driven evaluation method ensures the objectivity and credibility of the results. Through the combination of quantitative indicators and visual display, not only the effectiveness of the energy optimization strategy is verified, but also strong support is provided for management decisions. Practice shows that regular effectiveness evaluation and feedback can promote the continuous improvement of energy management level and help building groups move forward on the road of energy saving and emission reduction. The evaluation results can also serve as an important basis for applying for green building certification, carbon trading quota, etc., and produce more extensive economic and social benefits.

[0076] The above description is only the preferred embodiment of the present application and the explanation of the applied technical principles. Those skilled in the art should understand that the scope of the invention involved in the present application is not limited to the technical solutions formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the concept of the present application. For example, the above features can be replaced with the technical features disclosed in the present application (but not limited to) with similar functions to form technical solutions.

Claims

1. A digital-twin-based building energy consumption real-time monitoring and carbon emission assessment method, characterized in that, The method comprises: predicting fluctuation trends of the regional energy supply real-time data and the building group load demand data to obtain short-term supply and demand dynamic prediction results; determining node connection strength and energy transmission efficiency according to the short-term supply and demand dynamic prediction results, combining spatial distribution characteristics of building group energy consumption, judging overall supply and demand matching degree of the network, and obtaining reconstruction demand evaluation results; adjusting connection structure of the building group energy flow network through a minimum spanning tree algorithm according to the reconstruction demand evaluation results, optimizing energy distribution path between nodes, and obtaining a reconstructed network topology dynamic configuration; extracting energy supply and demand data of key nodes from the reconstructed network topology dynamic configuration, combining combined heat and power system operation parameters, calculating energy efficiency and carbon emissions under different load distribution ratios, and forming a mode switching optimization scheme; adjusting operation mode of the combined heat and power system according to the mode switching optimization scheme, matching distributed energy output changes, obtaining adjusted system operation parameters, and feeding back to the digital twin model; obtaining the adjusted system operation parameters and the distributed energy output changes, scheduling internal load distribution of the building group through a particle swarm optimization algorithm, combining differences and flexibility of building group load demand data, and obtaining an optimized load distribution scheme; extracting carbon emission data from the optimized load distribution scheme, combining fluctuations of the regional energy supply real-time data, predicting carbon emission total amount and peak change trend, adjusting carbon emission control threshold, updating building group energy dispatching instructions, and obtaining a final energy dispatching scheme.

2. The building energy consumption real-time monitoring and carbon emission evaluation method based on digital twinning according to claim 1, characterized in that, The method comprises: obtaining the regional energy supply real-time data, collecting the building group load demand data, classifying and marking according to building function types, identifying supply and demand imbalance period distribution characteristics by calculating supply and load amount deviation values of each period, obtaining supply and demand deviation time series data; for the supply and demand deviation time series data, extracting fluctuation rate characteristic parameters, fitting the fluctuation rate characteristic parameters and historical deviation data through a time series model, calculating supply prediction value and load prediction value of a future period, obtaining supply and demand prediction time series; according to the supply and demand prediction time series, calculating supply prediction value change rate and load prediction value change rate, marking supply and demand fluctuation key periods with change rates exceeding a threshold, adjusting prediction parameter weights of a next period, obtaining adjusted supply and demand prediction data; using the adjusted supply and demand prediction data, calculating a ratio of supply prediction value and load prediction value of each period as matching degree, judging supply and demand imbalance risk periods with matching degree below a threshold, and generating short-term supply and demand dynamic prediction results.

3. The building energy consumption real-time monitoring and carbon emission evaluation method based on digital twinning according to claim 1, characterized in that, The method comprises: According to the short-term supply and demand dynamic prediction result, an energy network topology graph is constructed, energy flow direction and transmission flow between each building node are calculated, node connection strength is determined, and if the node connection strength is lower than a preset threshold, the node is marked as a weak node to obtain network connectivity data containing a weak node set; according to the weak node position in the network connectivity data, in combination with the energy consumption intensity value of each building node, the transmission path loss rate between the weak node and the high energy consumption node is calculated.

4. The building energy consumption real-time monitoring and carbon emission evaluation method based on digital twinning according to claim 1, characterized in that, According to the reconstruction demand evaluation result, the connection structure of the building group energy flow network is adjusted by a minimum spanning tree algorithm, and the energy distribution path between nodes is optimized, including: According to the reconstruction demand evaluation result, a high supply node and a high demand node set are obtained, a transmission loss value between node pairs is calculated, and a candidate line set is generated; a minimum spanning tree algorithm is used to select a minimum transmission loss connection scheme, and new and deleted lines are adjusted to obtain a reconstructed network topology dynamic configuration.

5. The building energy consumption real-time monitoring and carbon emission evaluation method based on digital twinning according to claim 1, characterized in that, According to the reconstructed network topology dynamic configuration, a digital twin model is constructed, the node connection degree and the betweenness centrality index are calculated, the key nodes are screened, and the energy supply and demand data and the combined cooling heating and power system power parameters of the key nodes are extracted; a linear programming method is used to construct a target function with cooling, heating and electric load ratio as variables, to calculate the energy efficiency and carbon emission under different ratio combinations, and to obtain a load distribution scheme set; according to the load distribution scheme set, an optimal scheme that meets the conditions of energy efficiency higher than the historical average value and carbon emission lower than the standard is selected, the time period load demand change is matched, and a mode switching optimization scheme containing switching time, load ratio, expected energy efficiency and carbon emission value is generated. According to the mode switching optimization scheme, the operation mode of the combined cooling heating and power system is adjusted, the distributed energy output change is matched, the adjusted system operation parameters are obtained, and the adjusted system operation parameters are fed back to the digital twin model, including:

6. The building energy consumption real-time monitoring and carbon emission evaluation method based on digital twinning according to claim 1, characterized in that, According to the mode switching optimization scheme, the real-time output of the distributed energy and the total load demand of the building group are obtained, and the supply and demand deviation is calculated; a power adjustment instruction is generated by using a reinforcement learning algorithm, and the power of the combined cooling heating and power system is adjusted, and the digital twin model is updated synchronously. The particle swarm optimization algorithm is used to schedule the load distribution in the building group, including:

7. The building energy consumption real-time monitoring and carbon emission evaluation method based on digital twinning according to claim 1, characterized in that, The adjusted system operation parameters and the real-time output data of the distributed energy are obtained, the particle swarm optimization algorithm is used, the total energy consumption and carbon emission corresponding to each particle are calculated as fitness values, the particle position set with the minimum fitness is obtained by iterative optimization, and the optimal solution in the particle position set is analyzed to obtain the power, heat and cold distribution values corresponding to each building. The total carbon emission and peak change trend are predicted, the carbon emission control threshold is adjusted, the building group energy dispatching instruction is updated, and the final energy dispatching scheme is obtained, including:

8. The building energy consumption real-time monitoring and carbon emission evaluation method based on digital twinning according to claim 1, characterized in that, ​ Carbon emission data is extracted from the optimized load distribution scheme, combined with real-time data fluctuations of the regional energy supply, to predict carbon emission trends; the carbon emission threshold is adjusted according to the predicted peak value, the energy supply structure is redistributed, and a final energy scheduling scheme is generated.

9. The building energy consumption real-time monitoring and carbon emission evaluation method based on digital twinning according to claim 1, characterized in that, The method further comprises: obtaining operation data according to the final energy scheduling scheme, calculating the energy utilization efficiency and carbon emission reduction amount of the building group through the preset energy efficiency evaluation method in the digital twin model, and obtaining the execution effect evaluation result of the scheduling scheme.

10. The building energy consumption real-time monitoring and carbon emission assessment method based on digital twinning according to claim 9, characterized in that, The method further comprises: obtaining operation data according to the final energy scheduling scheme, calculating the energy utilization efficiency and carbon emission reduction amount of the building group through the preset energy efficiency evaluation method in the digital twin model, and obtaining the execution effect evaluation result of the scheduling scheme. The method further comprises: obtaining operation data according to the final energy scheduling scheme, calculating the energy utilization efficiency and carbon emission reduction amount of the building group through the preset energy efficiency evaluation method in the digital twin model, and obtaining the execution effect evaluation result of the scheduling scheme. The method further comprises: obtaining operation data according to the final energy scheduling scheme, calculating the energy utilization efficiency and carbon emission reduction amount of the building group through the preset energy efficiency evaluation method in the digital twin model, and obtaining the execution effect evaluation result of the scheduling scheme. The method further comprises: obtaining operation data according to the final energy scheduling scheme, calculating the energy utilization efficiency and carbon emission reduction amount of the building group through the preset energy efficiency evaluation method in the digital twin model, and obtaining the execution effect evaluation result of the scheduling scheme. The method further comprises: obtaining operation data according to the final energy scheduling scheme, calculating the energy utilization efficiency and carbon emission reduction amount of the building group through the preset energy efficiency evaluation method in the digital twin model, and obtaining the execution effect evaluation result of the scheduling scheme.

Images