A digital twin modeling method and system for a central air conditioning cold station

Abstract

The present application relates to a kind of digital twin modeling method and system of central air conditioning cold station, including steps S1: building demand-side cold load model, obtain the demand cooling capacity of building end;S2: building topological perception hydraulic model of delivery and distribution side, obtain the actual delivery and distribution cold capacity of delivery and distribution pipe network;S3: according to cold water chiller equipment parameter, build the cold water chiller equipment performance model of supply side, obtain the actual cooling capacity of cold water chiller equipment;S4: the demand cooling capacity, actual delivery and distribution cold capacity and actual cooling capacity are carried out uniform sampling period and time label alignment;Based on the balance relationship of cooling capacity, the demand cooling capacity, actual delivery and distribution cold capacity and actual cooling capacity are associated calculation, and build cold station digital twin model, based on the output twin prediction value of the cold station digital twin model;S5: the twin prediction value and corresponding measured value are compared, obtain the data residual between twin prediction value and measured value, based on the data residual the cold station digital twin model is modified.

Description

Technical Field

[0001] This invention relates to the field of digital twin modeling technology for central air conditioning chiller plants, specifically to a digital twin modeling method and system for central air conditioning chiller plants. Background Technology

[0002] With the development of building energy-saving technologies and digital twin technologies, establishing digital models for central air conditioning chiller plants has gradually become an important means of improving the operational efficiency of building energy systems. Central air conditioning chiller plants typically generate chilled water from chiller units, which is then transported through a pipeline network to the building's end units and released into various cooled spaces (such as corridors, offices, and elevator lobbies). Therefore, the actual operating status of the chiller plant depends not only on the cooling capacity of the equipment supply side but also on the combined influence of the pipeline network's condition and the cooling demand at the building's end units.

[0003] However, existing modeling methods for central air conditioning chiller plants typically treat the building terminals, the pipeline network, and the equipment supply end as three independent modules. The lack of effective coupling between these models leads to fragmentation, failing to accurately reflect the supply-demand matching and cooling capacity transfer relationships during actual operation. For example, in practice, the predicted cooling demand at the building terminals is often used to replace the actual cooling capacity of the equipment, and operating indicators such as equipment load rate and coefficient of performance (COP) are calculated. However, this ignores the differences between the actual cooling demand at the building terminals, losses during pipeline transmission, and the actual operating status of the equipment. This results in several problems: when the actual cooling demand is lower than the predicted demand, the equipment load rate is overestimated; when equipment ages or its performance deteriorates, the energy efficiency indicators calculated based on the predicted cooling demand fail to reflect the true performance of the equipment. Ultimately, this leads to inaccurate energy consumption calculations and distorted supply-demand assessments, making it difficult to provide a reliable computing environment for chiller plant operation monitoring and energy efficiency diagnosis.

[0004] Therefore, it is necessary to improve upon existing technologies and provide a digital twin modeling method and system for central air conditioning chiller plants that can better couple the building end, transmission pipeline network and equipment supply end to truly reflect the actual operating status of the chiller plant. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a digital twin modeling method and system that can more realistically reflect the actual operating status of a central air conditioning chiller plant.

[0006] To solve the above-mentioned technical problems, the technical solution used in this invention is as follows:

[0007] A digital twin modeling method for a central air conditioning chiller plant includes the following steps:

[0008] S1: Based on the operational data of the building terminals, construct a demand-side cooling load model to obtain the required cooling capacity of the building terminals;

[0009] S2: Construct a topology-aware hydraulic model on the transmission and distribution side to obtain the actual transmission and distribution cooling capacity of the transmission and distribution network;

[0010] S3: Based on the parameters of the chiller unit equipment, construct a performance model of the chiller unit equipment on the supply side to obtain the actual cooling capacity of the chiller unit equipment;

[0011] S4: Align the sampling period and time label of the required cooling capacity, actual distributed cooling capacity and actual supplied cooling capacity with a unified system; perform correlation calculations on the required cooling capacity, actual distributed cooling capacity and actual supplied cooling capacity based on the cooling capacity balance relationship, and construct a digital twin model of the cooling station; output twin prediction values ​​based on the digital twin model of the cooling station.

[0012] S5: Compare the twin prediction value with the corresponding measured value to obtain the data residual between the twin prediction value and the measured value, and dynamically correct the digital twin model of the cold station based on the data residual.

[0013] Preferably, in step S1, the building terminal operation data includes one or more of the following: meteorological data, historical cooling load, indoor status, and time characteristics;

[0014] The required cooling capacity at the building's terminal can be expressed as:

[0015]

[0016] Where t represents the time variable, and H represents the prediction range. This represents the demand-side time-series KAN model. Indicates demand-side timing input. Indicates model parameters.

[0017] Preferably, step S2 further includes constructing a chilled water pipe network topology map, which specifically includes the following steps:

[0018] S201: Show chiller units, water pumps, valves, main pipes and key terminals as nodes in the diagram;

[0019] S202: The pipe segments connecting each node are represented as graph edges;

[0020] S203: Embed the parameter information of the pipe segments into the graph structure to form a chilled water pipe network topology graph. Construct the topology-aware hydraulic model based on the chilled water pipe network topology graph. The expression formula of the chilled water pipe network topology graph is:

[0021]

[0022] in, This represents the set of nodes consisting of chiller units, pumps, valves, main pipes, and key terminals. This represents the set of pipe segments that connect the nodes.

[0023] Preferably, the parameter information of the pipe section includes one or more of the following: pipe diameter, length, and resistance coefficient.

[0024] Preferably, the topology-aware hydraulic model reconstructs the hydraulic state of the transmission and distribution network based on the dynamic operating parameters of the chilled water and the chilled water network topology. The hydraulic state includes one or more of the following: chilled water flow rate, main pipe differential pressure, and return water temperature.

[0025] The dynamic operating parameters of the chilled water include one or more of the following: pump frequency, equipment start / stop status, valve opening, differential pressure target, and water supply temperature.

[0026] Obtain the chilled water flow rate and supply-return water temperature difference at the distribution boundary; wherein, the chilled water flow rate and supply-return water temperature difference are the reconstructed values ​​and / or measured values ​​obtained by reconstructing through the topology-sensing hydraulic model;

[0027] Based on the chilled water flow rate and the supply and return water temperature difference, the actual cooling capacity of the distribution network is determined as follows:

[0028]

[0029] Where t represents the time variable, This indicates the density of water. This represents the specific heat capacity of water at constant pressure. Indicates the volumetric flow rate of chilled water. Indicates the return water temperature. This indicates the water supply temperature.

[0030] Preferably, in step S3, the formula for calculating the actual cooling capacity of the chiller unit is:

[0031]

[0032] Where t represents the time variable, C represents the density of water. P This represents the specific heat capacity of water at constant pressure. This indicates the chilled water volumetric flow rate on the evaporator side of the chiller unit. This indicates the inlet water temperature on the evaporator side. This indicates the outlet water temperature on the evaporator side.

[0033] Preferably, the actual load rate and coefficient of performance (COP) of the chiller unit are determined based on the actual cooling capacity of the chiller unit.

[0034]

[0035] in, Indicates the actual load rate. Indicates the coefficient of performance for cooling. This indicates the actual cooling capacity of the chiller unit. This indicates the rated cooling capacity of the chiller unit. This indicates the power of the chiller unit.

[0036] Preferably, in step S4, the formulas for the difference in cooling capacity between the required cooling capacity and the actual cooling capacity distributed, and the residual difference in cooling capacity between the actual cooling capacity supplied and the actual cooling capacity distributed, are as follows:

[0037]

[0038]

[0039] in, This indicates the actual cooling capacity distributed on the distribution side. This indicates the cooling demand on the demand side. This represents the actual cooling capacity supplied by the equipment, where t represents the time variable.

[0040] The cooling capacity status of the central air conditioning chiller is determined based on the cooling capacity difference between the demand cooling capacity and the actual cooling capacity delivered, and the cooling capacity residual difference between the actual cooling capacity supplied and the actual cooling capacity delivered.

[0041] Preferably, the dynamic correction of the chiller plant digital twin model includes incremental updates to the chiller unit equipment performance model based on an elastic weight consolidation mechanism, wherein the objective function of the incremental update is:

[0042]

[0043] in, This represents the total loss from incremental updates. This represents the fitting loss for the newly added field data. Indicates constraint strength. Indicates the importance of the parameter. Indicates the current model parameters. This represents the basic model parameters.

[0044] A digital twin modeling system for a central air conditioning chiller station, executing the modeling method as described in any of the preferred embodiments above, characterized in that the system comprises:

[0045] The demand-side modeling module is used to construct a demand-side cooling load model based on the operational data of the building's terminals, and to obtain the required cooling capacity of the building's terminals.

[0046] The transmission and distribution side modeling module is used to construct a topology-aware hydraulic model of the transmission and distribution side to obtain the actual transmission and distribution cooling capacity of the transmission and distribution network;

[0047] The supply-side modeling module is used to construct a performance model of the chiller equipment on the supply side based on the parameters of the chiller equipment, and to obtain the actual cooling capacity of the chiller equipment.

[0048] The three-sided coupling module is used to align the sampling period and time label of the required cooling capacity, the actual cooling capacity distributed, and the actual cooling capacity supplied, and to perform correlation calculations on the three based on the cooling capacity balance relationship to construct a digital twin model of the cooling station and output twin prediction values.

[0049] The calibration module is used to compare the twin prediction value with the corresponding field measured value, obtain the data residual between the twin prediction value and the field measured value, and dynamically correct the digital twin model of the cooling station based on the data residual.

[0050] The beneficial effects of this invention are mainly reflected in the following aspects: By unifying and calculating the demand-side cooling demand, the actual cooling capacity distributed on the transmission and distribution side, and the actual cooling capacity supplied by the equipment on the supply side, this invention eliminates the need to treat the demand side, transmission and distribution side, and supply side as independent modules. When changes occur in building terminal demand, pipeline hydraulic conditions, or equipment performance, the model can reflect the continuous impact of changes in one link on other links. This avoids the use of predicted cooling capacity on the demand side to replace the actual cooling capacity of the equipment in calculating load rate, coefficient of performance, or equipment power, as is done in existing technologies. This further improves the accuracy of supply and demand status judgment and energy consumption calculation for chiller plants. Simultaneously, it provides a more complete and reliable operational calculation environment for subsequent chiller plant operation monitoring, energy efficiency diagnosis, and operating condition simulation. Attached Figure Description

[0051] The above and other objects, features, and advantages of the invention will become clearer through a more detailed description of the preferred embodiments illustrated in the accompanying drawings. The same reference numerals denote the same parts throughout the drawings, and the drawings are not intentionally drawn to scale with actual dimensions; the focus is on illustrating the gist of the invention.

[0052] Figure 1 This is a schematic diagram illustrating the steps of the digital twin modeling method for a central air conditioning chiller station according to the present invention.

[0053] Figure 2 This is a schematic diagram of the overall process of the digital twin modeling method for central air conditioning chiller stations according to the present invention;

[0054] Figure 3 This is a schematic diagram showing the physical boundaries and relationships between the required cooling capacity, the actual cooling capacity distributed, and the actual cooling capacity supplied by the equipment in this invention.

[0055] Figure 4This is a schematic diagram of the demand-side cooling load modeling structure in this invention;

[0056] Figure 5 This is a schematic diagram of the topology-sensing hydraulic operation reconstruction on the transmission and distribution side in this invention;

[0057] Figure 6 This is a schematic diagram of supply-side equipment performance modeling and incremental updates in this invention;

[0058] Figure 7 This is a schematic diagram of the three-sided coupling and calibration process in this invention. Detailed Implementation

[0059] To facilitate understanding of the present invention, a more comprehensive description of the present invention will be given below with reference to the accompanying drawings.

[0060] It should be noted that when a component is considered to be "connected" to another component, it can be directly connected to and integrated with the other component, or there may be an intervening component present. The terms "mounted," "one end," "the other end," and similar expressions used in this document are for illustrative purposes only.

[0061] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0062] refer to Figure 1-7 This invention provides a digital twin modeling method for a central air conditioning chiller plant, comprising the following steps:

[0063] S1: Based on the operational data of the building's terminals, a demand-side cooling load model is constructed to obtain the required cooling capacity of the building's terminals. Specifically, this can be achieved by connecting with the building's control system, sensor network, and meteorological service API to obtain operational data from the building's terminals. This operational data includes one or more of the following: meteorological data (e.g., outdoor dry-bulb temperature, wet-bulb temperature, humidity, solar radiation intensity, etc.), historical cooling load, indoor conditions (e.g., indoor temperature, humidity, and other indoor condition parameters in various areas of the building, as well as occupancy density distribution, etc.), and time characteristics (e.g., hour, day of the week, whether it is a holiday, etc.). It can also include feedback information from end-users, such as user-manually adjusted air conditioning settings. The demand-side cooling load model utilizes the above data input and learns through a time-series KAN model. The model consists of multiple layers of learnable spline functions, enabling it to adaptively learn the nonlinear mapping relationship between each input feature and the cooling load. It also outputs the cooling demand for a future preset time period (e.g., Figure 4 As shown, it is a structural diagram of demand-side time-series cooling load modeling, which aims to accurately predict the cooling demand of a building in a future preset period (e.g., the next 24 hours).

[0064] S2: Construct a topology-aware hydraulic model on the transmission and distribution side to obtain the actual transmission and distribution cooling capacity of the transmission and distribution network;

[0065] S3: Based on the parameters of the chiller unit equipment, construct a performance model of the chiller unit equipment on the supply side to obtain the actual cooling capacity of the chiller unit equipment;

[0066] S4: Align the required cooling capacity, actual distributed cooling capacity, and actual supplied cooling capacity with a unified sampling period and time stamp. For example, these three types of cooling capacity can be unified into a data point every 5 minutes and aligned with the same timestamp, allowing for comparison and calculation of the three types of cooling capacity at the same point in time. After alignment, based on the cooling capacity balance relationship, perform correlation calculations on the required cooling capacity, actual distributed cooling capacity, and actual supplied cooling capacity. The cooling capacity balance relationship refers to the ideal situation where the required cooling capacity at the building's end, the actual distributed cooling capacity of the distribution network, and the actual supplied cooling capacity of the equipment should satisfy the cooling capacity transfer and balance relationship. However, in reality, there will be differences among the three due to factors such as transmission losses and heat storage and release in the network. By performing correlation calculations on the three types of cooling capacity under the same time reference, the supply and demand matching relationship, cooling capacity transfer relationship, and operational status differences among the three sides (i.e., the demand side, the distribution side, and the supply side) can be better determined.

[0067] Furthermore, based on this, the demand-side cooling capacity, the distribution-side cooling capacity, and the supply-side cooling capacity are coupled to construct a digital twin model of the chiller plant. This digital twin model is used to digitally represent the actual operating status of the central air conditioning chiller plant, and outputs twin prediction values ​​based on the model. These twin prediction values ​​include one or more of the following: demand-side cooling capacity, actual distribution-side cooling capacity, actual equipment cooling capacity, chilled water flow rate, main pipe pressure difference, return water temperature, equipment power, and coefficient of performance (COP). This forms a continuous operational model between the demand side, distribution side, and supply side. Traditional methods typically model these three modules independently, creating a fragmented approach that easily substitutes predicted demand for actual cooling capacity in calculating load factor, COP, or equipment power, leading to estimation errors in supply and demand status and energy consumption status. In this embodiment, by unifying the demand for cooling capacity, the actual distribution of cooling capacity, and the actual supply of cooling capacity by the equipment under the same sampling period and time label, and performing correlation calculations on the three based on the cooling capacity balance relationship, the demand side, the distribution side, and the supply side are no longer treated as independent modules. Instead, they can continuously represent the entire process of cooling capacity generated by the building's end demand, distributed through the chilled water network, and supplied by the chiller unit equipment within the same cooling capacity transfer framework.

[0068] S5: Compare the twin prediction value with the corresponding measured value to obtain the data residual between the twin prediction value and the measured value. This comparison means comparing the twin prediction value output by the digital twin model of the chiller plant with the measured value of the corresponding physical quantity on-site, thereby obtaining the data residual between the two. The measured value of the corresponding physical quantity on-site mainly refers to the data measured by on-site sensors and other acquisition equipment, including but not limited to the measured cooling capacity, measured flow rate, measured pressure difference, and measured temperature corresponding to the twin prediction value. For example, the chilled water flow rate output by the model can be compared with the measured flow rate on-site to obtain the flow rate residual.

[0069] When the data residual exceeds a preset threshold, it indicates a deviation between the output of the cold station digital twin model and the actual operating status of the cold station. In this case, the cold station digital twin model can be dynamically corrected based on the data residual. In this embodiment, dynamic correction of the cold station digital twin model includes one or more of the following: correcting the model state, reducing the weight of abnormal measurement points, and updating model parameters. This ensures that the cold station digital twin model is continuously updated and can more realistically reflect the operating status of the cold station. The residual threshold, the abnormal measurement point weighting method, and the parameter update intensity can be set according to the on-site working conditions, measurement reliability, and model deviation degree. Specific configurations do not constitute a limitation of this invention.

[0070] Overall, it can be used as a reference Figure 2 This paper illustrates the overall process of the digital twin modeling method for central air conditioning chiller plants according to the present invention. The core of this method lies in dividing the complex central air conditioning chiller plant system into three interconnected but independently modelable subsystems: the demand side, the transmission and distribution side, and the supply side. Through a calibration mechanism, the long-term consistency between the entire twin model and the physical entity is ensured.

[0071] Therefore, the digital twin modeling method for central air conditioning chiller stations provided by this invention has the following advantages compared with the prior art:

[0072] The digital twin modeling method for central air conditioning chiller plants provided by this invention unifies and correlates the demand-side cooling capacity, the actual distribution-side cooling capacity, and the actual supply-side cooling capacity. This eliminates the need to treat the demand, distribution, and supply sides as independent modules. When changes occur in building terminal demand, network hydraulic conditions, or equipment performance, the model can reflect the continuous impact of changes in one aspect on other aspects. This avoids the use of predicted cooling capacity from the demand side to calculate load factor, coefficient of performance (COP), or equipment power, as is common in existing technologies. This further improves the accuracy of chiller plant supply and demand status assessment and energy consumption calculation. Simultaneously, it provides a more complete and reliable operational computing environment for subsequent chiller plant operation monitoring, energy efficiency diagnosis, and operational simulation.

[0073] In a preferred embodiment, the required cooling capacity at the building terminal is calculated using the following formula:

[0074]

[0075] Where t represents the time variable, and H represents the prediction range. This represents the demand-side time-series KAN model. Indicates demand-side timing input. Indicates model parameters.

[0076] Specifically, demand-side timing input It can consist of operational data from the building's terminals, including one or more of the following: meteorological data, historical cooling load, indoor status, time characteristics, and terminal feedback. For example, meteorological data may include outdoor dry-bulb temperature, wet-bulb temperature, humidity, solar radiation intensity, etc.; historical cooling load includes the actual cooling load of the building's terminals over a past period; indoor status includes indoor temperature, humidity, and occupancy density distribution in various areas of the building; time characteristics may include hour, day of the week, and whether it is a holiday, etc.; terminal feedback may include feedback information such as users manually adjusting air conditioning settings.

[0077] Demand-side time series KAN model This is used to establish a nonlinear time-series mapping relationship between the aforementioned multi-source operational data inputs and the cooling demand at the building's terminal. Specifically, the demand-side time-series KAN model can be composed of multiple layers of learnable spline functions. Through these learnable spline functions, it adaptively learns the nonlinear relationship between factors such as meteorological disturbances, building thermal inertia, changes in indoor conditions, and changes in occupant activity and cooling demand, thereby outputting the cooling demand sequence for a future preset time period. ;

[0078] in, This is a predicted sequence of future cooling demand, where 't' represents the time variable, indicating the current sampling moment, and 'H' represents the prediction range. Therefore, 't+1' can be understood as the next sampling moment after the current moment, and 't+H' as the Hth sampling moment after the current moment. For example, when the sampling period is 5 minutes and the prediction range is 1 hour, H can be 12, meaning the model outputs a cooling demand sequence for the next 12 moments. In practical applications, the prediction range H can be determined based on actual operational needs, data sampling period, and prediction accuracy requirements; the specific value does not constitute a limitation of this invention. The specific network size, spline function settings, and training parameters of the temporal KAN can be determined based on building data characteristics, prediction time domain, and accuracy requirements; the specific parameter configuration does not constitute a limitation of this invention.

[0079] Using the above formula for calculating building end-point cooling demand, the model can also output medium- to long-term load variation trends and comfort operating boundaries to guide operation (e.g., the acceptable range of indoor temperature fluctuation while ensuring occupant comfort). The predicted cooling demand... This can serve as the "target" for subsequent system operation. It should be noted that the demand-side model only outputs the cooling demand at the building's terminal and the comfort operation boundary. It does not use the predicted demand as the actual cooling capacity of the chiller unit, nor does it directly output the equipment control actions, in order to ensure a clear boundary between the demand state and the hydraulic execution state.

[0080] In a preferred embodiment, step S2 further includes constructing a chilled water pipe network topology map, which specifically includes the following steps:

[0081] S201: Show chiller units, water pumps, valves, main pipes and key terminals as nodes in the diagram;

[0082] S202: The pipe segments connecting each node are represented as graph edges;

[0083] S203: Embed the parameter information of the pipe segments (such as pipe diameter, length, resistance coefficient, etc.) into the graph structure to form a chilled water pipe network topology graph. Construct the topology-aware hydraulic model based on the chilled water pipe network topology graph. The expression formula of the chilled water pipe network topology graph is as follows:

[0084]

[0085] in, This represents the set of nodes consisting of chiller units, pumps, valves, main pipes, and key terminals. This represents the set of pipe segments connecting the nodes. Specifically, the chilled water network of a central air conditioning chiller plant typically consists of chiller units, chilled water pumps, valves, main pipes, and building terminals, all connected by pipe segments. To utilize graph neural networks to learn the hydraulic transmission patterns in the network, this embodiment abstracts the aforementioned equipment and key hydraulic locations as graph nodes, and the pipe segments connecting the nodes as graph edges, thereby transforming the actual chilled water network into a graph structure. In this network topology diagram, nodes V represent physical devices, such as chillers, chilled water pumps, valves, and critical air conditioning terminals; edges E represent pipe segments connecting these devices and are assigned physical attributes, such as pipe diameter, length, and local resistance coefficients generated by valves, etc. Since using only nodes and edges to represent the network connections is insufficient to describe the network's hydraulic characteristics, it is necessary to embed the physical attributes of the pipe segments into the graph structure. This ensures that the chilled water network topology diagram includes not only connections but also the physical parameters required for hydraulic calculations.

[0086] In a preferred embodiment, the topology-aware hydraulic model reconstructs the hydraulic state of the transmission and distribution network based on the dynamic operating parameters of the chilled water and the chilled water network topology. The hydraulic state includes one or more of the following: chilled water flow rate, main pipe differential pressure, and return water temperature. The dynamic operating parameters of the chilled water include one or more of the following: pump frequency, equipment start-up and shutdown status, valve opening degree, differential pressure target, and supply water temperature.

[0087] It is understandable that, due to the possibility that some key nodes in the chilled water network may lack sensors or have sparse measuring points, the overall hydraulic status of the network is difficult to obtain directly. Therefore, in this embodiment, the chilled water network topology map is input into a topology-aware hydraulic model, along with dynamic operating parameters of the chilled water. These dynamic operating parameters include one or more of the following: pump frequency, equipment start / stop status, valve opening, differential pressure target, and supply water temperature. These dynamic operating parameters reflect the current hydraulic operating conditions of the chiller plant, such as... Figure 5 As shown, Figure 5 The diagram illustrates the structure of topology-aware hydraulic operation state reconstruction on the transmission and distribution side. Based on the pipeline network topology and the aforementioned dynamic operating parameters, the topology-aware hydraulic model can calculate the hydraulic state at each node (equipment) and each edge (pipe segment), even if no sensors are installed at these locations. This reconstructs hydraulic states that are difficult to measure directly, such as chilled water flow rate, main pipe differential pressure, and return water temperature. The specific number of layers, attention heads, hidden dimensions, and training parameters of the graph attention network can be determined based on the pipeline network size, topology complexity, and measurement point configuration. Specific parameter configurations do not constitute a limitation of this invention.

[0088] The chilled water flow rate and supply / return water temperature difference at the distribution boundary are obtained. These values ​​are the reconstructed values ​​and / or measured values ​​obtained through the topology-aware hydraulic model. The distribution boundary refers to the physical boundary where chilled water is transported from the chiller plant to the building's end, such as the chilled water main outlet or manifold. At the distribution boundary, if flow meters and temperature sensors are installed, the measured chilled water flow rate and supply / return water temperature difference can be directly obtained. If the measurement points are not covered or the measurement data is unavailable, the chilled water flow rate and supply / return water temperature difference reconstructed using the topology-aware hydraulic model can be used. In other words, the chilled water flow rate and supply / return water temperature difference can be entirely from actual measurements, entirely from reconstruction, or partially from actual measurements and partially from reconstruction. Based on the chilled water flow rate and supply / return water temperature difference, the actual chilled water capacity of the distribution network is determined, based on the reconstructed or measured chilled water main flow rate. and the supply and return water temperatures of the main pipe and It can accurately calculate the actual cooling capacity transported by the chilled water network at time t, that is, the actual cooling capacity distributed. :

[0089]

[0090] Where t represents the time variable, This indicates the density of water. This represents the specific heat capacity of water at constant pressure. Indicates the volumetric flow rate of chilled water. Indicates the return water temperature. Indicates the water supply temperature and the actual cooling capacity distributed. As a bridge connecting the demand side and the supply side, it represents the actual cold energy transported by the water system.

[0091] In a preferred embodiment, the formula for calculating the actual cooling capacity of the chiller unit in step S3 is as follows:

[0092]

[0093] Where t represents the time variable, This indicates the density of water. This represents the specific heat capacity of water at constant pressure. This indicates the chilled water volume flow rate on the evaporator side of the chiller unit. This indicates the inlet water temperature on the evaporator side. This indicates the outlet water temperature on the evaporator side. It's understood that during the operation of a chiller unit, heat exchange typically occurs between the evaporator and chilled water, transferring cooling capacity to the chilled water system. To obtain the actual cooling capacity of the chiller unit, effective chilled water flow rate and inlet / outlet water temperature measurements need to be obtained on the evaporator side. When a flow meter and temperature sensor are installed on the evaporator side, the chilled water volumetric flow rate on the evaporator side can be directly obtained. Inlet water temperature and outlet water temperature When some measuring points are missing, the corresponding state can be reconstructed by combining the performance model of the supply-side equipment or the hydraulic model of the transmission and distribution side.

[0094] It is understandable that in the above formula, the temperature difference between the evaporator inlet and outlet water... This reflects the degree to which the chilled water is cooled after passing through the evaporator. Through the above calculations, the actual cooling capacity of the chiller unit at each sampling moment can be obtained. Therefore, in this embodiment, the actual cooling capacity is calculated based on the actual operating parameters of the chiller unit's evaporator side, rather than using the predicted cooling capacity from the demand side. Consequently, the load factor, coefficient of performance (COP), and energy consumption indicators calculated based on the actual cooling capacity can also reflect the true operating status of the chiller unit, avoiding data distortion caused by substituting predicted cooling capacity for actual cooling capacity.

[0095] In a preferred embodiment, the actual load rate and coefficient of performance of the chiller unit are determined based on the actual cooling capacity of the chiller unit:

[0096]

[0097] in, Indicates the actual load rate. Indicates the coefficient of performance for cooling. This indicates the actual cooling capacity of the chiller unit. This indicates the rated cooling capacity of the chiller unit. This represents the power of the chiller unit, and t also represents the time variable.

[0098] Understandably, after obtaining the actual cooling capacity of the equipment, the actual load rate and coefficient of performance (COP) of the chiller unit can be further determined. The actual load rate characterizes the ratio of the chiller unit's current actual cooling capacity to its rated cooling capacity; the COP characterizes the cooling capacity provided per unit power of the chiller unit and is an important indicator for measuring equipment energy efficiency.

[0099] Since both of the above indicators are based on the actual cooling capacity of the equipment This calculation uses the actual cooling capacity supplied by the equipment as the basis, rather than relying on the predicted cooling capacity demanded by the demand side. Therefore, it can more accurately reflect the actual operating load and energy efficiency level of the chiller unit. For example, when the actual cooling capacity used by a building is lower than the predicted cooling capacity demanded, calculating the load rate based on the predicted demand would overestimate the equipment load rate; however, in this embodiment, calculating the load rate based on the actual cooling capacity supplied by the equipment avoids this problem. Similarly, when the equipment ages, scales, or its performance deteriorates, the coefficient of performance (COP) calculated based on the actual cooling capacity supplied by the equipment can also reflect the true performance changes of the equipment, providing a reliable basis for subsequent energy efficiency diagnosis.

[0100] In a preferred embodiment, in step S4, the formulas for the difference in cooling capacity between the required cooling capacity and the actual cooling capacity distributed, and the residual difference in cooling capacity between the actual cooling capacity supplied and the actual cooling capacity distributed, are as follows:

[0101]

[0102]

[0103] in, This indicates the actual cooling capacity distributed on the distribution side. This indicates the cooling demand on the demand side. This represents the actual cooling capacity supplied by the equipment, where t represents the time variable.

[0104] The cooling capacity status of a central air conditioning chiller is determined based on the cooling capacity difference between demanded cooling capacity and actual distributed cooling capacity, and the cooling capacity residual between actual supplied cooling capacity and actual distributed cooling capacity. For example, the cooling capacity status of a central air conditioning chiller includes, but is not limited to, supply and demand balance, excessive cooling supply, insufficient cooling supply, hydraulic imbalance, and excessive distribution.

[0105] Specifically, after unifying the demand for cooling capacity, the actual cooling capacity distributed, and the actual cooling capacity supplied by the equipment to the same sampling period and time label, the cooling capacity difference between the demand side and the distribution side, as well as the cooling capacity residual between the supply side and the distribution side, can be calculated separately.

[0106] in, This represents the difference between the required cooling capacity and the actual cooling capacity distributed, used to characterize the degree of supply-demand balance between building end-user demand and actual chilled water distribution. When the temperature is close to zero, it indicates that the actual cooling capacity supplied and distributed is basically matched with the cooling capacity demanded, and supply and demand are in balance; when... When the value remains positive, it indicates that the actual cooling capacity being supplied exceeds the required cooling capacity, indicating an over-cooling problem; while when... A continuously negative value indicates that the actual cooling capacity delivered is less than the required cooling capacity, resulting in insufficient cooling supply.

[0107] This represents the residual cooling capacity between the actual cooling capacity supplied by the equipment and the actual cooling capacity distributed. It characterizes the degree of inconsistency between the cooling capacity at the supply-side boundary and the distribution-side boundary. This residual may be caused by factors such as boundary differences, heat storage in the pipeline network, measurement errors, or dynamic delays. Ideally, this value should be close to zero (considering pipeline heat loss); when... A consistently positive value may indicate heat intrusion or bypass in the pipe network; when A negative value may indicate a cold storage effect in the pipeline network. Analyzing these differences allows for a clearer assessment of complex system-level operating conditions such as supply and demand imbalance, hydraulic imbalance, or equipment inefficiency, providing a basis for subsequent energy efficiency diagnostics and online calibration.

[0108] In a preferred embodiment, dynamically correcting the digital twin model of the chiller plant includes incrementally updating the performance model of the chiller unit based on an elastic weight consolidation mechanism. The objective function of the incremental update is:

[0109]

[0110] in, This represents the total loss from incremental updates. This represents the fitting loss for the newly added field data. Indicates constraint strength. Indicates the importance of the parameter. Indicates the current model parameters. This represents the basic model parameters.

[0111] In actual operation, chiller units experience performance drift due to scaling, wear, and other factors after prolonged use, causing the equipment performance model trained on historical data to gradually deviate from the actual equipment condition. Therefore, to improve the long-term adaptability of the model, this embodiment performs incremental updates to the chiller unit equipment performance model based on an elastic weight consolidation mechanism. For example... Figure 6 The diagram illustrates the supply-side equipment performance modeling and incremental updates. In this embodiment, a two-stage modeling strategy can be adopted: the first stage involves basic performance modeling. That is, using performance sample data and common rated parameters provided by the manufacturer at the time of equipment delivery, a basic equipment performance model is trained offline. This model is preferably a KAN model, as it can capture the complex nonlinear relationship between equipment performance and changes in load, water temperature, flow rate, etc., with fewer parameters. The second stage involves appropriate on-site adjustments. As mentioned above, during actual operation, equipment performance will slowly decay or drift due to scaling, wear, etc. To ensure the model can keep up with these changes, a constrained incremental update mechanism is introduced in this embodiment.

[0112] Specifically, during incremental updates, the equipment performance model is adapted using newly added field operational data, whereby... This represents the fitting loss for newly added field data, used to adapt the model to new operating conditions and changes in equipment performance. Simultaneously, to prevent the model from forgetting patterns learned from historical data when learning new data, an elastic weight consolidation term is introduced. .

[0113] in, It indicates the importance of parameters and is used to measure the importance of each model parameter to existing working condition laws; Represents the basic model parameters. Indicates the current model parameters; The constraint strength is used to balance learning new knowledge and retaining old knowledge. Through the above objective function, the model can adapt to equipment performance drift while limiting changes in key parameters and retaining the performance patterns learned under existing operating conditions. Therefore, the chiller unit equipment performance model can continuously output state variables such as equipment power consumption, coefficient of performance (COP), and outlet water temperature, improving the long-term accuracy of the chiller plant digital twin model.

[0114] A digital twin modeling system for a central air conditioning chiller station, performing the modeling method as described in any of the preferred embodiments above, the system comprising:

[0115] The demand-side modeling module is used to construct a demand-side cooling load model based on the operational data of the building's terminals, and to obtain the required cooling capacity of the building's terminals.

[0116] The transmission and distribution side modeling module is used to construct a topology-aware hydraulic model of the transmission and distribution side to obtain the actual transmission and distribution cooling capacity of the transmission and distribution network;

[0117] The supply-side modeling module is used to construct a performance model of the chiller equipment on the supply side based on the parameters of the chiller equipment, so as to obtain the actual cooling capacity of the chiller equipment.

[0118] The three-sided coupling module is used to align the sampling period and time label of the required cooling capacity, actual distributed cooling capacity, and actual supplied cooling capacity, and to perform correlation calculations on the three based on the cooling capacity balance relationship to construct a digital twin model of the cooling station and output twin prediction values. Figure 3 As shown, Figure 3 The invention clearly defines the three types of cooling capacity and their physical boundaries. Among them, the required cooling capacity... Generated at the building's end, this is the target the system should meet; the actual cooling capacity distributed... It exists within the chilled water piping network and is the actual carrier for cold energy transfer; the actual cooling capacity supplied by the equipment Originating in the evaporator of the chiller unit, it is the source of cooling capacity. The coupling module establishes a unified state interface for the demand side, distribution side, and supply side to receive outputs from the three-sided modeling modules. Based on the cooling capacity balance and hydraulic transfer relationships, it forms a three-sided coupled operating state. Through the coupling module, complex operating states such as supply and demand imbalance, hydraulic imbalance, or equipment inefficiency can be identified. Finally, the module outputs a unified operating state that includes the states of each side, coupling relationships, and state identification results, and outputs twin prediction values. For example, it outputs predicted total power consumption of the chiller system, temperature at key points, flow rate, and other data.

[0119] The calibration module is used to compare the twin prediction values ​​with the corresponding field measured values ​​to obtain the data residuals between the twin prediction values ​​and the field measured values, and to dynamically correct the cooling plant digital twin model based on the data residuals. Figure 7 As shown, this diagram illustrates the coupling of three operating states and online calibration. The calibration module continuously compares the predicted values ​​output by the twin model (e.g., chiller power consumption predicted by the supply-side model, total pipe flow reconstructed by the distribution-side model, cooling load predicted by the demand-side model, etc.) with the measured values ​​collected from the physical system (e.g., measured by sensors) and calculates the residual between the two. When the residual exceeds a preset threshold, the online calibration mechanism is triggered. This mechanism can perform one or more of the following operations depending on the source and characteristics of the residual:

[0120] 1. State Deviation Correction: For some short-term deviations caused by the cumulative error of the model, the current state output of the model is directly corrected to align it with the measured value.

[0121] 2. Abnormal Measurement Point Reduction: If it is diagnosed that the expansion of a certain residual is caused by a specific sensor failure or data quality problem, the weight of the measurement point data will be temporarily reduced in subsequent calculations and model updates, or even isolated, to prevent contamination of the model.

[0122] III. Constrained Update of Model Parameters: If it is diagnosed that the residuals are caused by model parameters no longer accurately reflecting physical reality (such as equipment performance drift), then an update of the corresponding model parameters is triggered. For example, the elastic weight consolidation constraint update described above is performed on the parameters of the supply-side model.

[0123] This enables the final digital twin model to operate synchronously and adaptively with the physical central air conditioning chiller, providing a more realistic and computable virtual environment for subsequent operation monitoring, energy efficiency diagnosis, and optimized control.

[0124] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A digital twin modeling method for a central air conditioning chiller station, characterized in that: Includes the following steps, S1: Based on the operational data of the building terminals, construct a demand-side cooling load model to obtain the required cooling capacity of the building terminals; S2: Construct a topology-aware hydraulic model on the transmission and distribution side to obtain the actual transmission and distribution cooling capacity of the transmission and distribution network; S3: Based on the parameters of the chiller unit equipment, construct a performance model of the chiller unit equipment on the supply side to obtain the actual cooling capacity of the chiller unit equipment; S4: Align the sampling period and time stamp of the required cooling capacity, actual distributed cooling capacity and actual supplied cooling capacity; Based on the cooling balance relationship, the required cooling capacity, the actual cooling capacity distributed, and the actual cooling capacity supplied are correlated and calculated, and a digital twin model of the cooling station is constructed. Based on the digital twin model of the cooling station, the twin prediction value is output. S5: Compare the twin prediction value with the corresponding measured value to obtain the data residual between the twin prediction value and the measured value, and dynamically correct the digital twin model of the cold station based on the data residual.

2. The digital twin modeling method for a central air conditioning chiller station as described in claim 1, characterized in that, In step S1, the building terminal operation data includes one or more of the following: meteorological data, historical cooling load, indoor status, and time characteristics; The required cooling capacity at the building's terminal can be expressed as: Where t represents the time variable, and H represents the prediction range. This represents the demand-side time-series KAN model. Indicates demand-side timing input. Indicates model parameters.

3. The digital twin modeling method for a central air conditioning chiller station as described in claim 1, characterized in that, Step S2 also includes constructing a chilled water pipe network topology map, which specifically includes the following steps: S201: Show chiller units, water pumps, valves, main pipes and key terminals as nodes in the diagram; S202: The pipe segments connecting each node are represented as graph edges; S203: Embed the parameter information of the pipe segments into the graph structure to form a chilled water pipe network topology graph. Construct the topology-aware hydraulic model based on the chilled water pipe network topology graph. The expression formula of the chilled water pipe network topology graph is: in, This represents the set of nodes consisting of chiller units, pumps, valves, main pipes, and key terminals. This represents the set of pipe segments that connect the nodes.

4. The digital twin modeling method for a central air conditioning chiller station as described in claim 3, characterized in that, The parameters of the pipe section include one or more of the following: pipe diameter, length, and resistance coefficient.

5. The digital twin modeling method for a central air conditioning chiller station as described in claim 3, characterized in that, The topology-aware hydraulic model reconstructs the hydraulic state of the transmission and distribution network based on the dynamic operating parameters of the chilled water and the chilled water network topology. The hydraulic state includes one or more of the following: chilled water flow rate, main pipe pressure difference, and return water temperature. The dynamic operating parameters of the chilled water include one or more of the following: pump frequency, equipment start / stop status, valve opening, differential pressure target, and water supply temperature. Obtain the chilled water flow rate and supply-return water temperature difference at the distribution boundary; wherein, the chilled water flow rate and supply-return water temperature difference are the reconstructed values ​​and / or measured values ​​obtained by reconstructing through the topology-sensing hydraulic model; Based on the chilled water flow rate and the supply and return water temperature difference, the actual cooling capacity of the distribution network is determined as follows: Where t represents the time variable, This indicates the density of water. This represents the specific heat capacity of water at constant pressure. Indicates the volumetric flow rate of chilled water. Indicates the return water temperature. This indicates the water supply temperature.

6. The digital twin modeling method for a central air conditioning chiller station as described in claim 1, characterized in that, In step S3, the formula for calculating the actual cooling capacity of the chiller unit is: Where t represents the time variable, This indicates the density of water. This represents the specific heat capacity of water at constant pressure. This indicates the chilled water volume flow rate on the evaporator side of the chiller unit. This indicates the inlet water temperature on the evaporator side. This indicates the outlet water temperature on the evaporator side.

7. The digital twin modeling method for a central air conditioning chiller station as described in claim 6, characterized in that, The actual load rate and coefficient of performance of the chiller unit are determined based on the actual cooling capacity of the chiller unit. in, Indicates the actual load rate. Indicates the coefficient of performance for cooling. This indicates the actual cooling capacity of the chiller unit. This indicates the rated cooling capacity of the chiller unit. This indicates the power of the chiller unit.

8. The digital twin modeling method for a central air conditioning chiller station as described in claim 1, characterized in that, In step S4, the formulas for the difference in cooling capacity between the required cooling capacity and the actual cooling capacity distributed, and the residual difference in cooling capacity between the actual cooling capacity supplied and the actual cooling capacity distributed, are as follows: in, This indicates the actual cooling capacity distributed on the distribution side. This indicates the cooling demand on the demand side. This represents the actual cooling capacity supplied by the equipment, where t represents the time variable. The cooling capacity status of the central air conditioning chiller is determined based on the cooling capacity difference between the demand cooling capacity and the actual cooling capacity delivered, and the cooling capacity residual difference between the actual cooling capacity supplied and the actual cooling capacity delivered.

9. The digital twin modeling method for a central air conditioning chiller station as described in claim 1, characterized in that, Dynamically correcting the digital twin model of the chiller plant includes incrementally updating the performance model of the chiller unit equipment based on an elastic weight consolidation mechanism. The objective function of the incremental update is: in, This represents the total loss from incremental updates. This represents the fitting loss for the newly added field data. Indicates constraint strength. Indicates the importance of the parameter. Indicates the current model parameters. This represents the basic model parameters.

10. A digital twin modeling system for a central air conditioning chiller station, executing the modeling method as described in any one of claims 1-9, characterized in that, The system includes: The demand-side modeling module is used to construct a demand-side cooling load model based on the operational data of the building's terminals, and to obtain the required cooling capacity of the building's terminals. The transmission and distribution side modeling module is used to construct a topology-aware hydraulic model of the transmission and distribution side to obtain the actual transmission and distribution cooling capacity of the transmission and distribution network; The supply-side modeling module is used to construct a performance model of the chiller equipment on the supply side based on the parameters of the chiller equipment, and to obtain the actual cooling capacity of the chiller equipment. The three-sided coupling module is used to align the sampling period and time label of the required cooling capacity, the actual cooling capacity distributed, and the actual cooling capacity supplied, and to perform correlation calculations on the three based on the cooling capacity balance relationship to construct a digital twin model of the cooling station and output twin prediction values. The calibration module is used to compare the twin prediction value with the corresponding field measured value, obtain the data residual between the twin prediction value and the field measured value, and dynamically correct the digital twin model of the cooling station based on the data residual.

Images