Central air-conditioning system load prediction method based on digital twinning
Through digital twin technology, the indoor temperature-cooling relationship model is established and dynamically updated, which solves the problem of low load prediction accuracy in the existing technology, and achieves more accurate load prediction and energy efficiency improvement.
Patent Information
- Application Number
- CN202411239622.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-05
- Publication Date
- 2024-11-15
AI Technical Summary
The existing central air-conditioning system load prediction model fails to accurately predict under interior design temperature conditions, and it is difficult to adapt to changes in different building characteristics, resulting in a decrease in prediction accuracy.
Using a digital twin method, a relationship model of indoor temperature-system cooling capacity is established by obtaining historical data, and converted into a cooling capacity prediction model, and continuously updated through actual operation data to avoid direct use of cooling capacity or heating capacity as Load indicators, achieve dynamic adjustment.
It improves the accuracy and reliability of load prediction of central air conditioning systems, adapts to various built environments, reduces energy consumption and improves energy utilization efficiency.
Smart Images

Figure CN118965809A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of building energy management and is applied to central air-conditioning systems, and specifically to a central air-conditioning system load prediction method based on digital twins. Background Art
[0002] In order to maintain the indoor air heat and humidity parameters within the design requirements, the heat that needs to be removed from the room per unit time is called the building cooling load. In the cooling season, the heat that the central air conditioning system needs to remove from the room mainly includes all the heat transferred from the outside to the inner surface of the enclosure structure and the heat dissipated by the indoor heat source under the action of external meteorological parameters, that is, the heat generated by external disturbances and internal disturbances respectively. The external disturbances mainly include external meteorological parameters such as outdoor air temperature and humidity, solar radiation, wind speed and direction, etc.; the internal disturbances mainly include indoor heat and humidity sources such as personnel, equipment and lighting.
[0003] The heat and moisture transfer through the envelope structure and the heat and moisture generation in the room include three forms: convective heat transfer (convective mass exchange), heat conduction (water vapor penetration) and radiation. The heat that enters the room under the action of internal and external disturbances at a certain moment is called the heat gain at that moment. The total heat generated by the indoor heat source basically depends on the calorific value of the heat source and is relatively easy to obtain. However, the total heat passing through the envelope structure is not only related to the external meteorological parameters and indoor air parameters, but also affected by other surface conditions in the room. Due to the thermal inertia of the envelope structure, the change law of its heat and moisture process is very complicated, and there is also an attenuation and delay relationship between the heat gain through the envelope structure and the external disturbance. Therefore, in theory, the calculation of heat gain has the characteristics of nonlinearity, dynamics and time lag. In the actual calculation of heat gain, certain simplified assumptions are often required. Since the building cooling load is basically equal to the heat gain in the absence of significant radiation heat gain, it is difficult to accurately calculate and predict the cooling load through physical modeling.
[0004] Compared with physical models, digital twin models do not require the establishment of building heat balance equations, so little or no building physical information is required when applied to central air-conditioning system load forecasting. However, the current data-driven algorithms used for load forecasting are all based on the time-invariant characteristics of the system. For buildings with heavy enclosure structures and thermal inertia, the impact of indoor and outdoor environmental parameters on the load of the central air-conditioning system has long-term time correlation and memory, and the existence and changes of system dynamic characteristics will greatly reduce the accuracy of load forecasting. At the same time, the current extraction of central air-conditioning system load data generally directly uses the system cooling capacity without considering the combination of other factors; however, only when the indoor temperature is at the design temperature condition, the central air-conditioning system load is equal to the system cooling capacity. Therefore, directly using the cooling capacity data to extract the cooling load data without processing or prediction will lead to a decrease in the accuracy of the relevant model and fail to achieve the expected effect of central air-conditioning system load forecasting. Summary of the invention
[0005] Based on the current status of the background technology, the purpose of the present invention is to solve the problem that the existing system load prediction model is not based on the indoor design temperature conditions and is difficult to meet the changes in different building characteristics. Therefore, a central air-conditioning system load prediction method based on digital twins is proposed. The method of the present invention can follow the changes in building characteristics, avoiding the method of directly replacing the central air-conditioning system load with cooling or heating in the modeling process, thereby reflecting the actual load situation and improving the prediction accuracy.
[0006] The present invention adopts the following technical solutions to achieve the purpose:
[0007] A method for predicting the load of a central air-conditioning system based on digital twins comprises the following steps:
[0008] S1. Obtain historical data related to the operation of the central air-conditioning system;
[0009] S2. Based on historical data, a first relationship model representing indoor temperature and system cooling capacity is established, and the first relationship model is made to output the temperature at the next moment under the input relative to the temperature at the current moment;
[0010] S3, converting the first relationship model into a second relationship model, and making the second relationship model output the system cooling capacity at the current moment relative to the temperature change at different moments;
[0011] S4, using the indoor design temperature to replace the next moment temperature relative to the current moment temperature in the second relationship model, to obtain a system cooling capacity prediction model, and using the output of the system cooling capacity prediction model as a predicted value of the central air conditioning system load;
[0012] S5. Based on the system cooling capacity prediction model, actual operation data is obtained, a data evolution model is established, and data update of the system cooling capacity prediction model is realized;
[0013] S6. The central air-conditioning system continues to operate and acquire data. The system cooling capacity prediction model is continuously updated based on the actual operating data to obtain the predicted value of the central air-conditioning system load at the corresponding time, thus completing the central air-conditioning system load prediction based on digital twins.
[0014] In summary, due to the adoption of this technical solution, the beneficial effects of the present invention are as follows:
[0015] The method of the present invention can be dynamically adjusted according to the different characteristics of the building, but the required parameters do not need physical modeling and consideration of complex physical changes, so the load demand of the central air-conditioning system can be more accurately predicted in a relatively simplified manner. This flexibility enables the model to better adapt to various types of building environments and improve the reliability and practicality of the prediction results.
[0016] By modeling based on actual design temperature conditions, the present invention avoids the errors caused by using cooling or heating as substitute indicators in traditional methods. This method is closer to the actual operation, so the prediction results are more accurate, which can more effectively guide the design and operation scheduling of central air-conditioning systems, reduce energy consumption and improve energy efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] Figure 1 It is a schematic diagram of the overall process of the method of the present invention;
[0018] Figure 2 Schematic diagram of input and output of the first relational model in the present invention;
[0019] Figure 3 is an input and output schematic diagram of the second relational model in the present invention;
[0020] Figure 4 It is a schematic diagram of input and output of the system cooling capacity prediction model in the present invention;
[0021] Figure 5 A schematic diagram of the processing logic of the data evolution model in the present invention;
[0022] Figure 6 It is a schematic diagram of the prediction effect of the example scenario in the present invention. DETAILED DESCRIPTION
[0023] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the parts of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.
[0024] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0025] Example 1
[0026] A load forecasting method for central air-conditioning system based on digital twin. Figure 1 The steps of the method can be summarized as follows:
[0027] S1. Obtain historical data related to the operation of the central air-conditioning system;
[0028] S2. Based on historical data, a first relationship model representing indoor temperature and system cooling capacity is established, and the first relationship model is made to output the temperature at the next moment under the input relative to the temperature at the current moment;
[0029] S3, converting the first relationship model into a second relationship model, and making the second relationship model output the system cooling capacity at the current moment relative to the temperature change at different moments;
[0030] S4, using the indoor design temperature to replace the next moment temperature relative to the current moment temperature in the second relationship model, to obtain a system cooling capacity prediction model, and using the output of the system cooling capacity prediction model as a predicted value of the central air conditioning system load;
[0031] S5. Based on the system cooling capacity prediction model, actual operation data is obtained, a data evolution model is established, and data update of the system cooling capacity prediction model is realized;
[0032] S6. The central air-conditioning system continues to operate and acquire data. The system cooling capacity prediction model is continuously updated based on the actual operating data to obtain the predicted value of the central air-conditioning system load at the corresponding time, thus completing the central air-conditioning system load prediction based on digital twins.
[0033] In this embodiment, the first relationship model and the second relationship model can be established through statistical regression analysis, machine learning algorithms such as neural networks, or other mathematical means; this method of modeling does not require detailed consideration and analysis of the house's enclosure structure, and does not need to be implemented through complex physical model modeling, that is, there is no need to establish a building thermal balance equation, etc., and while reducing the complexity of the method operation, it ensures the relative accuracy of the final prediction results.
[0034] The key to the conversion between the first relationship model and the second relationship model is to adjust the model structure so that the second relationship model can predict the current system cooling capacity according to the temperature changes at different times, so as to be more directly used in the design and optimization of the control strategy in the subsequent business, which is also the application purpose of predicting the load of the central air-conditioning system. The conversion process can be achieved by reverse derivation of mathematical transformation or refitting the model. The specific operation is relatively conventional and will not be repeated in this embodiment.
[0035] This embodiment first establishes a first relationship model and then converts it to a second relationship model. The logic behind this is to gradually approach the core of the problem to ensure the accuracy and applicability of the model. The advantage of this method is to make better use of existing historical data to ensure the stability and robustness of the model. Predicting the temperature at the next moment from the current moment is a relatively intuitive problem, and it can be solved through a variety of modeling methods such as linear regression and time series analysis. This simplified method can help to understand the basic behavior pattern of the central air-conditioning system more quickly. At the same time, predicting future temperatures is also a relatively easy task to verify. The correctness of the first relationship model can be evaluated in advance, so that it can make more full and accurate use of relevant data.
[0036] In most cases, how to reach the indoor design temperature by adjusting the system cooling capacity is the main problem, and the central air conditioning system load prediction is also to solve this problem. Based on the first relationship model, the system cooling capacity can be inferred from the temperature change to establish the second relationship model, ensuring that the knowledge and information data in the first relationship model are fully utilized. Therefore, the process of steps S2 and S3 of this embodiment is an effective way to gradually approach the solution to the problem, which not only takes advantage of historical data, but also ensures the simplicity, accuracy and practicality of model establishment.
[0037] In this embodiment, for the data evolution model, it can use actual operation data to continuously correct the established prediction model to further ensure its accuracy; new data collected during the continuous operation of the central air-conditioning system is also used to further optimize the prediction model, so as to better adapt to environmental changes and user needs. The core of the entire method is therefore to use historical data to build prediction models, and continuously improve these models through real-time data feedback, and ultimately achieve accurate and timely prediction of the central air-conditioning system load. This method not only helps to improve the efficiency of the system, but also reduces energy waste and improves user experience.
[0038] Example 2
[0039] On the basis of Example 1, this example introduces in detail the process of establishing the system cooling capacity prediction model in the method, that is, the contents corresponding to steps S1 to S4 of the method.
[0040] In step S1 of this embodiment, when the central air-conditioning system is running indoors, the historical data collected include temperature change data at different times, environmental variable data related to internal and external disturbances, system cooling capacity data at different times, and time lag data between system load and internal and external disturbances.
[0041] In step S2 of this embodiment, a functional relationship between indoor temperature and system cooling capacity is first established, and then a first relationship model is established based on the functional relationship; the functional relationship between indoor temperature and system cooling capacity is as follows:
[0042]
[0043] In the formula, the subscript n represents indoors, and the superscript k represents the time. represents the indoor temperature at time k, represents the indoor temperature at time k+1; represents the cooling capacity of the system at time k; f T Represents a nonlinear functional relationship; represents environmental variables related to internal disturbances, represents the environmental variables related to external disturbances, where the superscripts i and j are as follows:
[0044]
[0045] Where, T deiayin Represents the time lag between system load and internal disturbance, T delayex Represents the time lag between system load and external disturbance.
[0046] Based on the functional relationship, a first relationship model is established using a data-driven approach; Figure 2 As shown, the input layer of the first relational model includes environmental variables related to internal disturbance Environmental variables related to external disturbances Indoor temperature at time k System cooling capacity at time k The output layer of the first relational model includes the indoor temperature at time k+1
[0047] In step S3 of this embodiment, the first relational model is converted into a second relational model according to the input layer and output layer data of the first relational model; Figure 3 As shown, the input layer of the second relational model includes environmental variables related to internal disturbance Environmental variables related to external disturbances Indoor temperature at time k Indoor temperature at time k+1 The output layer of the second relational model includes the system cooling capacity at time k
[0048] In step S4 of this embodiment, the indoor design temperature T design Replace the indoor temperature at time k+1 in the input layer of the second relational model The system cooling capacity prediction model is obtained. Its input and output can be found in Figure 4 's hint.
[0049] Example 3
[0050] On the basis of the above-mentioned embodiment, this embodiment provides a detailed description of the relevant parts of the actual operation data when the central air-conditioning system is continuously running, that is, the contents corresponding to steps S5 to S6 of the method.
[0051] In step S5 of this embodiment, when the central air-conditioning system is operating indoors, the actual operating data collected and acquired include the current temperature data, environmental variable data related to internal and external disturbances, the time lag data between the system load and the internal and external disturbances, and the system cooling capacity data corresponding to the current time.
[0052] According to the system cooling capacity prediction model and the actual operation data, the system cooling capacity prediction model outputs the system cooling capacity corresponding to the current moment as the predicted value; when the current temperature is the same as the indoor design temperature, the system cooling capacity corresponding to the current moment in the actual operation data is used as the measured value; the error between the predicted value and the measured value is calculated, and the range of the load prediction error threshold is preset to [-σ%,σ%], and then a data evolution model is established.
[0053] like Figure 5 As shown, in the data evolution model, the database update method is used to realize the data update of the system cooling capacity prediction model; for the error between the calculated prediction value and the measured value, when the error is within the range of the load prediction error threshold, the actual operation data is merged with the historical data used when establishing the first relationship model, and stored as a new database; when the error is not within the range of the load prediction error threshold, data evolution is performed, firstly analyzing the similarity between the actual operation data and the historical data, eliminating the determined similar data, and storing the remaining dissimilar parts together as a new database; the data evolution model uses the new database to establish the system cooling capacity prediction model again to complete its data update.
[0054] In step S6 of this embodiment, the data evolution model continuously updates the system cooling capacity prediction model, and obtains the predicted value output by the current system cooling capacity prediction model at the corresponding moment of each update, thereby realizing a continuous digital twin process and continuously predicting the load of the central air-conditioning system.
[0055] Taking the indoor scene of a high-speed railway station with an area of about 15,000 square meters as an example, an air source heat pump unit is used as the main unit of the central air-conditioning system for summer cooling. The central air-conditioning system load prediction method of the present invention is applied to this scene, and multiple data evolutions are performed to achieve a stable digital twin prediction process, and the following is obtained: Figure 6 The predicted effect evolution results are shown in Figure 2. Figure 6It can be seen that the load prediction achieved by the system cooling capacity prediction model established by the method of the present invention has a good follow-up performance to the building characteristics. The mean absolute percentage error (MAPE) of the initial established model without evolution in continuous operation is about 8%. However, if the initial established model is evolved multiple times, the stabilized system cooling capacity prediction model can achieve a model MAPE of less than 1%, which proves its good accuracy.
Claims
1. A central air-conditioning system load forecasting method based on digital twin, characterized in that: The method comprises the following steps: S1. Obtain historical data related to the operation of the central air-conditioning system; S2. Based on historical data, a first relationship model representing indoor temperature and system cooling capacity is established, and the first relationship model is made to output the temperature at the next moment under the input relative to the temperature at the current moment; S3, converting the first relationship model into a second relationship model, and making the second relationship model output the system cooling capacity at the current moment relative to the temperature change at different moments; S4, using the indoor design temperature to replace the next moment temperature relative to the current moment temperature in the second relationship model, to obtain a system cooling capacity prediction model, and using the output of the system cooling capacity prediction model as a predicted value of the central air conditioning system load; S5. Based on the system cooling capacity prediction model, actual operation data is obtained, a data evolution model is established, and data update of the system cooling capacity prediction model is realized; S6. The central air-conditioning system continues to operate and acquire data. The system cooling capacity prediction model is continuously updated based on the actual operating data to obtain the predicted value of the central air-conditioning system load at the corresponding time, thus completing the central air-conditioning system load prediction based on digital twins.
2. The method for predicting central air-conditioning system load based on digital twin according to claim 1 is characterized in that: In step S1, when the central air-conditioning system is running indoors, the historical data collected include temperature change data at different times, environmental variable data related to internal and external disturbances, system cooling capacity data at different times, and time lag data between system load and internal and external disturbances.
3. The method for predicting central air-conditioning system load based on digital twin according to claim 1 is characterized in that: In step S2, a functional relationship between indoor temperature and system cooling capacity is first established, and then a first relationship model is established based on the functional relationship; the functional relationship between indoor temperature and system cooling capacity is as follows: In the formula, the subscript n represents indoors, and the superscript k represents the time. represents the indoor temperature at time k, represents the indoor temperature at time k+1; represents the cooling capacity of the system at time k; f T Represents a nonlinear functional relationship; represents environmental variables related to internal disturbances, represents the environmental variables related to external disturbances, where the superscripts i and j are as follows: i=[k+1-T delayin :k+1] j=[k+1-T delayex :k+1] Where, T delayin Represents the time lag between system load and internal disturbance, T delayex Represents the time lag between system load and external disturbance.
4. The method for predicting central air-conditioning system load based on digital twin according to claim 3 is characterized in that: Based on the functional relationship, a first relationship model is established in a data-driven manner; the input layer of the first relationship model includes environmental variables related to internal disturbances Environmental variables related to external disturbances Indoor temperature at time k System cooling capacity at time k The output layer of the first relational model includes the indoor temperature at time k+1 5. The method for predicting central air-conditioning system load based on digital twin according to claim 4 is characterized in that: In step S3, the first relational model is converted into a second relational model based on the input layer and output layer data of the first relational model; the input layer of the second relational model includes environmental variables related to the internal disturbance Environmental variables related to external disturbances Indoor temperature at time k Indoor temperature at time k+1 The output layer of the second relational model includes the system cooling capacity at time k 6. The method for central air-conditioning system load forecasting based on digital twin according to claim 5 is characterized in that: In step S4, the indoor design temperature T design Replace the indoor temperature at time k+1 in the input layer of the second relational model That is, the system cooling capacity prediction model is obtained.
7. The method for predicting central air-conditioning system load based on digital twin according to claim 1, characterized in that: In step S5, when the central air-conditioning system is running indoors, the actual operation data collected includes the current temperature data, the environmental variable data related to the internal and external disturbances, the time lag data between the system load and the internal and external disturbances, and the system cooling capacity data corresponding to the current time.
8. The method for predicting central air-conditioning system load based on digital twin according to claim 7 is characterized in that: According to the system cooling capacity prediction model and the actual operation data, the system cooling capacity prediction model outputs the system cooling capacity corresponding to the current moment as the prediction value; when the temperature at the current moment is the same as the indoor design temperature, the system cooling capacity corresponding to the current moment in the actual operation data is used as the measured value; The error between the predicted value and the measured value is calculated, and the load forecast error threshold is preset to the range of [-σ%,σ%], and then the data evolution model is established.
9. The method for predicting central air-conditioning system load based on digital twin according to claim 8 is characterized in that: In the data evolution model, the data update of the system cooling capacity prediction model is realized by using the database update method; for the error between the calculated prediction value and the measured value, when the error is within the range of the load prediction error threshold, the actual operation data is merged with the historical data used when establishing the first relationship model and stored as a new database; When the error is not within the range of the load forecast error threshold, data evolution is performed. First, the similarity between the actual operation data and the historical data is analyzed, and the determined similar data is eliminated. The remaining dissimilar parts are stored together as a new database; the data evolution model uses the new database to re-establish the system cooling capacity prediction model to complete its data update.
10. The method for central air-conditioning system load forecasting based on digital twin according to claim 9 is characterized in that: In step S6, the data evolution model continuously updates the system cooling capacity prediction model, and obtains the predicted value output by the current system cooling capacity prediction model at the corresponding time of each update, thereby realizing a continuous digital twin process and continuously predicting the load of the central air-conditioning system.
Citation Information
Cited By
Building operation and maintenance management system based on digital twinning
CN120806918A
