An Adaptive Commissioning Method for Ground Source Heat Pump Systems Based on Modelica

Abstract

This invention provides an adaptive commissioning method for a ground source heat pump system based on Modelica, comprising: collecting operating parameter data of the ground source heat pump system and analyzing its operating strategy; constructing a physical simulation proxy model of the ground source heat pump system based on Modelica according to its structure and operating strategy; adaptively calibrating the parameters of the physical simulation proxy model using the operating parameter data to obtain a calibration model; verifying the accuracy of the calibration model using operating parameter data from another independent time period to obtain a standardization model; performing adaptive virtual commissioning on the standardization model to optimize and determine the operating strategy; periodically collecting operating parameter data, calibrating the parameters of the standardization model, and updating the standardization model. This invention provides an adaptive commissioning method for a ground source heat pump system based on Modelica, constructing a high-fidelity dynamic model and transforming the commissioning process from a one-time verification to adaptive optimization.

Description

Technical Field

[0001] This invention belongs to the field of heating, ventilation, air conditioning and building energy management technology, specifically, it relates to an adaptive commissioning method for a ground source heat pump system based on Modelica. Background Technology

[0002] Ground source heat pump systems, as a highly efficient air conditioning and heating technology utilizing renewable energy, have been widely applied in the field of building energy conservation in recent years. Compared with traditional air conditioning systems, ground source heat pumps have advantages such as stable operation, high energy efficiency, and significant energy saving and emission reduction effects. However, the overall performance of the system depends not only on the design and installation quality of the equipment but also on various factors such as operating strategies, load characteristics, climate conditions, and equipment aging. Currently, the commissioning of ground source heat pump systems is mainly concentrated in the final acceptance stage, and the commissioning process is mostly based on one-time testing and parameter setting, which is difficult to cover the diverse operating conditions under different meteorological conditions throughout the year. This static, one-time commissioning method makes it difficult to adjust the operating strategy in a timely manner when faced with load fluctuations, environmental changes, and equipment performance degradation during operation, which can easily lead to problems such as decreased energy efficiency, increased energy consumption, or reduced indoor comfort in the long-term operation of the system. In addition, some optimized control strategies face high risks and implementation costs when directly tested on-site in actual buildings, and are often difficult to fully verify and apply in real-world environments.

[0003] In recent years, with the development of building simulation technology, physical mechanism-based system modeling methods have been increasingly applied to the modeling, analysis, and commissioning of HVAC systems. Among these, Modelica, due to its openness, cross-platform compatibility, and rich physical model library, has become an important tool for constructing simulation models of complex building energy systems. Existing methods typically build system models based on design parameters or limited historical operating data for theoretical performance evaluation or strategy verification under specific conditions. However, these models still have shortcomings in terms of parameter matching with actual operating characteristics and lifecycle optimization of operating strategies, making it difficult to maintain high accuracy and reliability in dynamically changing operating environments. Furthermore, existing technologies often remain at the one-time modeling stage, failing to achieve adaptive adjustment of model parameters based on actual operating data. This makes it difficult to construct dynamic models that can reflect the characteristics of the real system over the long term, resulting in model-based commissioning results that are difficult to maintain effectiveness and adaptability in actual operation. Therefore, there is an urgent need for a technical solution that can transform the commissioning process from one-time verification to adaptive optimization to achieve efficient and safe operation of ground source heat pump systems throughout their entire lifecycle. Summary of the Invention

[0004] The technical problem to be solved by this invention is to provide an adaptive commissioning method for ground source heat pump systems based on Modelica, which constructs a high-fidelity dynamic model that can reflect the characteristics of the real system over a long period of time, transforms the commissioning process from a one-time verification to adaptive optimization, and realizes the efficient and safe operation of the ground source heat pump system throughout its entire life cycle.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: This invention provides an adaptive commissioning method for a ground source heat pump system based on Modelica, comprising the following steps: Step 10: Collect operating parameter data of the ground source heat pump system and analyze the operating strategy of the ground source heat pump system; Step 20: Based on the structure and operation strategy of the ground source heat pump system, construct a physical simulation proxy model of the ground source heat pump system using Modelica; use the operating parameter data to adaptively calibrate the parameters of the physical simulation proxy model to obtain a calibrated model; Step 30: Verify the accuracy of the calibration model using operating parameter data from another independent time period to obtain the calibration model; Step 40: Perform adaptive virtual debugging on the calibration model to optimize and determine the operating strategy; Step 50: Periodically collect the operating parameter data of the ground source heat pump system, calibrate the parameters of the calibration model, and update the calibration model.

[0006] As a further improvement of the present invention, the operating parameters include the main pipe temperature at the inlet of the heat pump evaporator, the main pipe temperature at the outlet of the heat pump evaporator, the main pipe temperature at the inlet of the condenser, the main pipe temperature at the outlet of the condenser, the inlet temperature of the buried pipe loop, the outlet temperature of the buried pipe loop, the supply water temperature of the heating circuit between the building and the buffer water tank, the return water temperature between the building and the buffer water tank, the operating power of each heat pump, the water flow rate of the heating circuit between the heat pump and the buffer water tank, the water flow rate of the cooling circuit between the heat pump and the buffer water tank, the water flow rate of the heating circuit between the building and the buffer water tank, the water flow rate of the cooling circuit between the building and the buffer water tank, and the fluid flow rate of the buried pipe loop.

[0007] As a further improvement of the present invention, the operating strategy includes a heat pump start-up and shutdown strategy, a heating and hot water pump control strategy, a cooling water pump control strategy, a buried pipe loop fluid pump control strategy, and a buffer tank temperature regulation strategy.

[0008] As a further improvement of the present invention, step 20, based on the structure and operation strategy of the ground source heat pump system, constructs a physical simulation proxy model of the ground source heat pump system using Modelica, specifically including: Step 201: Based on the Modelica model library, call the relevant components of the ground source heat pump system and connect them according to the structural topology of the ground source heat pump system. Set the control components according to the operation strategy to obtain the preliminary structural framework of the physical simulation proxy model of the ground source heat pump system. Step 202: Based on the rated parameters of the equipment in the ground source heat pump system and the system design information, set the parameters of the model components in the preliminary structural framework of the physical simulation proxy model to obtain the physical simulation proxy model of the ground source heat pump system. The model components include heat pump components, buffer tank components, underground pipe heat exchanger components, and room components; The parameters of a heat pump component include heating capacity, cooling capacity, condenser design water flow rate, evaporator design water flow rate, evaporator inlet and outlet design temperature difference, condenser inlet and outlet design temperature difference, coefficient of performance (COP) under heating conditions, and coefficient of performance (COP) under cooling conditions. The parameters of the buffer tank assembly include the tank's volume, height, thickness of the external insulation layer, thermal conductivity of the insulation material, and mixing time constant inside the tank. The parameters of the buried pipe heat exchanger assembly include fluid flow rate under design conditions, borehole radius, soil type, packing type, total borehole depth, pipe pair spacing, pipe radius, pipe wall thickness, and local average annual soil temperature. The parameters of the room components include the number of air changes per minute.

[0009] As a further improvement of the present invention, in step 20, the parameters of the physical simulation proxy model are adaptively calibrated using the running parameter data to obtain a calibrated model, specifically including: Step 203: Based on the relevant operating parameter data of the heat pump, the parameters of the heat pump components are adaptively calibrated by using data fitting. Step 204: Based on simulated building load data, sensitivity analysis is used to screen parameters, and Bayesian inference methods are used to adaptively calibrate the parameter settings of room components.

[0010] As a further improvement of the present invention, step 203 specifically includes: Based on the collected operating parameter data, the heat pump heating capacity, heat pump cooling capacity, heat pump operating load rate under heating condition, and heat pump operating load rate under cooling condition are calculated. Then, the coefficient of performance (COP) and its decay coefficient under heating condition and the COP and its decay coefficient under cooling condition are calculated. The relationship between the decay coefficient under heating condition and the heat pump operating load rate under heating condition, and the relationship between the decay coefficient under cooling condition and the heat pump operating load rate under cooling condition are fitted. The calibration parameters obtained from the fitted parameters are set in the heat pump component of the physical simulation proxy model.

[0011] As a further improvement of the present invention, step 204 specifically includes: In the physical simulation proxy model, different combinations of building envelope and building thermal parameters are set to generate simulated building load data, and a substitute model is constructed using the random forest method. Based on the substitute model, the first-order sensitivity index and total effect sensitivity index of the building envelope heat transfer coefficient, heat capacity, thickness, area, density, and room volume are calculated using the improved Sobol sampling method, and a set of key parameters higher than a preset threshold are selected. For the key parameters, a uniform prior distribution is set and a likelihood function is constructed using the average absolute error between the predicted value and the actual measured value of the substitute model. The posterior distribution is solved using Bayes' theorem, and the posterior probability distribution of the key parameters is obtained using Markov chain Monte Carlo sampling. The parameter value at the maximum posterior probability density is selected as the final calibration result. The calibration parameters are set in the room component of the physical simulation proxy model.

[0012] As a further improvement of the present invention, step 30 specifically includes: Using the operating parameter data from another independent time period as input into the calibration model, the normalized average deviation error values ​​of all operating parameters are calculated using equation (16): Equation (16) In the formula, Indicates the first i One measurement value, Indicates the first i One simulated value, This represents the average value of the simulated values. n Indicates the number of data points; If the normalized average deviation error of all operating parameters is within ±15%, then the calibration model is used as the standardization model; otherwise, return to step 20.

[0013] Compared with the prior art, the technical solution of the present invention has the following beneficial effects: (1) The present invention provides an adaptive commissioning method for ground source heat pump system based on Modelica. By introducing a high-fidelity physical simulation proxy model and a periodic parameter calibration mechanism, it breaks through the limitations of the traditional ground source heat pump system commissioning method of one-time static verification. This enables the model to maintain consistency with the actual system characteristics under the conditions of fluctuating operating conditions, changes in meteorological conditions and equipment performance degradation, thus ensuring the long-term adaptability and efficiency of the operation strategy.

[0014] (2) In the modeling process, the heat pump operating performance is calibrated based on actual operating data. In addition, by combining sensitivity analysis and Bayesian inference, high-precision calibration is performed only on key parameters that have a significant impact on the system output. This reduces the parameter dimension, the computation and modeling complexity, and effectively reduces the risk of model overfitting. This significantly improves the model's ability to reflect the actual operating characteristics of the system.

[0015] In the adaptive calibration method for heat pump component operating parameters, by establishing a quantitative relationship between the heat pump operating load rate and the COP decay coefficient, the energy efficiency characteristics of the heat pump under different partial load conditions can be accurately characterized, overcoming the bias problem caused by traditional models relying solely on nominal operating parameters. This method improves the prediction accuracy of the model under different seasons and operating conditions, providing a reliable basis for optimizing operating strategies (such as heat pump number control and preheating control), thereby improving the overall system operating energy efficiency and reducing energy consumption costs. For the adaptive calibration of room component parameters, key building thermal parameters are screened through sensitivity analysis, and then high-precision calibration is performed using Bayesian inference methods. This not only avoids redundant calibration of non-sensitive parameters and reduces computational complexity, but also quantifies uncertainty while obtaining optimal parameter values. This method effectively improves the fitting accuracy of the room model to the building's thermal dynamic characteristics, enhancing the credibility and practicality of the virtual commissioning phase in building load prediction and indoor comfort assessment.

[0016] The method of this invention not only ensures the long-term accuracy and stability of the model, but also enables quantitative evaluation and optimization of different operating strategies in a risk-free virtual environment, achieving multiple effects such as improved system energy efficiency, reduced operating costs, and guaranteed indoor comfort, demonstrating significant advancement and practical value.

[0017] (3) The method of the present invention significantly improves the generalization ability and robustness of the model under different working conditions through the verification mechanism of independent time period datasets.

[0018] (4) In the operation strategy optimization stage, the method of the present invention relies on the virtual debugging function of the Modelica simulation platform to conduct multi-scheme comparative analysis of operating parameters such as the number of heat pumps turned on, water tank temperature setpoint, and heat pump turn-on time in a risk-free simulation environment, quantify the energy-saving potential of each scheme, and verify its comprehensive guarantee capability for building load, indoor thermal comfort and equipment safe operation, thereby avoiding the high risk and high cost caused by direct trial and error in the real system.

[0019] (5) The method of the present invention has good scalability and versatility. It is not only applicable to ground source heat pump systems, but can also be extended to other types of HVAC and renewable energy utilization systems, providing reliable technical support for the efficient operation of building energy systems throughout their entire life cycle. Attached Figure Description

[0020] Figure 1 A flowchart of an adaptive commissioning method for a ground source heat pump system based on Modelica, provided for this invention; Figure 2 This is a schematic diagram of a ground source heat pump system in a specific embodiment of the present invention; Figure 3 This is a data visualization analysis diagram of a specific example of the present invention; wherein, Figure 3 (a) is a visualization of the heat pump hot water return temperature and chilled water return temperature data. Figure 3 (b) A visualization of heat pump operating power data; Figure 4 This is a logical diagram of the actual system operation strategy based on actual data analysis in a specific embodiment of the present invention; Figure 5 This is the preliminary structural framework of the physical simulation proxy model of the ground source heat pump system built based on Modelica in a specific example of the present invention; Figure 6 This is a parameter calibration result diagram of the heat pump component in a specific embodiment of the present invention; wherein, Figure 6 (a) is a graph showing the parameter calibration results of the heat pump components in heating mode. Figure 6 (b) is a graph showing the parameter calibration results of the heat pump component in cooling mode; Figure 7 This is the fitting result of using the random forest method as an alternative model in a specific example of the present invention; Figure 8 The results of sensitivity analysis of the calibration parameters of the room components in a specific embodiment of the present invention; Figure 9 This is a diagram showing the calibration results of room component parameters in a specific embodiment of the present invention; wherein, Figure 9 (a) shows the calibration results of the area. Figure 9 (b) shows the calibration results for the thickness. Figure 9 (c) shows the calibration results of the heat transfer coefficient. Figure 9 (d) shows the calibration results of the hot melt; Figure 10 This is a comparison chart of the model simulation calculation results and the actual system measurement values ​​in a specific example of the present invention; wherein, Figure 10 (a) is a comparison chart of the simulation calculation results and actual measured values ​​of the heat pump operating power. Figure 10 (b) is a comparison chart of the simulation calculation results and actual measured values ​​of the heat pump hot water supply temperature. Figure 10 (c) Comparison of simulation results and actual measured values ​​of heat pump hot water return temperature; Figure 11This is a comparison chart of virtual commissioning results for different heat pump start-up strategies in a specific embodiment of the present invention; wherein, Figure 11 (a) Comparison of operating power results under different heat pump start-up strategies during virtual commissioning. Figure 11 (b) is a comparison chart of room temperatures under different heat pump start-up strategies during virtual commissioning on the coldest day. Detailed Implementation

[0021] The technical solution of the present invention will be described in detail below.

[0022] This invention provides an adaptive commissioning method for a ground source heat pump system based on Modelica, such as... Figure 1 As shown, it includes the following steps: Step 10: Collect operating parameter data of the ground source heat pump system and analyze the operating strategy of the ground source heat pump system.

[0023] Step 20: Based on the structure and operation strategy of the ground source heat pump system, construct a physical simulation proxy model of the ground source heat pump system using Modelica. Adaptively calibrate the parameters of the physical simulation proxy model using operating parameter data to obtain a calibrated model.

[0024] Step 30: Verify the accuracy of the calibration model using operating parameter data from another independent time period, different from that in Step 20, to obtain the calibration model.

[0025] Step 40: Perform adaptive virtual debugging on the calibration model to optimize and determine the operating strategy.

[0026] Step 50: Periodically collect the operating parameter data of the ground source heat pump system, calibrate the parameters of the calibration model, and update the calibration model.

[0027] The operating parameters include the main pipe temperature at the inlet of the heat pump evaporator, the main pipe temperature at the outlet of the heat pump evaporator, the main pipe temperature at the inlet of the condenser, the main pipe temperature at the outlet of the condenser, the inlet temperature of the buried pipe loop, the outlet temperature of the buried pipe loop, the supply water temperature of the heating circuit between the building and the buffer tank, the return water temperature between the building and the buffer tank, the operating power of each heat pump, the water flow rate of the heating circuit between the heat pump and the buffer tank, the water flow rate of the cooling circuit between the heat pump and the buffer tank, the water flow rate of the heating circuit between the building and the buffer tank, the water flow rate of the cooling circuit between the building and the buffer tank, and the fluid flow rate of the buried pipe loop.

[0028] The operating strategies include the start-up and shutdown strategies of heat pumps, the control strategies of heating and hot water pumps, the control strategies of chilled water pumps, the control strategies of buried pipe loop fluid pumps, and the temperature regulation strategies of buffer tanks.

[0029] Preferably, in step 20, based on the structure and operation strategy of the ground source heat pump system, a physical simulation proxy model of the ground source heat pump system is constructed using Modelica, specifically including: Step 201: Based on the Modelica model library, call the relevant components of the ground source heat pump system and connect them according to the structural topology of the ground source heat pump system. Set the control components according to the operation strategy to obtain the preliminary structural framework of the physical simulation proxy model of the ground source heat pump system.

[0030] Specifically, the "Carnot_y", "Stratified", "FlowControlled_m_flow", and "UTube" components developed in the open-source "Buidings library" are used to build the ground source heat pump, buffer tank, water pump or fluid pump, and buried pipe heat exchanger components in the model, respectively. The "RadiatorEN442_2", "MixingVolume", "SingleLayer", and "HeatCapacitor" components are used to construct the room components in the model. Finally, based on the actual system's sensor layout, equipment configuration, connection methods, and control strategies, the corresponding connection and control components from the "Buidings library" are connected according to the actual system topology to construct the preliminary structural framework of the system's physical model.

[0031] Step 202: Based on the rated parameters of the equipment in the ground source heat pump system and the system design information, set the parameters of the model components in the preliminary structural framework of the physical simulation proxy model to obtain the physical simulation proxy model of the ground source heat pump system.

[0032] The model components include a heat pump assembly, a buffer water tank assembly, a buried pipe heat exchanger assembly, and a room assembly.

[0033] The parameters of the heat pump component include heating capacity, cooling capacity, condenser design water flow rate, evaporator design water flow rate, evaporator inlet and outlet design temperature difference, condenser inlet and outlet design temperature difference, coefficient of performance (COP) under heating conditions, and coefficient of performance (COP) under cooling conditions. Preferably, the COP is used as the COP indicator.

[0034] The parameters of the buffer tank assembly include the tank's volume, height, thickness of the external insulation layer, thermal conductivity of the insulation material, and the mixing time constant inside the tank. Preferably, the mixing time constant inside the tank is 60 seconds.

[0035] The parameters of the buried pipe heat exchanger assembly include fluid flow rate under design conditions, borehole radius, soil type, packing type, total borehole depth, pipe pair spacing, pipe radius, pipe wall thickness, and local average annual soil temperature.

[0036] The parameters of the room components include the number of air changes per minute.

[0037] These parameters can be obtained from the design information of the ground source heat pump system.

[0038] Preferably, in step 20, the parameters of the physical simulation proxy model are adaptively calibrated using the running parameter data to obtain a calibrated model, specifically including: Step 203: Based on the relevant operating parameter data of the heat pump, adaptive calibration is performed on the relationship between the attenuation coefficient of the heating capacity, the attenuation coefficient of the cooling capacity, the coefficient of performance under heating conditions, the coefficient of performance under cooling conditions, the operating power of the heat pump and the actual load rate by means of data fitting.

[0039] Specifically, it includes: Based on the collected operating parameter data, the heat pump heating capacity is calculated using equation (1), the heat pump cooling capacity is calculated using equation (2), the heat pump operating load rate under heating conditions is calculated using equation (3), and the heat pump operating load rate under cooling conditions is calculated using equation (4).

[0040] Equation (1) Equation (2) Equation (3) Equation (4) In the formula, Indicates the heating capacity of the heat pump. Indicates the cooling capacity of the heat pump. Indicates the mass flow rate of hot water under heating conditions. This indicates the chilled water mass flow rate under refrigeration conditions. This indicates the hot water supply temperature under heating conditions. This indicates the hot water return temperature under heating conditions. This indicates the chilled water supply temperature under refrigeration conditions. This indicates the chilled water return temperature under refrigeration conditions. This indicates the nominal heating capacity of the heat pump. This indicates the nominal cooling capacity of the heat pump. This indicates the specific heat capacity of water. This indicates the heat pump operating load rate under heating conditions. This indicates the heat pump operating load rate under cooling conditions.

[0041] Then, the performance coefficient under heating conditions is calculated using equation (5), the attenuation coefficient is calculated using equation (7), the performance coefficient under cooling conditions is calculated using equation (6), and the attenuation coefficient is calculated using equation (8).

[0042] Equation (5) Equation (6) Equation (7) Equation (8) In the formula, Indicates the coefficient of performance under heating conditions. The coefficient of performance (COP) indicates the performance under refrigeration conditions. This represents the attenuation coefficient of COP under heating conditions. This represents the decay coefficient of COP under refrigeration conditions. This indicates the heating capacity of the heat pump. This indicates the cooling capacity of the heat pump. This indicates the nominal power of the heat pump under heating conditions. This indicates the nominal power of the heat pump under cooling conditions.

[0043] The relationship between the attenuation coefficient and the heat pump operating load rate under heating conditions, as shown in equation (9), and the relationship between the attenuation coefficient and the heat pump operating load rate under cooling conditions are fitted together.

[0044] Equation (9) In the formula, , , It is a constant obtained through identification.

[0045] Actual operating data was used for fitting, and the least squares method was selected. Finally, the relationship between the decay coefficient of the heat pump's operating COP and the heat pump's operating load rate was established, and the calibration parameters obtained from the fitting were set in the heat pump components of the physical simulation proxy model.

[0046] Step 204: Based on simulated building load data, sensitivity analysis is used to screen parameters, and Bayesian inference methods are used to adaptively calibrate the parameter settings of room components.

[0047] Specifically, it includes: Considering that information on parameters affecting building load, such as building volume, envelope area, heat transfer coefficient, heat capacity, and density, is usually difficult to obtain, this invention sets different combinations of envelope and building thermal parameters in the physical simulation proxy model to generate simulated building load data, and uses the random forest method to construct an alternative model, which is used for subsequent rapid analysis.

[0048] Based on the pre-trained alternative model, a parameter sensitivity analysis problem is constructed, including the set of parameters to be calibrated { x1 , x 2 , ..., x k} and its corresponding range of values ​​[ L i , U i ],in L i and U i The first i The lower and upper bounds of each parameter are determined. An improved Sobol sequence sampling method is used to generate a uniformly distributed sampling matrix in the parameter space. X Each set of sampled parameters is input into the alternative model to calculate the output. Y The Sobol method was used to calculate the heat transfer coefficient, heat capacity, thickness, area, density, and first-order sensitivity index of the room volume of the building envelope. S i Sensitivity index of total effect As shown in equations (10) and (11).

[0049] Equation (10) Equation (11) In the formula, Indicates except All parameter sets outside of this set. Represents variance. It represents the mathematical expectation.

[0050] according to The parameters are sorted by their magnitudes, and key parameters with a total effect sensitivity index higher than a preset threshold are selected as the parameter set for subsequent calibration. Preferably, the preset threshold is 0.005.

[0051] The set of key parameters obtained from the sensitivity analysis θ ={ θ 1 , θ 2 , ..., θ m}, set a uniform prior distribution, as shown in equation (12). Use this to replace the model predictions. y pred,j ( θ ) and actual measured value y true,j The average absolute error between the two is used as the objective function, as shown in Equation (13), and the likelihood function is constructed as shown in Equation (14). Combining the prior distribution and the likelihood function, the posterior distribution is obtained according to Bayes' theorem, as shown in Equation (15).

[0052] Equation (12) Equation (13) Equation (14) Equation (15) In the formula, This represents the load forecast from the alternative model. This represents the actual load measurement value, and N represents the total number of observation data samples. Describes the likelihood function, in the parameters The probability of the observed data D occurring; This represents the proportionality coefficient, used to adjust the degree of influence of MAE on the likelihood value; a value of 1 is preferred. Represents the prior distribution, which describes the distribution of parameters without the use of observational data. The probability assumption; Denotes the posterior distribution, with parameters given the observed data D. The probability distribution; The symbol indicates "proportional to", meaning that both sides of the equation are proportional.

[0053] The Markov chain Monte Carlo sampling method was used to numerically solve the posterior distribution, obtaining the posterior probability distribution curves of the building envelope's heat transfer coefficient, heat capacity, thickness, and area. The parameter value at the point of maximum posterior probability density was selected as the final calibration result. The calibration parameters were then set in the room components of the physical simulation proxy model.

[0054] Preferably, step 30 specifically includes: Input the operating parameter data from a different time period than that used in step 20 into the calibration model, and calculate the normalized average deviation error value of all operating parameters using equation (16): Equation (16) In the formula, Indicates the first i One measurement value, Indicates the first i One simulated value, This represents the average value of the simulated values. n Indicates the number of data points; If the normalized average deviation error of all operating parameters is within ±15%, it indicates that the model has high accuracy and can reflect the actual system. In this case, the calibration model will be used as the standardization model for subsequent debugging. Otherwise, return to step 20 and recalibrate the model parameters using more data over a longer period until the model validation error deviation for each individual data period is within ±15%.

[0055] In step 40, key operating parameters such as the number of heat pumps in operation, the water tank temperature setpoint, and the heat pump start-up time can be adjusted based on the model. By comparing and analyzing different parameter settings in the simulation platform, the energy-saving effect of each scheme is evaluated. Simulation data is then used to verify whether the scheme meets building load requirements, indoor thermal comfort requirements, and the reliability requirements for safe system operation, thus achieving a balance between efficiency and safety in the operating strategy. Finally, based on the results of this virtual debugging, the optimized operating strategy is applied to the actual control system to achieve efficient, safe, and stable operation of the actual system.

[0056] In step 50, key parameters of the physical simulation proxy model are periodically recalibrated based on actual system operation data. The updated model is then used to evaluate and optimize the operating strategy in a virtual environment. The optimized strategy is applied to the actual control system to ensure long-term protection of system energy efficiency, indoor thermal comfort, and operational safety. Preferably, the adaptive update cycle can be set to once a month.

[0057] The following is a specific example.

[0058] A ground-source heat pump system in an office building in London, England, has the following structure: Figure 2 As shown, the system comprises two ground source heat pumps configured in parallel, each selected for full-load design capacity and connected to a vertical perforated heat exchanger for heat exchange with the ground. Each heat pump has a cooling capacity of 148 kW and a heating capacity of 160 kW. While this configuration complies with standard redundancy design specifications, it presents inherent challenges in actual operation. Specifically, system efficiency drops significantly when the load is far below the design value. The system also includes a hot and cold water buffer tank. The system includes sensors for the main pipe temperature at the evaporator inlet and outlet (T5, T6); sensors for the main pipe temperature at the condenser inlet and outlet (T7, T8); sensors for the inlet and outlet temperature of the buried pipe loop (T3, T4); and sensors for the supply and return water temperatures of the heating and cooling loops between the building and the buffer tank (T1, T2, T9, T1). 10 The sensors include: a power meter (P) for each heat pump; water flow meters (F4, F5) for the heating and cooling circuits between the heat pump and the buffer tank; water flow meters (F1, F2) for the heating and cooling circuits between the building and the buffer tank; and a fluid flow meter (F3) for the underground pipe loop. Data monitored by these sensors is available from the building system, and the monitoring and recording time is 15 minutes.

[0059] In step 10, operating parameter data are collected through sensors and the building control system, including the main pipe temperature at the inlet and outlet of the heat pump evaporator, the main pipe temperature at the inlet and outlet of the condenser, the inlet and outlet temperatures of the buried pipe loop, the supply water temperature and return water temperature of the heating and cooling circuit between the building and the buffer water tank, the operating power of each heat pump, the water flow rate of the heating and cooling circuit between the heat pump and the buffer water tank, the water flow rate of the heating and cooling circuit between the building and the buffer water tank, and the fluid flow rate of the buried pipe loop.

[0060] The analysis focuses on the initial operation strategies of the actual system, including the start-up and shutdown strategies of the heat pump, the control strategies of the heating and hot water pumps, the chilled water pumps, the underground pipe loop fluid pumps, and the temperature regulation strategies of the buffer tank.

[0061] Specifically, the visualization analysis chart based on actual data from November 27th to 29th, 2022 is shown below. Figure 3 As shown. From Figure 3 As can be seen, when the heat pump is turned on, it adopts an equal load distribution operation strategy with simultaneous operation. Furthermore, the return water temperature of the chilled water in the heat pump is generally controlled between 9 and 12℃, and the return water temperature of the hot water in the heat pump is controlled between 44 and 47℃. Therefore, it can be analyzed that when the hot water return water temperature in the tank is below 44℃, the heat pump is turned on for heating; when it is above 47℃, the heat pump is turned off for heating, and the heating capacity of the heat pump is adjusted by PI to keep the hot water return water temperature above 45℃. When the cold water return water temperature in the tank is above 12℃, the heat pump is turned on for cooling; when it is below 9℃, the heat pump is turned off for cooling, and the heating capacity of the heat pump is adjusted by PI to keep the cold water return water temperature above 10℃. The heat pump in this example can provide both heating and cooling simultaneously. The chilled water pump, hot water pump, and underground pipe loop side fluid pump are fixed-frequency pumps, adopting a fixed-flow operation strategy. Therefore, the actual system operation strategy logic based on actual data analysis can be summarized as follows: Figure 4 As shown.

[0062] Step 201: Using the open-source "Buidings library," the "Carnot_y," "Stratified," "FlowControlled_m_flow," and "UTube" components are used to build models of the ground source heat pump, buffer tank, water pump or fluid pump, and buried pipe heat exchanger, respectively. Additionally, the "RadiatorEN442_2," "MixingVolume," "SingleLayer," and "HeatCapacitor" components are used to construct the room model. Finally, based on the actual sensor layout, actual equipment configuration, and connection methods of the system, and the analysis obtained in Step 10... Figure 4The control logic utilizes the corresponding connection components and control module components in the "Buildings library" to connect according to the actual system topology, ultimately constructing the preliminary structural framework of the physical simulation proxy model of the ground source heat pump system, as shown below. Figure 5 As shown in the figure. In this model, the simulation calculation can be achieved by using external meteorological conditions and the flow rate of the building-side water pump. That is, the inputs are the outdoor temperature at different times and the operating flow rate of the building-side water pump, and the outputs include, but are not limited to, the operating parameter data of the acquisition system described in step 10.

[0063] Step 202: A summary of these parameters set based on the actual system information is shown in Table 1.

[0064] Table 1 Summary of actual system information settings parameters

[0065] Step 203: Heating operation data from December 1st to 18th, 2022, and cooling operation data from August 12th to 25th, 2022, were selected. The heat pump heating capacity or cooling capacity at different times was calculated using equations (1) and (2). Then, the heat pump operating load rate under heating and cooling conditions was calculated using equations (3) and (4). Next, the heat pump operating COP under heating and cooling conditions was calculated using equations (5) and (6). Then, the attenuation coefficient of the heat pump operating COP under heating and cooling conditions was calculated using equations (7) and (8). Finally, based on the relationship in equation (9), the actual operating data was used for fitting. The parameter calibration results of the ground source heat pump unit model under heating and cooling conditions were obtained through fitting, as shown below. Figure 6 As shown. The heating condition coefficient was obtained through fitting. a 1, a 2, a The coefficients of performance (COPs) are 0.4101, 1.1365, and -0.5484, respectively. a 1, a 2, a The values ​​for 3 are 0.01, 0.5941, and -0.0455, respectively.

[0066] Step 204: Select 14 days of operational data from the actual system from August 30th to September 12th, 2022. First, set different combinations of building heat transfer coefficient, building envelope heat capacity, building envelope thickness, building envelope area, wall density, and room volume in the Modelica model, thereby generating partially simulated building load data. Based on this data, an alternative model is constructed using the random forest method. The fitting results are as follows: Figure 7As shown, the alternative model fits well and accurately reflects the simulation results of the actual Modelica model in this data. Subsequently, based on the trained random forest model, a parameter sensitivity analysis problem is constructed. The parameters to be calibrated include six parameters: building heat transfer coefficient, building envelope heat capacity, building envelope thickness, building envelope area, wall density, and room volume. Their corresponding value ranges are summarized in Table 2. A uniformly distributed sampling matrix is ​​generated in the parameter space using the Saltelli modified Sobol sequence sampling method. X Each set of sampled parameters is input into the alternative model to calculate the output. Y The first-order sensitivity exponent for each parameter is calculated using the Sobol method. S i Total effect sensitivity index As shown in equations (10) and (11) respectively. Sort the parameters by their size to obtain, as follows Figure 8 The results show that the building heat transfer coefficient, building envelope heat capacity, building envelope thickness, and building envelope area are highly sensitive, while the other variables are less sensitive, with values ​​all below the optimal threshold of 0.005. Therefore, the key variables selected are the building heat transfer coefficient, building envelope heat capacity, building envelope thickness, and building envelope area, totaling four.

[0067] Table 2 Sampling range of prior distribution of parameters to be calibrated heat transfer coefficient of building envelope <![CDATA[0.2-0.7 W / (m 2 ·K)]]> thermal capacity of building envelope 300-1000 J / (kg·K) Envelope thickness 0.2-0.4 m Envelope area <![CDATA[1000-4000 m 2 ]]> Wall density <![CDATA[1.6-3 g / cm 3 ]]> Room volume <![CDATA[10000-30000 m 3 ]]> A uniform prior distribution is set as shown in Equation (12). Specifically, the prior distribution settings for these four variables are shown in Table 2. The load forecast values ​​from the random forest model are used... y pred,j ( θ ) and actual load measurement value y true,j The average absolute error between the two is used as the objective function, as shown in Equation (13), and the likelihood function is constructed as shown in Equation (14). Combining the prior distribution and the likelihood function, the posterior distribution is obtained according to Bayes' theorem, as shown in Equation (16).

[0068] The Markov chain Monte Carlo sampling method was used to numerically solve the posterior distribution, obtaining the posterior probability distribution curves of each parameter. The parameter value at the point of maximum posterior probability density was selected as the final calibration result. (See...) Figure 9 Therefore, in Modelica, the building heat transfer coefficient, building envelope heat capacity, building envelope thickness, and building envelope area are set to 0.5932 W / (m²). 2 ·K), 890.59J / (kg·K), 0.2982m, 3416.77m 2 .

[0069] This leads to the calibration model.

[0070] Step 30: Data from December 1st to December 18th, 2022 (18 days) was selected. The parameters to be verified include the corresponding parameters of all the system operation data collected in Step 10. The normalized mean deviation error of all parameters was calculated using Equation (16). If all normalized mean deviation errors are within ±15%, the model accuracy is high, it can reflect the actual system, and it can be used for subsequent virtual debugging. If the error deviation exceeds ±15%, the process returns to Step 20, and the model parameters are recalibrated using more data with a longer period until the model verification error deviation for each data period is within ±15%. Figure 10 A comparison chart of some simulation results and actual system measurement data is given. The normalized average deviation of all simulation results and actual system measurement data can be summarized in Table 3. It can be seen that the normalized average deviation of all detection parameters is less than ±5.34%, and this error level can accurately reflect the actual situation of the system.

[0071] Table 3 Normalized average deviation of simulation results for each system parameter Total power of heat pump 3.36 Heat pump hot water return temperature 0.71 Heat pump hot water supply temperature 2.05 Hot water supply temperature between the building and the buffer tank -0.05 Hot water return temperature between the building and the buffer tank 0.65 Heat pump chilled water return temperature 2.66 Heat pump chilled water supply temperature 5.19 Cold water supply temperature between the building and the buffer tank 1.97 Cold water return temperature between the building and the buffer tank 5.34 Inlet water temperature of buried pipe side loop 2.06 Water outlet temperature of buried pipe side loop 2.84 Building load 4.05 Step 40: Analysis revealed that the heat pump control method was unreasonable during the heating season, resulting in a long-term low system load rate and low energy efficiency. The system's energy efficiency can be optimized and improved by adjusting the number of heat pumps in operation and their operating time. Virtual commissioning was conducted based on this calibration model. Due to the low system load rate, the original control method of running two heat pumps simultaneously was changed to running only one heat pump during virtual commissioning to increase the system load rate and thus improve equipment energy efficiency. To verify the reliability of the solution during virtual commissioning, the two coldest weeks in historical data—December 5th to December 18th, 2022—were selected. The results of the virtual commissioning were used to evaluate the reliability of the solution, specifically whether changing to single-pump operation met the building load and indoor thermal comfort requirements, assessing the system's safe operation, and quantifying the benefits of the optimized operating strategy. This virtual commissioning method solves the difficulties of long commissioning cycles, high commissioning risks, and high costs associated with traditional on-site optimization commissioning. Figure 11As shown, debugging revealed that simply switching to a single heat pump operation mode, while improving load rate and system energy efficiency, could violate the room's thermal comfort requirements in colder weather conditions, particularly in the morning. However, through simulation tests under different weather conditions and initial temperatures, operating only one heat pump while preheating the room 30 to 120 minutes in advance, analysis showed that preheating the room more than one hour in advance, even with only one heat pump, could meet the room's thermal comfort requirements. To avoid energy waste, the final strategy of operating only one heat pump with one hour of preheating was adopted. After implementing this strategy, the heat pump's operating load rate increased from the original 20%-50% to 50%-80%, consistently operating within the high-efficiency range. Comparison showed that the original control method consumed 10090.98 kW over 14 days, while the optimized strategy reduced energy consumption to 8575.81 kW, achieving an energy saving of 15.02%. Furthermore, this virtual commissioning method can simultaneously optimize multiple parameters, including various water supply temperature settings. It was found that lowering the water supply temperature by 5°C in mild weather ensures both comfort and improved heat pump efficiency. Flow analysis indicates that variable flow control of the pump can achieve moderate energy savings, although these energy-saving methods are not as significant as optimizing the number of heat pumps operating and their operating time.

[0072] Step 50: The adaptive update cycle can be set to once a month. Virtual commissioning was conducted using a room preheating method of more than one hour in advance, with only one heat pump running. Finally, after virtual commissioning evaluation demonstrated its reliability and excellent energy efficiency, the optimized commissioning scheme was implemented in the actual system. This confirmed that the optimization method maintained system reliability and room comfort. Furthermore, by using this control optimization alone, without any hardware investment costs, it successfully saved £22,000 annually, verifying the advanced nature and economic efficiency of the invention.

[0073] Compared with existing technologies, the method of this invention introduces a high-fidelity physical simulation proxy model and a periodic parameter calibration mechanism, resulting in higher model accuracy. Simultaneously, it overcomes the limitations of traditional ground source heat pump system commissioning methods, which rely on one-time, static verification. It can quantify the energy-saving effects of various commissioning schemes and verify the safe and stable operation of the system in a risk-free simulation environment, avoiding the risks of on-site commissioning. Without requiring hardware investment, it ensures the long-term adaptability and efficiency of the operating strategy and the long-term energy efficiency of the system, fully demonstrating the significant advantages of this invention.

[0074] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the specific embodiments described above. The specific embodiments and descriptions in the specification are merely for further illustrating the principles of the invention. The basic principles, main features, and advantages of the present invention have been shown and described above without departing from the spirit and scope of the invention. Those skilled in the art should understand that various changes and modifications will be made, and all such changes and modifications fall within the scope of the present invention as claimed.

Claims

1. An adaptive commissioning method for a ground source heat pump system based on Modelica, characterized in that, Includes the following steps: Step 10: Collect operating parameter data of the ground source heat pump system and analyze the operating strategy of the ground source heat pump system; Step 20: Based on the structure and operation strategy of the ground source heat pump system, construct a physical simulation proxy model of the ground source heat pump system using Modelica; The parameters of the physical simulation surrogate model are adaptively calibrated using runtime parameter data to obtain a calibrated model; Step 30: Verify the accuracy of the calibration model using operating parameter data from another independent time period to obtain the calibration model; Step 40: Perform adaptive virtual debugging on the calibration model to optimize and determine the operating strategy; Step 50: Periodically collect the operating parameter data of the ground source heat pump system, calibrate the parameters of the calibration model, and update the calibration model; In step 20, based on the structure and operation strategy of the ground source heat pump system, a physical simulation proxy model of the ground source heat pump system is constructed using Modelica, specifically including: Step 201: Based on the Modelica model library, call the relevant components of the ground source heat pump system and connect them according to the structural topology of the ground source heat pump system. Set the control components according to the operation strategy to obtain the preliminary structural framework of the physical simulation proxy model of the ground source heat pump system. Step 202: Based on the rated parameters of the equipment in the ground source heat pump system and the system design information, set the parameters of the model components in the preliminary structural framework of the physical simulation proxy model to obtain the physical simulation proxy model of the ground source heat pump system. The model components include heat pump components, buffer tank components, underground pipe heat exchanger components, and room components; In step 20, the parameters of the physical simulation proxy model are adaptively calibrated using the running parameter data to obtain a calibrated model, specifically including: Step 203: Based on the relevant operating parameter data of the heat pump, the parameters of the heat pump components are adaptively calibrated by using data fitting. Step 204: Based on simulated building load data, sensitivity analysis is used to screen parameters, and Bayesian inference methods are used to adaptively calibrate the parameter settings of room components.

2. The adaptive commissioning method for a ground source heat pump system based on Modelica according to claim 1, characterized in that, The operating parameters include the main pipe temperature at the inlet of the heat pump evaporator, the main pipe temperature at the outlet of the heat pump evaporator, the main pipe temperature at the inlet of the condenser, the main pipe temperature at the outlet of the condenser, the inlet temperature of the buried pipe loop, the outlet temperature of the buried pipe loop, the supply water temperature of the heating circuit between the building and the buffer tank, the return water temperature between the building and the buffer tank, the operating power of each heat pump, the water flow rate of the heating circuit between the heat pump and the buffer tank, the water flow rate of the cooling circuit between the heat pump and the buffer tank, the water flow rate of the heating circuit between the building and the buffer tank, the water flow rate of the cooling circuit between the building and the buffer tank, and the fluid flow rate of the buried pipe loop.

3. The adaptive commissioning method for a ground source heat pump system based on Modelica according to claim 1, characterized in that, The operating strategies include the start-up and shutdown strategies for heat pumps, the control strategies for heating and hot water pumps, the control strategies for chilled water pumps, the control strategies for buried pipe loop fluid pumps, and the temperature regulation strategies for buffer tanks.

4. The adaptive commissioning method for a ground source heat pump system based on Modelica according to claim 1, characterized in that, In step 20, the parameters of the heat pump component include the heating capacity, cooling capacity, condenser design water flow rate, evaporator design water flow rate, evaporator inlet and outlet design temperature difference, condenser inlet and outlet design temperature difference, coefficient of performance under heating conditions, and coefficient of performance under cooling conditions. The parameters of the buffer tank assembly include the tank's volume, height, thickness of the external insulation layer, thermal conductivity of the insulation material, and mixing time constant inside the tank. The parameters of the buried pipe heat exchanger assembly include fluid flow rate under design conditions, borehole radius, soil type, packing type, total borehole depth, pipe pair spacing, pipe radius, pipe wall thickness, and local average annual soil temperature. The parameters of the room components include the number of air changes per minute.

5. The adaptive commissioning method for a ground source heat pump system based on Modelica according to claim 1, characterized in that, Step 203 specifically includes: Based on the collected operating parameter data, the heat pump heating capacity, heat pump cooling capacity, heat pump operating load rate under heating condition, and heat pump operating load rate under cooling condition are calculated. Then, the coefficient of performance (COP) and its decay coefficient under heating condition and the COP and its decay coefficient under cooling condition are calculated. The relationship between the decay coefficient under heating condition and the heat pump operating load rate under heating condition, and the relationship between the decay coefficient under cooling condition and the heat pump operating load rate under cooling condition are fitted. The calibration parameters obtained from the fitted parameters are set in the heat pump component of the physical simulation proxy model.

6. The adaptive commissioning method for a ground source heat pump system based on Modelica according to claim 1, characterized in that, Step 204 specifically includes: In the physical simulation proxy model, different combinations of building envelope and building thermal parameters are set to generate simulated building load data, and a substitute model is constructed using the random forest method. Based on the substitute model, the first-order sensitivity index and total effect sensitivity index of the building envelope heat transfer coefficient, heat capacity, thickness, area, density, and room volume are calculated using the improved Sobol sampling method, and a set of key parameters higher than a preset threshold are selected. For the key parameters, a uniform prior distribution is set and a likelihood function is constructed using the average absolute error between the predicted value and the actual measured value of the substitute model. The posterior distribution is solved using Bayes' theorem, and the posterior probability distribution of the key parameters is obtained using Markov chain Monte Carlo sampling. The parameter value at the maximum posterior probability density is selected as the final calibration result. The calibration parameters are set in the room component of the physical simulation proxy model.

7. The adaptive commissioning method for a ground source heat pump system based on Modelica according to claim 1, characterized in that, Step 30 specifically includes: Using the operating parameter data from another independent time period as input into the calibration model, the normalized average deviation error values ​​of all operating parameters are calculated using equation (16): Equation (16) In the formula, Indicates the first i One measurement value, Indicates the first i One simulated value, This represents the average value of the simulated values. n Indicates the number of data points; If the normalized average deviation error of all operating parameters is within ±15%, then the calibration model is used as the standardization model; otherwise, return to step 20.

Images