Ground source heat pump buried pipe system design method based on building load change

Abstract

The invention provides a ground source heat pump buried pipe system design method based on building load change, comprising the following steps: S1, acquiring geological survey data, meteorological data and building load data, and presetting buried pipe heat exchange system parameters; s2, establishing a multi-physics field coupling numerical model; s3, transient simulation is carried out through a multi-physics field coupling numerical model, and system short-term thermal response characteristic data corresponding to each candidate length are acquired and stored; s4, constructing and training a long-term dynamic performance prediction model through the long and short-term memory network model; and S5, based on a multi-objective optimization algorithm, solving a multi-objective optimization problem, obtaining a group of Pareto optimal buried pipe total length solution sets, and selecting a final buried pipe optimal total length from the optimal solution sets. According to the method, the design precision of the ground source heat pump buried pipe can be remarkably improved, the long-term dynamic prediction capacity is achieved, the reliability, economical efficiency and environment friendliness of long-term operation of a ground source heat pump system can be guaranteed, and the design efficiency and reliability are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of ground-source heat pump systems, and in particular relates to a design method for a ground-source heat pump buried pipe system based on building load changes. Background Art

[0002] Ground-source heat pump (GSHP) systems, as a highly efficient, energy-saving, and environmentally friendly renewable energy technology, have been widely used in building heating and cooling. Ground heat exchangers (GHEs) are a key component of GSHP systems, and their performance directly determines the efficiency, stability, and economic viability of the entire system. GHEs exchange heat with the soil through buried pipes, and one of their core design parameters is the total buried pipe length.

[0003] The determination of buried pipe length has a crucial impact on the initial investment and long-term operational performance of a GSHP system. If the buried pipe length is designed too short, the heat transfer load per unit pipe length will be excessive, potentially causing soil temperatures to cumulatively increase (primarily due to heat rejection in summer) or decrease (primarily due to heat extraction in winter) year after year, resulting in soil thermal imbalance. This not only significantly reduces system heat transfer efficiency and increases operating energy consumption, but in severe cases may even cause soil freezing or thermal saturation, rendering the system ineffective, shortening its lifespan, and irreversibly impacting the local geothermal environment. Conversely, if the buried pipe length is designed too long, while it can ensure effective heat transfer and long-term soil temperature stability, it will significantly increase initial investment costs and the operating energy consumption of the circulation pump, reducing the overall economic efficiency of the system.

[0004] Currently, the main methods for determining GHE buried pipe length in engineering practice include empirical estimation and design manual lookup tables. These methods are based on project location, building type, load estimates, and other factors, using empirical values ​​or simplified charts. These methods are overly crude and cannot adapt to the complexity and diversity of specific project geological conditions, soil characteristics, and actual operating conditions. The resulting design results are subject to significant deviations and high risks.

[0005] Simplified analytical / semi-empirical models: These models, such as those based on linear or cylindrical heat source theory, typically assume uniform soil properties, ignore the effects of groundwater seepage and water migration on heat transfer, and are often based on steady-state or simplified transient analyses. These models struggle to accurately reflect the complex coupled heat and mass transfer processes within the soil, as well as the dynamic heat accumulation effects of long-term system operation. Summary of the Invention

[0006] In order to solve the above problems in the prior art, a design method for a ground source heat pump buried pipe system based on building load changes is provided.

[0007] The technical solution adopted by the present invention to solve its technical problem is: This technical solution proposes a design method for a ground source heat pump buried pipe system based on building load changes, including the following steps: S1: Obtain geological survey data, meteorological data and building load data, and preset the parameters of the buried pipe heat exchange system; S2: Based on the acquired data, a multi-physics field coupling numerical model is established. The multi-physics field coupling numerical model couples the heat transfer process and the water migration process in the soil; S3: Input representative short-term operating condition data, perform transient simulation using a multi-physics field coupling numerical model, and obtain and store the system's short-term thermal response characteristic data corresponding to each candidate length; S4: Based on the generated system short-term thermal response characteristic data and the project's long-term load and meteorological data, a long-term dynamic performance prediction model is constructed and trained using a long-term short-term memory network model. The long-term dynamic performance prediction model is used to predict the time series of key performance indicators of the system over the long-term operation cycle; S5: Based on the multi-objective optimization algorithm, solve the multi-objective optimization problem and obtain a set of Pareto optimal solutions for the total length of the buried pipes. Based on the multi-criteria decision analysis method, select the final optimal total length of the buried pipes from the optimal solution set according to the specific project requirements and decision preferences.

[0008] Preferably, in S1, the parameters of the buried pipe heat exchange system include: preset buried pipe array layout, drill hole spacing, drill hole diameter, buried pipe type, pipe material, pipe diameter, wall thickness, type of circulating fluid in the pipe and its physical properties, thermal conductivity and physical properties of the filling material, and system design flow rate.

[0009] Preferably, in S2, a finite element analysis software platform is used to establish a three-dimensional or two-dimensional axisymmetric numerical model including the buried pipe and the surrounding soil, and the axisymmetric numerical model reflects the geometric structure of the borehole, U-shaped pipe, backfill material and each layer of soil.

[0010] Preferably, the axisymmetric numerical model is set to solve: solid and porous media heat transfer physics, porous media water flow physics, porous media and groundwater flow physics; Activate the heat-moisture coupling mechanism, consider the relationship between soil thermal conductivity and specific heat capacity as a function of moisture content and temperature, as well as the contribution of water migration to heat transfer, consider groundwater seepage and convection heat transfer, and establish a multi-physics field coupling numerical model; The multi-physics field coupling numerical model solution includes: energy conservation equation and mass conservation equation. The energy conservation equation includes conduction, convection, and phase change latent heat, and the mass conservation equation includes moisture migration.

[0011] Preferably, the boundary conditions are set for the multi-physics field coupling numerical model: Surface boundary: Apply comprehensive energy flux boundary conditions that take into account convective heat transfer, solar radiation absorption, long-wave radiation heat transfer, and evaporation latent heat loss; Far-field boundary: Based on the annual average ground temperature or geothermal gradient, the lateral and bottom boundaries are set as adiabatic, and the hydrological boundary is set according to actual conditions; Buried pipe inner wall boundary: This applies boundary conditions for heat exchange with the fluid inside the pipe. The boundary conditions are related to the fluid temperature, flow rate, and the convective heat transfer coefficient between the pipe wall and the fluid. The inner pipe boundary conditions reflect the operating logic of the heat pump unit and the outlet water temperature constraints. Grid division: Perform fine grid division in the area near the buried pipe and gradually transition to coarse grid towards the far field.

[0012] Preferably, in said S3, Simulation working condition setting: Select a series of representative candidate buried pipe total lengths L i , for each candidate length L i , using the established multi-physics coupled numerical model, representative short-term building loads and meteorological data were input to perform transient simulations; Data extraction: Run the simulation and record the length of each candidate L i The simulation results at key time nodes include the inlet and outlet fluid temperatures in the pipe, the average soil temperature around the buried pipe, the heat transfer per unit pipe length, and the soil moisture distribution. Database construction: All candidate lengths L i The corresponding short-term simulation results are organized into a structured database, which reflects the impact of different buried pipe lengths on the short-term thermal response characteristics of the system.

[0013] Preferably, in S4, the method for building a long-term dynamic performance prediction model includes: Model selection: Long short-term memory network model is selected, which is a recurrent neural network used to process and predict long time series dependency problems; Model input features: Design the input feature set of the long short-term memory network model, including time step, hourly building load forecast values ​​for many years in the future, typical meteorological year data, and the total length of candidate buried pipes. L i ; Model output target: Set the prediction target of the long short-term memory network model, including the future operation cycle, each candidate length L i The corresponding time series of key performance indicators include the fluid temperature at the underground pipe outlet and the average soil temperature around the underground pipe; Model training: Based on the structured database generated in S3 and combined with long-term load and meteorological data, the long short-term memory network model is trained. The long short-term memory network model learns and replaces the finite element analysis model. Rapidly predict the long-term thermal performance evolution trend under different buried pipe lengths. Use standard machine learning processes for training, validation, and testing, and adjust the network structure and hyperparameters to achieve optimal prediction accuracy.

[0014] Preferably, in S5, solving the multi-objective optimization problem includes the following steps: Optimization variable: Set the total length of the buried pipe L is the main optimization design variable; Objective function definition: Based on long-term prediction results, define conflicting optimization objective functions. f 1( L ), f 2( L )and f 3( L ); f 1( L ) To minimize the initial investment cost, the investment cost is proportional to the total drilling length. The objective formula is expressed as: min f 1( L )=( C drill + C material ) L + C fixed (1); Where, C drill is the drilling cost per unit length, C material is the pipe and backfill cost per unit length, C fixed is a fixed cost; f 2( L ) To minimize long-term operating costs, the operating costs are composed of water pump energy consumption and heat pump compressor energy consumption. The water pump energy consumption is proportional to the total pipe length. L The heat pump efficiency is related to the flow rate and the temperature of the fluid entering and exiting the buried pipe. Based on the long-term hourly outlet fluid temperature predicted by the long-short-term memory network model, the total operating energy consumption or average operating efficiency over the entire life cycle is calculated. The target formula is expressed as: minf 2( L )= min ( Energy_total ( L )) or maxf 2(L )= max ( COP_avg ( L )) (2); Where, Energy_total ( L ) is the total operating energy consumption, COP_avg ( L ) is the average operating efficiency; f 3( L ): Maximize soil thermal sustainability, used to measure the impact of long-term operation on the thermal balance of the underground environment. Based on the long-term soil temperature predicted by the long-short-term memory network model, the change in the average soil temperature at the end of the operation cycle relative to the initial temperature is calculated. The formula is expressed as: min f 3( L )=∣ T soil-avg ( t = end , L )- T soil-avg ( t =0)|(3); Where, T soil-avg ( t = end , L ) is the long-term soil temperature predicted by the LSTM model at the end of the operation period, T soil-avg ( t =0) is the initial soil average temperature, T soil The change in the average soil temperature at the end of the operation cycle relative to the initial temperature reflects the impact of the long-term operation of the buried pipe system on the thermal balance of the underground environment; Constraints: Set actual engineering constraints, including the outlet fluid temperature limit of the buried pipe. During the entire operating cycle, the outlet temperature must be maintained within the allowable operating range of the heat pump unit. The soil temperature limit is set. The soil temperature change caused by long-term operation must be within the environmentally acceptable range. The maximum drilling depth limit is the product of the maximum allowable depth of a single hole and the number of holes drilled.

[0015] Preferably, a second generation non-dominated sorting genetic algorithm is selected to handle multi-objective optimization problems, including: setting the algorithm parameters: population size, number of iterations, crossover probability, and mutation probability; Execution optimization: During its iteration, the algorithm generates a series of candidate pipe lengths. L For each candidate L , call the trained long short-term memory network model to quickly predict its corresponding long-term performance indicators; Calculate three objective functions based on the prediction results f 1( L ), f 2( L )and f 3( L ) value, the population is continuously evolved through selection, crossover, and mutation operations, and finally converges to obtain a set of Pareto optimal solutions for the total length of buried pipes.

[0016] Preferably, Pareto front visualization: the Pareto optimal solution set found by the second generation non-dominated sorting genetic algorithm is visualized in the multidimensional target space, and each point on the Pareto front represents a non-dominated solution; Decision support: Analyze the Pareto front to reveal the quantitative trade-offs between different design objectives. Based on the specific needs of the project, budget constraints, environmental requirements, and the preferences of decision makers, a multi-criteria decision analysis method is used, or interactive selection is made directly on the Pareto front to select the optimal length of the buried pipe from the optimal solution set that best meets the comprehensive project requirements.

[0017] Compared with the prior art, the present invention has the following advantages: 1. This application establishes a refined soil heat and moisture coupling multi-physics field model and uses finite element analysis units for numerical simulation, which can more accurately reflect the complex heat and mass transfer processes underground, especially the impact of moisture migration on long-term heat transfer performance, significantly improves the simulation accuracy, and has high precision.

[0018] 2. This application uses an LSTM neural network model, combined with short-term high-precision simulation data for training, to quickly and accurately predict the dynamic performance evolution of the buried pipe system (such as changes in fluid temperature and soil temperature) over a decades-long operating cycle. This overcomes the difficulty of traditional methods in making long-term accurate predictions and has long-term dynamic prediction capabilities.

[0019] 3. This application clearly defines three key and mutually influential optimization objectives: initial investment cost, long-term operating efficiency (or cost), and soil thermal sustainability. Using the NSGA-II multi-objective optimization algorithm, it systematically seeks a set of Pareto optimal solutions that achieve the best balance between these objectives, rather than a suboptimal solution for a single objective, thus achieving systematic multi-objective optimization.

[0020] 4. This application combines high-precision simulation, rapid prediction models and automated optimization algorithms, which greatly reduces the tedious manual trial and error process in traditional design and improves design efficiency. At the same time, the optimization results obtained based on long-term prediction and multi-objective trade-offs can better ensure the long-term reliability, economy and environmental friendliness of the ground source heat pump system, thereby improving design efficiency and reliability.

[0021] 5. The Pareto optimal solution set and its visualization provided in this application provide decision makers with a clear quantitative trade-off basis, which helps them make more scientific and reasonable final design decisions based on the specific circumstances of the project. The method framework is flexible and can adapt to different geological conditions, climate characteristics, building load patterns and GHE system configurations. It has strong engineering application value and is widely applicable. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments with reference to the accompanying drawings, in which: Figure 1 It is the overall flow chart of the present invention; Figure 2 It is a schematic diagram of the calculation domain in the present invention; Figure 3 This is a schematic diagram of the LSTM prediction results in the present invention; Figure 4 Schematic diagram of the Pareto front in the present invention. DETAILED DESCRIPTION

[0023] The following describes embodiments of the present invention in detail. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended only to explain the present invention and are not to be construed as limiting the present invention.

[0024] Example 1 like Figure 1-Figure 4 As shown, this embodiment proposes a design method for a ground source heat pump buried pipe system based on building load changes, including the following steps: S1: Obtain geological survey data, meteorological data and building load data, and preset the ground heat exchange (GHE) system parameters; S2: Based on the acquired data, a multi-physics field coupling numerical model is established. The multi-physics field coupling numerical model couples the heat transfer process and the water migration process in the soil; S3: Input representative short-term operating condition data, perform transient simulation using a multi-physics field coupling numerical model, and obtain and store the system's short-term thermal response characteristic data corresponding to each candidate length; S4: Based on the generated system short-term thermal response characteristic data and the project's long-term load and meteorological data, a long-term dynamic performance prediction model is constructed and trained using a long-term short-term memory network model. The long-term dynamic performance prediction model is used to predict the time series of key performance indicators of the system over the long-term operation cycle; S5: Based on the multi-objective optimization algorithm, solve the multi-objective optimization problem and obtain a set of Pareto optimal solutions for the total length of the buried pipes. Based on the multi-criteria decision analysis method, the final optimal total length of the buried pipes is selected from the optimal solution set according to the specific project requirements and decision preferences.

[0025] In S1, geological survey data include: site soil layer structure, thickness of each soil layer, rock and soil type, density, specific heat capacity, thermal conductivity (dry and wet states), permeability, porosity, moisture content and other thermophysical and hydrogeological parameters.

[0026] Meteorological data include: hourly or daily meteorological parameters of the project location over many years, including ambient temperature, solar radiation intensity, wind speed, humidity, etc.

[0027] Building load data includes: hourly heating / cooling load data of the target building obtained through detailed energy consumption simulation or actual monitoring, which must cover at least one full year and reflect future operating modes (such as operating schedule, set temperature, etc.).

[0028] Ground heat exchange (GHE) system parameters include: preset ground pipe array layout (such as rectangular, L-shaped, etc.), drill hole spacing, drill hole diameter, ground pipe type (single U, double U, coaxial), pipe material (such as HDPE), pipe diameter, wall thickness, type of circulating fluid in the pipe (such as water, ethylene glycol solution) and its physical properties, thermal conductivity and physical properties of the filling material (such as bentonite cement slurry), and system design flow rate.

[0029] In S2, a finite element analysis software platform, specifically COMSOL Multiphysics®, was used to create a three-dimensional or two-dimensional axisymmetric numerical model of the buried pipe (single hole or array) and the surrounding soil. The axisymmetric numerical model reflects the geometry of the borehole, U-shaped pipe, backfill material, and various soil layers.

[0030] Set up solutions for axisymmetric numerical models: heat transfer in solids and porous media, flow in porous media (Richards equations), and flow in porous media and groundwater. Enable the thermal-humidity coupling mechanism, consider the relationship between soil thermal conductivity and specific heat capacity as a function of moisture content and temperature, as well as the contribution of water migration (liquid water and gaseous water) to heat transfer (sensible heat and latent heat), consider groundwater seepage and convection heat transfer (if there is a significant effect), and establish a multi-physics field coupling numerical model; The multi-physics coupled numerical model solution includes: energy conservation equation and mass conservation equation. The energy conservation equation includes conduction, convection, and phase change latent heat. The mass conservation equation includes moisture migration, such as the Philip and de Vries model or similar heat and moisture coupling models.

[0031] For the multi-physics field coupling numerical model, boundary conditions are set: Surface boundary: Apply comprehensive energy flux boundary conditions that consider convective heat transfer (based on air temperature and wind speed), solar radiation absorption, longwave radiation heat transfer, and evaporative latent heat loss (based on humidity and soil surface moisture content); Far-field boundary: Based on the annual average ground temperature or geothermal gradient, the lateral and bottom boundaries are set as adiabatic (based on the annual average ground temperature or geothermal gradient), and the hydrological boundary is set according to actual conditions; Buried pipe inner wall boundary: This applies boundary conditions for heat exchange with the fluid inside the pipe. The boundary conditions are related to the fluid temperature, flow rate, and the convective heat transfer coefficient between the pipe wall and the fluid. The inner pipe boundary conditions reflect the operating logic of the heat pump unit and the outlet water temperature constraints. Grid division: Perform fine grid division in the area near the buried pipe and gradually transition to coarse grid towards the far field.

[0032] In S3, the simulation working condition setting is: a series of representative candidate buried pipe total lengths are selected L i , for each candidate length L i , using the established multi-physics coupled numerical model, input representative short-term (such as a full year or a characteristic period containing extreme loads) building load and meteorological data to perform transient simulation; Data extraction: Run the simulation and record the length of each candidate L i Simulation results at key time points (e.g., hourly or daily), including the inlet and outlet fluid temperatures, the average soil temperature around the buried pipe (or the temperature at a specific depth / radius), the heat transfer per unit pipe length, and the soil moisture distribution; Database construction: All candidate lengths L i The corresponding short-term simulation results (time series data) are organized into a structured database, which reflects the impact of different buried pipe lengths on the short-term thermal response characteristics of the system.

[0033] In S4, the method for building a long-term dynamic performance prediction model includes: Model selection: Long Short-Term Memory (LSTM) is used as a recurrent neural network (RNN) to process and predict long time series dependencies. Model input features: Design the input feature set of the long short-term memory network model, including time step (or date information), hourly building load forecast values ​​for many years in the future, typical meteorological year data, and the total length of candidate buried pipes. L i ; Model output target: Set the prediction target of the long short-term memory network model, including the future operation period (such as 20-30 years), each candidate length L i The corresponding time series of key performance indicators include the temperature of the fluid at the outlet of the buried pipe ( T out ( t , L i ))、Average soil temperature around buried pipes( T soil_avg ( t , L i )); Model training: Based on the structured database generated in S3 and combined with long-term load and meteorological data, the long short-term memory network model is trained. The long short-term memory network model learns and replaces the finite element analysis model. Rapidly predict the long-term (decades) thermal performance evolution trend of different buried pipe lengths. Use standard machine learning processes for training, validation, and testing, and adjust the network structure (number of layers and units) and hyperparameters to achieve optimal prediction accuracy.

[0034] In S5, solving the multi-objective optimization problem includes the following steps: Optimization variable: Set the total length of the buried pipe L is the main optimization design variable; Objective function definition: Based on long-term prediction results, define conflicting optimization objective functions. f 1( L ), f 2( L )and f 3( L ); f 1( L ) To minimize initial investment cost and maximize operating efficiency, the investment cost is proportional to the total drilling length. The objective formula is: min f 1( L )=( C drill + C material ) L + C fixed (1); Where, C drill is the drilling cost per unit length, C material is the pipe and backfill cost per unit length, C fixedis a fixed cost; f 2( L ) To minimize long-term operating costs, the operating costs are composed of water pump energy consumption and heat pump compressor energy consumption. The water pump energy consumption is proportional to the total pipe length. L The heat pump efficiency (COP / EER) is related to the flow rate and the fluid temperature entering and exiting the buried pipe. Based on the long-term hourly outlet fluid temperature (Tout(t,L)) predicted by the long-term short-term memory network model, the total operating energy consumption or average operating efficiency over the entire life cycle is calculated. The target formula is expressed as: minf 2( L )= min ( Energy_total ( L )) or maxf 2( L )= max ( COP_avg ( L )) (2); Where, Energy_total ( L ) is the total operating energy consumption, COP_avg ( L ) is the average operating efficiency; f 3( L ): Maximize soil thermal sustainability, used to measure the impact of long-term operation on the thermal balance of the underground environment. Based on the long-term soil temperature predicted by the long-short-term memory network model, calculate the change in the average soil temperature at the end of the operation cycle (such as the end of the 20th year) relative to the initial temperature. The formula is expressed as: min f 3( L )=∣ T soil-avg ( t = end , L )- T soil-avg ( t =0)|(3); Where, T soil-avg ( t = end , L ) is the long-term soil temperature predicted by the LSTM model at the end of the operation period, T soil-avg ( t =0) is the initial soil average temperature, T soil The change in the average soil temperature at the end of the operation cycle relative to the initial temperature reflects the impact of the long-term operation of the buried pipe system on the thermal balance of the underground environment; Constraints: Set actual engineering constraints and limit the temperature of the fluid at the outlet of the buried pipe. During the entire operation cycle, the outlet temperature must be maintained within the operating range allowed by the heat pump unit (such as T min ≤ T out ( t , L )≤ T max ); soil temperature limit, the soil temperature change caused by long-term operation must be within the acceptable range of the environment (such as Δ T soil ≤Δ T limit The maximum drilling depth limit is the product of the maximum allowable depth of a single hole and the number of holes to be drilled: L / N boreholes ≤ D max ( N boreholes is the number of holes drilled, D max is the maximum allowable depth of a single hole).

[0035] The second generation of the Non-dominated Sorting Genetic Algorithm (NSGA-II) was used to solve the multi-objective optimization problem, including setting the algorithm parameters: population size (e.g., 100 individuals), number of iterations / generations (e.g., 200 generations), crossover probability (e.g., 0.9), and mutation probability (e.g., 0.1). Execution optimization: During its iteration, the algorithm generates a series of candidate pipe lengths. L For each candidate L , call the trained long short-term memory network model to quickly predict its corresponding long-term performance indicators ( T out ( t , L ), T soil_avg ( t , L )); Calculate three objective functions based on the prediction results f 1( L ), f 2( L )and f 3( L ) value, the population is continuously evolved through selection, crossover, and mutation operations, and finally converges to obtain a set of Pareto optimal solutions for the total length of buried pipes.

[0036] Pareto front visualization: The Pareto optimal solution set found by the second-generation non-dominated sorting genetic algorithm is visualized in a multidimensional objective space (e.g., a two-dimensional scatter plot showing cost versus efficiency, or cost versus sustainability). Each point on the Pareto front represents a non-dominated solution (i.e., no objective can be improved without sacrificing other objectives). Decision support: Analyze the Pareto front to reveal the quantitative trade-offs between different design objectives. Based on the specific project needs, budget constraints, environmental requirements, and the decision maker's preferences, use multi-criteria decision analysis methods (such as weighted scoring, TOPSIS, and ideal point methods), or make interactive selections directly on the Pareto front to select the optimal length of the buried pipe from the optimal solution set that best meets the comprehensive project requirements.

[0037] Example 2 like Figure 1-Figure 4 As shown, this embodiment proposes a design method for a ground source heat pump buried pipe system based on building load changes, including the following steps: First, a sophisticated multi-physics model of coupled soil heat and moisture is established using finite element software (such as COMSOL Multiphysics®) for short-term, high-precision simulations. The simulated data is then used to train a long short-term memory (LSTM) machine learning model to rapidly predict the long-term dynamic performance (e.g., fluid and soil temperatures) of buried pipe systems with varying pipe lengths over the next several decades. Next, a multi-objective optimization problem is defined, encompassing initial investment, operational efficiency, and soil sustainability. This problem is solved using a non-dominated sorting genetic algorithm (NSGA-II) combined with an LSTM prediction model to obtain a Pareto-optimal solution set. Finally, a decision analysis is performed to select the final pipe length from this optimal solution set.

[0038] Basic data preparation: Collect the project's geological survey report (obtaining soil stratification, thermophysical parameters, and hydrogeological information), local typical annual meteorological data, use the building energy consumption simulation software DeST to calculate the building's hourly cooling and heating loads throughout the year, and preset GHE parameters (such as double U-tubes, HDPE material, 150mm hole diameter, 5m hole spacing, rectangular arrangement, backfill thermal conductivity of 1.5W / (m·K), and design flow rate of 0.5L / s).

[0039] Establish a multi-physics model: Start COMSOL Multiphysics® software. Establish a 3D geological model based on the geological survey data, including the surface, soil layers, boreholes, double U pipes and backfill areas ( Figure 2 is a schematic diagram of the calculation domain in the present invention, Figure 2 Schematic diagram of the computational domain of the buried pipe and surrounding soil area used for multiphysics coupled simulation, showing the key geometric structures, physical fields, and boundary conditions.

[0040] Select the Subsurface Flow module to simulate water flow in porous media (accounting for unsaturated properties and using the Richards equation), and the Heat Transfer in Porous Media module to simulate heat transfer. Enable the Multiphysics interface to couple water flow and heat transfer, and set soil thermal conductivity and specific heat capacity functions that vary with moisture content and temperature.

[0041] The surface setting includes complex boundary conditions of convection, radiation, and evaporation ( Figure 2 Mark), the far-field boundary is set as adiabatic and without water flux, a convective heat transfer boundary is applied to the inner wall of the U-tube, the heat source / sink terms are associated with the temperature and flow rate of the fluid in the tube, and fine meshing is performed.

[0042] Short-term simulation and data generation: A series of candidate total pipe lengths are selected, e.g. L =[100m,150m,200m,…,250m]. L i , using COMSOL model, input the hourly load of the first year and typical annual meteorological data, and conduct a one-year transient simulation. Li The hourly inlet and outlet water temperature, pipe wall temperature, surrounding soil temperature field, moisture content field and other data are collected to form a short-term feature database.

[0043] Training LSTM prediction model: Build an LSTM network whose input layer receives the time step, hourly load for the next 20 years, meteorological data, and candidate pipe lengths. L i The output layer predicts the hourly outlet water temperature for the next 20 years T out ( t , L i ) and average soil temperature T soil_avg ( t , L i ).

[0044] Use the generated structured database (it may be necessary to expand the simulation period or use characteristic daily / weekly simulation data to cover more dynamics) and long-term input data to train and verify the LSTM. After training, the LSTM model can be used to calculate the pipe length according to the input. L , quickly predict its performance in the next 20 years ( Figure 3 This is a schematic diagram of the LSTM prediction results in the present invention. Figure 3 This is a schematic diagram of the LSTM model prediction results, showing the changing trend of the fluid temperature or average soil temperature at the underground pipe outlet in the coming years under different candidate pipe lengths.

[0045] Perform multi-objective optimization: Define optimization variables: total pipe length L， Define the objective function: min f 1( L )=( C drill + C material ) L + C fixed (1); min f 2( L )= min ( Energy_total ( L )) (2); min f 3( L )=∣ T soil-avg ( t = end , L )- T soil-avg ( t =0)|(3); f 2( L )= ∑[ P pump ( L , t ) + P compressor ( T in ( L , t ), T out ( L , t ))]·∆ t , minimize the total operating energy consumption for 20 years, where T in , T out Prediction by LSTM, P compressor Depending on the heat pump performance curve, set the constraints: for example, 4°C ≤ T out (t,L)≤45°C; |Δ T soil |≤3°C.

[0046] Perform NSGA-II optimization: set the population size to 100 and iterate 200 times. In each iteration, for each individual in the population (representing a tube length L), calling the trained LSTM model to predict its long-term performance, calculating the values ​​of two objective functions, and checking the constraints. The algorithm drives the population toward the Pareto frontier through genetic operations (selection, crossover, and mutation).

[0047] Decision: After the optimization is completed, a set of Pareto optimal solutions are obtained. Figure 4 , Figure 4 is a schematic diagram of the Pareto frontier in the present invention, Figure 4 A schematic diagram of exemplary NSGA-II optimization results, showing the Pareto optimal frontier in a two-dimensional objective space (e.g., initial cost and long-term operating efficiency comparison).

[0048] Figure 4 The trade-off between initial cost and operating energy consumption is shown, and decision makers can make decisions based on the project budget (limit f 1) and the degree of concern for operating costs and environmental impact (influence f 2 and f 3), and select an optimal point on the Pareto frontier.

[0049] For example, point A represents the lowest initial cost but higher operating energy consumption; point B represents the lowest operating energy consumption but the highest initial cost; and point C represents a compromise between the two. f Considering 3, the optimal total length of buried pipe is finally determined L optimal .

[0050] Through the above steps, the present invention can provide a specific engineering project with an optimal length of buried pipe obtained through detailed simulation, long-term prediction and multi-objective trade-off optimization, which is significantly better than traditional design methods.

[0051] By establishing a refined soil thermal and moisture coupling multi-physics field model and using finite element analysis units for numerical simulation, the complex underground heat and mass transfer processes, especially the impact of moisture migration on long-term heat transfer performance, can be more accurately reflected, significantly improving the simulation accuracy and achieving high precision.

[0052] By adopting the LSTM neural network model and combining it with short-term high-precision simulation data for training, it is possible to quickly and accurately predict the dynamic performance evolution of the buried pipe system (such as changes in fluid temperature and soil temperature) over a decades-long operating cycle. This overcomes the difficulty of traditional methods in making long-term accurate predictions and has the ability to make long-term dynamic predictions.

[0053] By clearly defining three key and mutually influential optimization objectives: initial investment cost, long-term operating efficiency (or cost), and soil thermal sustainability, and applying the NSGA-II multi-objective optimization algorithm, we can systematically find a set of Pareto optimal solutions that achieve the best balance between these objectives, rather than a suboptimal solution for a single objective, thus achieving systematic multi-objective optimization.

[0054] By combining high-precision simulation, rapid prediction models and automated optimization algorithms, the tedious manual trial-and-error process in traditional design has been greatly reduced, and design efficiency has been improved. At the same time, the optimization results obtained based on long-term prediction and multi-objective trade-offs can better ensure the long-term reliability, economy and environmental friendliness of the ground source heat pump system, thereby improving design efficiency and reliability.

[0055] The provided Pareto optimal solution set and its visualization provide decision makers with a clear quantitative trade-off basis, which helps them make more scientific and reasonable final design decisions based on the specific conditions of the project. The method framework is flexible and can adapt to different geological conditions, climate characteristics, building load patterns and GHE system configurations. It has strong engineering application value and is widely applicable.

[0056] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

Claims

1. A design method for a ground source heat pump buried pipe system based on building load changes, characterized in that: The following steps are involved: S1: Obtain geological survey data, meteorological data and building load data, and preset the parameters of the buried pipe heat exchange system; S2: Based on the acquired data, a multi-physics field coupling numerical model is established. The multi-physics field coupling numerical model couples the heat transfer process and the water migration process in the soil; S3: Input representative short-term operating condition data, perform transient simulation using a multi-physics field coupling numerical model, and obtain and store the system's short-term thermal response characteristic data corresponding to each candidate length; S4: Based on the generated system short-term thermal response characteristic data and the project's long-term load and meteorological data, a long-term dynamic performance prediction model is constructed and trained using a long-term short-term memory network model. The long-term dynamic performance prediction model is used to predict the time series of key performance indicators of the system over the long-term operation cycle; S5: Based on the multi-objective optimization algorithm, solve the multi-objective optimization problem and obtain a set of Pareto optimal solutions for the total length of the buried pipes. Based on the multi-criteria decision analysis method, select the final optimal total length of the buried pipes from the optimal solution set according to the specific project requirements and decision preferences.

2. A method for designing a ground source heat pump buried pipe system based on building load changes according to claim 1, characterized in that: In S1, the buried pipe heat exchange system parameters include: preset buried pipe array layout, drilling spacing, drilling diameter, buried pipe type, pipe material, pipe diameter, wall thickness, type of circulating fluid in the pipe and its physical properties, thermal conductivity and physical properties of filling material, and system design flow rate.

3. The method for designing a ground source heat pump buried pipe system based on building load changes according to claim 1, characterized in that: In S2, a finite element analysis software platform is used to establish a three-dimensional or two-dimensional axisymmetric numerical model including the buried pipe and the surrounding soil. The axisymmetric numerical model reflects the geometric structure of the borehole, U-shaped pipe, backfill material and each layer of soil.

4. The method for designing a ground source heat pump buried pipe system based on building load changes according to claim 3, characterized in that: Set up solutions for axisymmetric numerical models: heat transfer in solids and porous media physics, water flow in porous media physics, and water flow in porous media and subsurface physics; Activate the heat-moisture coupling mechanism, consider the relationship between soil thermal conductivity and specific heat capacity as a function of moisture content and temperature, as well as the contribution of water migration to heat transfer, consider groundwater seepage and convection heat transfer, and establish a multi-physics field coupling numerical model; The multi-physics field coupling numerical model solution includes: energy conservation equation and mass conservation equation. The energy conservation equation includes conduction, convection, and phase change latent heat, and the mass conservation equation includes moisture migration.

5. The method for designing a ground source heat pump buried pipe system based on building load changes according to claim 4, characterized in that: For the multi-physics field coupling numerical model, boundary conditions are set: Surface boundary: Apply comprehensive energy flux boundary conditions that take into account convective heat transfer, solar radiation absorption, long-wave radiation heat transfer, and evaporation latent heat loss; Far-field boundary: Based on the annual average ground temperature or geothermal gradient, the lateral and bottom boundaries are set as adiabatic, and the hydrological boundary is set according to actual conditions; Buried pipe inner wall boundary: This applies boundary conditions for heat exchange with the fluid inside the pipe. The boundary conditions are related to the fluid temperature, flow rate, and the convective heat transfer coefficient between the pipe wall and the fluid. The inner pipe boundary conditions reflect the operating logic of the heat pump unit and the outlet water temperature constraints. Grid division: Perform fine grid division in the area near the buried pipe and gradually transition to coarse grid towards the far field.

6. The method for designing a ground source heat pump buried pipe system based on building load changes according to claim 1, characterized in that: In the S3, Simulation working condition setting: Select a series of representative candidate buried pipe total lengths L i , for each candidate length L i , using the established multi-physics coupled numerical model, representative short-term building loads and meteorological data were input to perform transient simulations; Data extraction: Run the simulation and record the length of each candidate L i The simulation results of key time nodes include , the temperature of the fluid inlet and outlet of the pipe, the average temperature of the soil around the buried pipe, the heat exchange per unit pipe length, and the distribution of soil moisture content; Database construction: All candidate lengths L i The corresponding short-term simulation results are organized into a structured database, which reflects the impact of different buried pipe lengths on the short-term thermal response characteristics of the system.

7. The method for designing a ground source heat pump buried pipe system based on building load changes according to claim 6, characterized in that: In S4, the method for building a long-term dynamic performance prediction model includes: Model selection: Long short-term memory network model is selected, which is a recurrent neural network used to process and predict long time series dependency problems; Model input features: Design the input feature set of the long short-term memory network model, including time step, hourly building load forecast values ​​for many years in the future, typical meteorological year data, and the total length of candidate buried pipes. L i ; Model output target: Set the prediction target of the long short-term memory network model, including the future operation cycle, each candidate length L i The corresponding time series of key performance indicators include the fluid temperature at the underground pipe outlet and the average soil temperature around the underground pipe; Model training: Based on the structured database generated in S3 and combined with long-term load and meteorological data, the long short-term memory network model is trained. The long short-term memory network model learns and replaces the finite element analysis model. Rapidly predict the long-term thermal performance evolution trend under different buried pipe lengths. Use standard machine learning processes for training, validation, and testing, and adjust the network structure and hyperparameters to achieve optimal prediction accuracy.

8. The method for designing a ground source heat pump buried pipe system based on building load changes according to claim 1, characterized in that: In S5, solving the multi-objective optimization problem includes the following steps: Optimization variable: Set the total length of the buried pipe L is the main optimization design variable; Objective function definition: Based on long-term prediction results, define conflicting optimization objective functions. f 1( L ), f 2( L )and f 3( L ); f 1( L ) To minimize the initial investment cost, the investment cost is proportional to the total drilling length. The objective formula is expressed as: min f 1( L )=（ C drill ＋ C material ）· L + C fixed （1）； Where, C drill is the drilling cost per unit length, C material is the pipe and backfill cost per unit length, C fixed is a fixed cost; f 2( L ) To minimize long-term operating costs, the operating costs are composed of water pump energy consumption and heat pump compressor energy consumption. The water pump energy consumption is proportional to the total pipe length. L The heat pump efficiency is related to the flow rate and the temperature of the fluid entering and exiting the buried pipe. Based on the long-term hourly outlet fluid temperature predicted by the long-short-term memory network model, the total operating energy consumption or average operating efficiency over the entire life cycle is calculated. The target formula is expressed as: minf 2( L )= min ( Energy_total ( L )) or maxf 2( L )= max ( COP_avg ( L )) (2); Where, Energy_total ( L ) is the total operating energy consumption, COP_avg ( L ) is the average operating efficiency; f 3( L ): Maximize soil thermal sustainability, used to measure the impact of long-term operation on the thermal balance of the underground environment. Based on the long-term soil temperature predicted by the long-short-term memory network model, the change in the average soil temperature at the end of the operation cycle relative to the initial temperature is calculated. The formula is expressed as: min f 3( L )=∣ T soil-avg ( t = end ， L )－ T soil-avg ( t =0)∣ （3）； Where, T soil-avg ( t = end , L ) is the long-term soil temperature predicted by the LSTM model at the end of the operation period, T soil-avg ( t =0) is the initial soil average temperature, T soil The change in the average soil temperature at the end of the operation cycle relative to the initial temperature reflects the impact of the long-term operation of the buried pipe system on the thermal balance of the underground environment; Constraints: Set actual engineering constraints, including the outlet fluid temperature limit of the buried pipe. During the entire operating cycle, the outlet temperature must be maintained within the allowable operating range of the heat pump unit. The soil temperature limit is set. The soil temperature change caused by long-term operation must be within the environmentally acceptable range. The maximum drilling depth limit is the product of the maximum allowable depth of a single hole and the number of holes drilled.

9. The method for designing a ground source heat pump buried pipe system based on building load changes according to claim 8, characterized in that: The second generation of non-dominated sorting genetic algorithm is used to deal with multi-objective optimization problems, including setting the algorithm parameters: population size, number of iterations, crossover probability, and mutation probability; Execution optimization: During its iteration, the algorithm generates a series of candidate pipe lengths. L For each candidate L , call the trained long short-term memory network model to quickly predict its corresponding long-term performance indicators; Calculate three objective functions based on the prediction results f 1( L ), f 2( L )and f 3( L ) value, the population is continuously evolved through selection, crossover, and mutation operations, and finally converges to obtain a set of Pareto optimal solutions for the total length of buried pipes.

10. The method for designing a ground source heat pump buried pipe system based on building load changes according to claim 9, characterized in that: Pareto front visualization: Visualize the Pareto optimal solution set found by the second generation non-dominated sorting genetic algorithm in a multidimensional target space. Each point on the Pareto front represents a non-dominated solution. Decision support: Analyze the Pareto front to reveal the quantitative trade-offs between different design objectives. Based on the specific needs of the project, budget constraints, environmental requirements, and the preferences of decision makers, a multi-criteria decision analysis method is used, or interactive selection is made directly on the Pareto front to select the optimal length of the buried pipe from the optimal solution set that best meets the comprehensive project requirements.

Images