Intelligent control method for building energy-saving wall and lattice structure wall applying the method

Abstract

The present application relates to an intelligent control method for building energy-saving walls and a lattice structure wall applying this method, which solves the problem that traditional heat insulation materials can reduce heat conduction, but have a relatively single function and cannot actively adjust heat. The method includes: setting weights for various input data according to different building types, usage scenarios and seasons, and calculating the total sum of input data by using the weighted average method; analyzing and determining the level corresponding to the thermal condition according to the mapping relationship between the interval range where the total sum of input data falls and the level; analyzing and determining the overall control plan according to the mapping relationship between the thermal condition level and the overall control plan, and implementing the determined overall control plan. The overall control plan includes a phase change material control plan, a lattice structure control plan, and a heat exchange fluid control plan. The present application has the following effects: accurately controlling the building thermal condition, improving the wall thermal performance, and achieving efficient energy saving and intelligent management.

Description

Technical Field

[0001] The present invention relates to the field of building energy conservation, and particularly to an intelligent control method for building energy-saving walls and a lattice structure wall applying the method. Background Art

[0002] Under the background of the increasingly severe global energy situation and the continuous improvement of people's requirements for the comfort of the building environment, the development of building energy-saving technologies is crucial. As an important field of energy consumption, buildings account for a relatively high proportion of the total social energy consumption. Therefore, improving the building energy-saving level has become a key link in alleviating the energy crisis and achieving sustainable development. As the core part of the building envelope structure, the wall plays a crucial role in building energy conservation, and its thermal performance directly affects the indoor thermal environment quality and the building energy consumption situation.

[0003] Currently, in the field of building wall energy-saving technologies, there are mainly two types of technical means. One is the traditional heat insulation and preservation technology, which improves the thermal resistance of the wall by using heat insulation materials such as polystyrene boards and rock wool to reduce heat conduction. These materials can reduce the heating and cooling loads of buildings to a certain extent and are widely used in the field of building energy conservation. The other is the application technology of phase change materials in walls. Phase change materials (PCMs) are introduced into building walls by virtue of their characteristics of absorbing and releasing a large amount of latent heat during the solid-liquid phase change process to alleviate indoor temperature fluctuations, reduce air conditioning and heating energy consumption, and thus improve indoor thermal comfort.

[0004] However, these related technologies have many drawbacks in practical applications. Although traditional heat insulation materials can reduce heat conduction, their functions are relatively single and they cannot actively regulate heat. At the same time, the current application of phase change materials in building walls lacks intelligent control means and cannot be dynamically adjusted and replaced according to seasonal changes, indoor temperature requirements, and the performance attenuation of phase change materials, which severely restricts the improvement of the wall thermal performance. Summary of the Invention

[0005] In order to accurately control the building thermal situation, improve the wall thermal performance, and achieve efficient energy conservation and intelligent management, the present application provides an intelligent control method for building energy-saving walls and a lattice structure wall applying the method.

[0006] In the first aspect, the present application provides an intelligent control method for building energy-saving walls, adopting the following technical solutions:

[0007] An intelligent control method for building energy-saving walls, comprising:

[0008] Obtaining environmental data and user behavior data, and performing data preprocessing;

[0009] Input environmental data and user behavior data into a pre - constructed heat demand prediction model to output building heat demand information;

[0010] Collect real - time monitoring data of parameters related to heat transfer, and input the real - time monitoring data into a pre - constructed multi - physical - field coupled heat transfer model. The multi - physical - field coupled heat transfer model calculates the heat transfer process of fine units of the wall structure based on the finite element analysis method. By solving the equations of multi - physical - field coupling, the analysis results of the heat transfer model are obtained. The analysis results of the heat transfer model include the temperature distribution, heat transfer rate at different positions of the wall at different times, and the phase change state of the phase change material;

[0011] Standardize the data included in the building heat demand information and the data included in the analysis results of the heat transfer model to make their ranges unified;

[0012] Set weights for each input data according to different building types, usage scenarios and seasons, and calculate the sum of input data using the weighted average method;

[0013] Analyze and determine the level corresponding to the heat condition according to the mapping relationship between the interval range where the sum of input data falls and the level;

[0014] Analyze and determine the overall regulation plan according to the mapping relationship between the heat condition level and the overall regulation plan, and execute the determined overall regulation plan. The overall regulation plan includes the phase change material regulation plan, the lattice structure regulation plan, and the heat exchange fluid regulation plan.

[0015] Optionally, the construction steps of the heat demand prediction model are as follows:

[0016] Collect environmental data and user behavior data, and perform data pre - processing;

[0017] Build a Transformer model based on spatio - temporal decoupling, randomly initialize the parameters of the model, and determine the hyperparameters of the model;

[0018] Divide the pre - processed data into a training set and a validation set according to a preset ratio;

[0019] Input the training set data into the Transformer model based on spatio - temporal decoupling, calculate the prediction results through forward propagation, use the weighted mean square error loss function to calculate the difference between the prediction results and the true heat demand values, and perform Dropout operations during each forward propagation to obtain prediction results under different parameter samples;

[0020] Based on the results of the loss function, calculate the gradient through the backpropagation algorithm and update the model parameters;

[0021] After each training cycle, calculate the validation set loss. If the change in the validation set loss is less than the set threshold for a preset number of consecutive cycles, it is considered that the training is complete.

[0022] Optionally, the steps for constructing the multi-physical field coupled heat transfer model are as follows:

[0023] Collect and obtain multi-source data and perform data preprocessing;

[0024] Build model architectures at three levels: microscopic, mesoscopic, and macroscopic;

[0025] According to the corresponding relationship between the building thermal management scenario and the physical field coupling method and intensity setting scheme, analyze and determine the physical field coupling method and intensity of the current building thermal management scenario as the starting state of model training;

[0026] Integrate the non-equilibrium heat flux equation into the model heat transfer calculation system, where is the heat flux density, is the thermal conductivity, is the temperature gradient, is the pressure gradient, is the mass concentration gradient, is the heat transfer coefficient related to pressure, is the heat transfer coefficient related to mass concentration;

[0027] Select a deep neural network with a multi-layer perceptron structure having 3 hidden layers and 100 neurons in each layer, and integrate it into the model using a residual connection fusion strategy. The input layer of the deep neural network receives temperature, humidity, and building structure parameter data. The hidden layer uses the ReLU function to extract features, and the output layer predicts heat transfer related parameters;

[0028] Input the preprocessed data into the model with the constructed architecture, determined coupling method, integrated non-equilibrium equation, and integrated deep neural network;

[0029] During training, calculate the deviation between the model prediction value and the true value using the mean square error loss function, and use the stochastic gradient descent method to iteratively update the model parameters at a learning rate of 0.01 to reduce the loss function value;

[0030] After every 10 parameter updates, calculate the influence weight of each physical field coupling intensity on the loss function based on the model prediction results and true data under the latest parameters. According to the weight, adjust the coupling intensity coefficient by 10% to reduce the loss function value;

[0031] According to the dynamic change results of the scenario within each time interval range, change the physical field coupling intensity through a preset coupling intensity adjustment function;

[0032] Continuously monitor the mean squared error. When the change values of the mean squared error in consecutive preset rounds of training are all less than the preset change value, it is determined that the training is completed.

[0033] In a second aspect, the present application provides a lattice structure wall, adopting the following technical solution:

[0034] A lattice structure wall includes:

[0035] A lattice structure framework;

[0036] A phase change material, filled in the internal space of the lattice structure framework. The phase change material is paraffin - based or hydrated salt - based, and is used to store and release heat;

[0037] A macroscopic encapsulation structure, used to encapsulate the phase change material inside the lattice structure framework;

[0038] Concrete, poured into a formwork embedded with the lattice structure framework, so that the lattice is tightly combined with the concrete.

[0039] Optionally, the lattice structure framework is a double - layer lattice structure framework. The inner layer lattice provides a stable filling space for the phase change material, and the outer layer lattice is arranged around the inner layer and is provided with a heat - exchange fluid circulation channel, and the channel is filled with a heat - exchange fluid. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 is a schematic flow chart of the intelligent control method for the building energy - saving wall in the embodiment of the present application.

[0041] Figure 2 is a schematic diagram of a single - layer lattice structure in the embodiment of the present application.

[0042] Figure 3 is Figure 2 the enlarged schematic diagram at position a in

[0043] Figure 4 is a schematic diagram of a double - layer lattice structure in the embodiment of the present application.

[0044] Figure 5 is Figure 4 the enlarged schematic diagram at position b in DETAILED DESCRIPTION OF THE EMBODIMENTS

[0045] The following further describes the present application in detail with reference to the accompanying drawings.

[0046] Referring to Figure 1 , an intelligent control method for a building energy - saving wall disclosed in the present application includes:

[0047] Step S100, obtain environmental data and user behavior data, and perform data pre - processing.

[0048] Environmental data refers to data related to the external environment of the building, including but not limited to external temperature, humidity, sunlight intensity, etc. These data will affect the building's thermal demand and the heat transfer process of the wall. Environmental data is obtained by reasonably arranging various sensors around the building. For example, temperature sensors are used to measure the external temperature, humidity sensors collect humidity data, and sunlight sensors monitor sunlight intensity. These sensors transmit the real-time monitored data to the data acquisition device.

[0049] User behavior data: data related to users’ activities in the building, such as entry and exit times, usage of indoor equipment (such as air conditioning on / off times, frequency of electrical appliance use), etc. These behaviors will change the indoor thermal environment.

[0050] The general process is as follows: various sensors and monitoring equipment continuously collect environmental data and user behavior data. After the data acquisition device collects these raw data, it first performs preliminary screening to eliminate erroneous or abnormal data, such as the obvious deviation value of the temperature sensor. Then, the data is formatted and standardized, and data of different types and dimensions are converted into a form suitable for subsequent analysis. Finally, pre-processed data is obtained to provide reliable data support for subsequent steps such as predicting building heat demand.

[0051] Step S200: inputting environmental data and user behavior data into a pre-built heat demand prediction model, and outputting building heat demand information.

[0052] Heat demand prediction model: Based on the Transformer model with time-space decoupling, it can learn the complex relationship between environmental data, user behavior data and building heat demand. Through training on a large amount of historical data, the model can predict the heat demand information of the building in the future based on the input current environment and user behavior data.

[0053] Building heat demand information: refers to the amount of heat that a building needs within a certain period of time to maintain a suitable indoor thermal environment, including heating, cooling and other needs. This information is crucial for building thermal management and energy-saving regulation.

[0054] Step S300, collect and obtain real-time monitoring data of heat transfer related parameters, and input the real-time monitoring data into a pre-built multi-physics field coupled heat transfer model. The multi-physics field coupled heat transfer model calculates the heat transfer process of the fine units of the wall structure based on the finite element analysis method, and obtains the heat transfer model analysis results by solving the multi-physics field coupling equations. The heat transfer model analysis results include the temperature distribution at different locations of the wall at different times, the heat transfer rate, and the phase change state of the phase change material.

[0055] Real-time monitoring data of heat transfer related parameters: Refers to various data directly related to heat transfer collected in real time during the heat transfer process of building walls, such as the temperature at different positions of the wall, heat flux density, phase change state of phase change materials, etc. These data reflect the real-time dynamic situation of heat transfer in the wall.

[0056] Multi-physical-field coupled heat transfer model: A heat transfer model that comprehensively considers the interactions of multiple physical fields. In the present invention, this model integrates physical processes such as heat conduction, convection, radiation, and the phase change process of phase change materials, and is used to accurately simulate the heat transfer behavior of the wall structure. It is based on the finite element analysis method, and divides the wall structure into fine units for calculation.

[0057] Finite element analysis method: A numerical calculation method that discretizes the continuous wall structure into a finite number of units, conducts mechanical and thermal analyses on each unit, and then combines these units to solve the mechanical and thermal responses of the entire structure. In this step, it is used to calculate the heat transfer process of the fine units of the wall structure.

[0058] Analysis results of the heat transfer model: Results obtained by calculating through the multi-physical-field coupled heat transfer model, including information such as the temperature distribution at different positions of the wall at different times, heat transfer rate, and phase change state of phase change materials. These results provide a key basis for building thermal management.

[0059] For the specific steps to obtain the analysis results of the heat transfer model, reference can be made to step S310 to step S350, which will not be elaborated here.

[0060] Step S400, standardize the data included in the building heat demand information and the data included in the analysis results of the heat transfer model to make their ranges unified.

[0061] Standardization processing: A data processing method whose purpose is to transform data with different characteristics, different dimensions, and different value ranges into a unified scale or range through specific mathematical transformations, so as to facilitate subsequent comparison, analysis, and calculation. In this step, it is to eliminate the interference of data differences on subsequent weighted calculations and thermal condition level analyses.

[0062] Step S500, set weights for each input data according to different building types, usage scenarios, and seasons, and calculate the total sum of input data using the weighted average method. Each input data refers to various data related to building thermal management in the intelligent control method for energy-saving building walls, including environmental data, user behavior data, real-time monitoring data of heat transfer related parameters, data included in the building heat demand information, and data included in the analysis results of the heat transfer model.

[0063] Building type: Classification based on the usage function, structural characteristics, etc. of the building, such as residential buildings, office buildings, industrial factories, etc. The heat demands and heat transfer characteristics of different types of buildings are different. For example, residential buildings mainly meet the comfort needs of residents' lives, while industrial factories may pay more attention to the suitable environment for equipment operation, which makes the emphasis on various data in heat management vary.

[0064] Usage scenario: Refers to the specific situation during the actual use of the building, including personnel density, equipment usage frequency, indoor activity types, etc. For example, during a meeting, a conference room is densely populated and equipment is used frequently, so its heat demand is higher than during normal office hours; the heat demands of a shopping mall are also very different during business hours and non-business hours. These scenario differences will affect heat management strategies and thus affect the weight setting of input data.

[0065] Season: Different time periods in a year, such as spring, summer, autumn, and winter. Factors such as environmental temperature and sunshine duration change significantly in different seasons, which have a great impact on the heat demand of buildings and the heat transfer process of walls. For example, in summer it is hot and the building's cooling demand is large; in winter it is cold and the heating demand is high, which makes the importance of various input data for heat management decisions different in different seasons.

[0066] Weight: A value used to measure the relative importance of each input data when calculating the sum of input data. According to the differences in building type, usage scenario, and season, the influence degree of each data on heat management decisions is different, and the weight will be adjusted accordingly. For example, in summer, the environmental temperature has a greater impact on the cooling demand, and its weight may be relatively high.

[0067] Step S600, based on the mapping relationship between the range of the sum of input data and the grade, analyze and determine the grade corresponding to the heat condition.

[0068] Range: A pre-set numerical range of the sum of input data for facilitating the classification of heat conditions. Different ranges represent different heat management situations, and each range corresponds to a heat condition grade.

[0069] Grade: A classification identifier for heat conditions, used to distinguish different states of building heat management. There is a corresponding relationship between the heat condition grade and the range of the sum of input data. By judging the range to which the sum of input data belongs, the heat condition grade can be determined so as to take corresponding control measures subsequently.

[0070] Heat condition: Describes the heat state of a building at a certain moment, including indoor temperature, wall heat transfer situation, etc. The division of heat condition grades helps to quickly understand the heat management needs of the building, so as to formulate targeted control strategies.

[0071] The general process is as follows: Obtain the total input data obtained in step S500, compare it with the preset range, determine which interval the total input data falls into, and then determine the corresponding thermal condition level.

[0072] Step S700: Analyze and determine the overall control plan according to the mapping relationship between the thermal condition level and the overall control plan, and execute the determined overall control plan.

[0073] Thermal condition level: A classification identifier used to characterize the current thermal state of a building, determined according to the range of the total input data, which reflects the degree of demand for building thermal management. Different levels correspond to different thermal management strategies.

[0074] Overall control plan: A series of control measures formulated for different thermal condition levels, including a phase change material control plan, a lattice structure control plan, and a heat exchange fluid control plan, aiming to achieve efficient thermal management by adjusting the relevant parameters of the building wall to meet the requirements of indoor comfort and energy conservation.

[0075] Phase change material control plan: Mainly focuses on adjusting the phase change material filled in the lattice structure of the wall. The phase change material can absorb or release a large amount of latent heat during the solid-liquid phase change process, which has a significant impact on the thermal performance of the wall. This plan selects and replaces phase change materials with different melting points according to the season change, external temperature change, and heat demand of the building. For example, in summer when the temperature is high, a phase change material with a low melting point is selected to quickly absorb heat when the temperature rises and reduce the indoor temperature; in winter, it is replaced with a phase change material with a high melting point to facilitate heat storage and slow release to maintain the indoor warmth.

[0076] Lattice structure control plan: Focuses on optimizing the 3D-printed lattice structure. The lattice structure provides a stable filling space for the phase change material, and its geometric shape and structural parameters directly affect the heat contact area between the phase change material and the outside world and the heat conduction efficiency. The control plan may include changing the pore size of the lattice, the arrangement of lattice units, etc., to enhance the heat conduction effect and improve the heat storage and heat dissipation capabilities of the wall.

[0077] Heat exchange fluid control plan: Precisely controls the fluid in the outer lattice heat exchange fluid circulation channel. The fluid circulates in the channel and plays an important role in heat transfer. This plan adjusts parameters such as the circulation rate and temperature of the fluid to achieve efficient heat transfer and dissipation of the wall heat. In summer, increasing the fluid circulation rate and reducing the fluid temperature can quickly take away the excess heat absorbed by the wall; in winter, appropriately reducing the circulation rate and increasing the fluid temperature can assist the phase change material to release heat and keep the indoor temperature stable.

[0078] The general process is as follows: When the system obtains the thermal condition level determined by step S600, it will immediately search in the database for the corresponding overall regulation plan. For example, if the thermal condition level indicates that the current building is in a high-temperature and high-load state, the overall regulation plan found by the system may be: in terms of phase change material regulation, select low-melting-point hydrated salt phase change materials for replacement; in terms of lattice structure regulation, adjust the parameters of the 3D printing equipment to increase the pores of the lattice to increase the thermal contact area; in terms of heat exchange fluid regulation, start the circulation pump, increase the fluid circulation rate to the maximum value, and reduce the fluid temperature. After determining the regulation plan, the system will automatically send control instructions to the relevant execution devices. For the replacement of phase change materials, control the dedicated filling and discharging device to discharge the original phase change materials and inject new phase change materials that meet the requirements; for the adjustment of the lattice structure, control the 3D printing equipment (if it has a real-time adjustment function) or manufacture the lattice structure according to the new parameters in the subsequent production process; in terms of heat exchange fluid regulation, adjust the fluid circulation rate by controlling the motor speed of the circulation pump, and adjust the fluid temperature using heating or cooling devices. By implementing these regulation measures, the thermal performance of the building wall is optimized, enabling the building to achieve good thermal management effects under different thermal conditions and meeting the dual requirements of indoor comfort and energy conservation.

[0079] The steps for constructing the heat demand prediction model are as follows:

[0080] Step S201, collect environmental data and user behavior data, and perform data preprocessing.

[0081] Step S202, build a Transformer model based on spatio-temporal decoupling, randomly initialize the parameters of the model, and determine the hyperparameters of the model.

[0082] Transformer model based on spatio-temporal decoupling: The Transformer model is a deep learning model based on the attention mechanism and is widely used in fields such as natural language processing. In the present invention, the Transformer model based on spatio-temporal decoupling is improved for the building heat demand prediction scenario. Spatio-temporal decoupling means separating the processing of time and space features, which can more effectively capture the change laws of environmental data and user behavior data in the time series, as well as the correlation of data at different spatial positions (such as different areas of the building), thereby improving the accuracy of building heat demand prediction.

[0083] Model parameters: are the values that the model needs to learn and adjust during the training process, such as weights and biases in a neural network. These parameters determine the function and performance of the model. By continuously optimizing the parameters, the model can better fit the training data and accurately predict the building heat demand.

[0084] Hyperparameters: These are parameters set in advance before model training and cannot be directly learned through training. Hyperparameters affect the model's structure, training process, and performance. For example, the number of model layers, the number of neurons in the hidden layer, the learning rate, etc. are all hyperparameters. Reasonable setting of hyperparameters is crucial for the training effect of the model.

[0085] The general process is as follows: First, select a suitable deep learning framework (such as TensorFlow or PyTorch) as the basis for building the model. Then, according to the design principle of the Transformer model based on spatio-temporal decoupling, write code in the framework to construct the model structure. During the construction process, define the input layer, hidden layer, and output layer of the model. The hidden layer adopts the structural unit of the Transformer and incorporates the processing logic of spatio-temporal decoupling. After the model is built, use the functions provided by the framework to randomly initialize the model parameters so that the model has an initial state at the beginning of training. Finally, based on the understanding and experience of the building heat demand prediction task, set the hyperparameters of the model, such as determining the number of model layers, the number of neurons in the hidden layer, the learning rate, etc., to prepare for subsequent model training.

[0086] Step S203: Divide the preprocessed data into a training set and a validation set according to a preset ratio.

[0087] Preset ratio: This is a ratio preset before model training for dividing the training set and the validation set. Common preset ratios are 70:30, 80:20, etc. It determines the distribution relationship between the amount of data used for training the model and the amount of data used for evaluating the model's performance.

[0088] Training set: This is a subset of data divided from the preprocessed data for training the model. During the training process, the model continuously adjusts its own parameters by learning the training set data to improve the accuracy of predicting the building heat demand. Validation set: This is a subset of data used to evaluate the training effect of the model. During the model training process, the validation set is used to calculate the loss value of the model at regular training intervals. By observing the change in the validation set loss, it is judged whether the model is overfitting or has achieved a good training effect.

[0089] Step S204: Input the training set data into the Transformer model based on spatio-temporal decoupling, calculate the prediction result through forward propagation, use the weighted mean square error loss function to calculate the difference between the prediction result and the true heat demand value, and perform the Dropout operation during each forward propagation to obtain the prediction results under different parameter samples.

[0090] Among them, forward propagation: In a deep learning model, the process in which data enters from the input layer, successively undergoes calculations and transformations in each hidden layer, and finally reaches the output layer to generate a prediction result. In this process, the model performs a series of mathematical operations on the input data according to the current parameters, such as matrix multiplication, activation function operations, etc.

[0091] Prediction result: The result output by the model after forward propagation. In this scenario, it refers to the predicted value of the building's heat demand by the Transformer model based on spatio-temporal decoupling. This value will be compared with the true heat demand value.

[0092] Weighted mean squared error loss function: A function used to measure the difference between the model's prediction result and the true value. The mean squared error is the average of the squares of the differences between the predicted value and the true value, while the weighted mean squared error assigns different weights to different data points when calculating the mean squared error to highlight the importance of certain data. In building heat demand prediction, the model's parameters are optimized by minimizing this loss function.

[0093] Dropout operation: A technique to prevent model overfitting. During model training, Dropout randomly "drops out" (i.e., sets the neuron output to 0) a part of the neurons in the hidden layer, so that the model cannot overly rely on certain specific neuron connections, thereby enhancing the model's generalization ability and reducing the overfitting phenomenon.

[0094] The general process is as follows: Assume that the Transformer model based on spatio-temporal decoupling has been built and there is a training set containing 100 data points. First, a batch of data (such as 10 data points as a batch) is taken from the training set and input into the model. The data enters the model from the input layer and successively undergoes calculations in the hidden layer according to the forward propagation calculation rules. For example, the neurons in the hidden layer perform a weighted sum on the input data and are transformed through an activation function, and finally a prediction result is obtained at the output layer. This prediction result is an estimated value of the building's heat demand.

[0095] Step S205, based on the result of the loss function, calculate the gradient through the backpropagation algorithm and update the model parameters.

[0096] Loss function: In step S204, the weighted mean squared error loss function is used to measure the difference between the model's prediction result and the true heat demand value. The value of the loss function reflects the accuracy of the current model prediction. The smaller the value, the closer the model's prediction result is to the true value.

[0097] Backpropagation algorithm: It is the core algorithm used to optimize model parameters in the training of deep learning models. Based on the chain rule of differentiation, starting from the output layer, it backpropagates the gradients of the loss function with respect to the parameters of each layer, calculates the gradient value of the loss function with respect to each parameter, and these gradient values indicate how much a small change in the parameter will affect the loss function.

[0098] Gradient: Mathematically, a gradient is a vector whose direction points to the direction of the fastest change of a function. In model training, the gradient of the loss function with respect to the model parameters indicates the direction in which the parameters should be adjusted to minimize the value of the loss function to the greatest extent.

[0099] The general process is as follows: 1. Receive the loss value: Obtain the loss value calculated by the weighted mean square error loss function from step S204, and this value reflects the deviation between the model prediction and the true heat demand. 2. Start backpropagation: Start backpropagation from the output layer of the model. Since the loss function is calculated based on the prediction results of the output layer, the impact of each parameter on the loss can be directly calculated from here. 3. Calculate the gradient of the output layer: Use the chain rule of differentiation to calculate the gradient of the loss function with respect to the input of the output layer neurons, that is, the gradient of the output layer. 4. Propagate the gradient to the hidden layer: Backpropagate the output layer gradient to the hidden layer and calculate the gradient of this layer in combination with the hidden layer activation function. 5. Process multiple hidden layers: If there are multiple hidden layers, sequentially backpropagate the gradient to each layer until the input layer. 6. Calculate the parameter gradients: Calculate the gradients of the loss function with respect to all the model parameters (weights and biases) based on the gradients of the neurons in each layer. 7. Set the learning rate: The learning rate controls the step size of parameter update and needs to be determined in advance according to experience or experiments. If the learning rate is too large, the optimal solution may be missed; if it is too small, the training will be slow. 8. Update the model parameters: Update the model parameters using the gradient descent algorithm according to the parameter gradients and the learning rate. Repeat steps S204 and S205 to continuously optimize the model parameters and improve the accuracy of heat demand prediction.

[0100] In step S206, after each training cycle is completed, calculate the validation set loss. If the change in the validation set loss for a continuously preset number of cycles is less than the set threshold, it is considered that the training is completed.

[0101] Among them, the training cycle (Epoch): In the process of model training, the process of completely inputting the entire training set data into the model and completing one forward propagation and backpropagation to update the parameters is called a training cycle. After each training cycle is completed, the model has completed one round of learning on the training set data.

[0102] Validation set loss: Use the validation set data to input the trained model, and obtain the loss value by calculating the difference between the prediction result and the true value of the validation set. Usually, the weighted mean square error loss function is also used for calculation. The validation set loss is used to evaluate the performance of the model on unseen data and reflects the generalization ability of the model.

[0103] Preset number of epochs: A predefined integer used to measure the change in the validation set loss over a consecutive number of training epochs, thereby determining whether the model has converged. For example, if it is preset to 5, it means observing the change in the validation set loss over 5 consecutive training epochs.

[0104] Set threshold: A small value determined in advance, used to judge whether the change in the validation set loss is small enough. If the change in the validation set loss is less than this threshold for a consecutive preset number of epochs, it is considered that the model training has achieved a good effect and the training can be stopped. The setting of this threshold needs to be determined according to specific problems and experience, and generally takes a small value, such as 0.001.

[0105] The steps for constructing a multi-physics field coupled heat transfer model are as follows:

[0106] Step S301: Collect and obtain multi-source data and perform data preprocessing.

[0107] Among them, multi-source data: refers to various different types of data related to the heat transfer process of building walls. These data have a wide range of sources, including environmental data (such as external temperature, humidity, solar radiation intensity, etc.), building structure data (such as wall materials, thickness, lattice structure parameters, etc.), phase change material property data (such as latent heat of phase change, melting point, thermal conductivity, etc.), and heat transfer related monitoring data (such as temperature and heat flux density at different positions of the wall). These data affect the heat transfer process of the wall from different aspects and are the basis for constructing a multi-physics field coupled heat transfer model.

[0108] Data preprocessing: Perform a series of processing operations on the collected original multi-source data, aiming to improve the data quality and make it more suitable for model training and analysis. It mainly includes data cleaning (removing outliers and duplicate values), data filling (filling in missing data), data standardization (unifying data with different dimensions and value ranges into a specific interval, such as [0, 1]), etc., to ensure the accuracy, integrity, and consistency of the data.

[0109] Step S302: Build a model architecture at three levels: microscopic, mesoscopic, and macroscopic.

[0110] Microscopic level model architecture: Focus on the microscopic structure and physical phenomena of materials, used to describe the heat transfer mechanism at the atomic and molecular scales, such as the influence of the thermal motion of internal molecules and lattice vibrations in phase change materials on heat transfer, and explain the microscopic process of heat transfer from a fundamental level.

[0111] Mesoscopic-level model architecture: It lies between the microscopic and macroscopic levels and mainly studies the relationship between the mesoscopic structural characteristics of materials and heat transfer. For example, the distribution of phase change materials in the lattice structure, the influence of micro-pores or defects on heat conduction, etc. It is a bridge connecting microscopic and macroscopic heat transfer phenomena.

[0112] Macroscopic-level model architecture: It describes the heat transfer behavior of building walls as a whole, considering macroscopic factors such as the overall geometry of the wall, boundary conditions, and heat exchange with the external environment. For example, the overall temperature distribution of the wall, the inflow and outflow of heat, etc., reflecting the performance of heat transfer at the macroscopic scale.

[0113] Take the construction of a heat transfer model for a lattice structure wall containing phase change materials as an example.

[0114] At the microscopic level, based on material science principles, determine the thermal motion model of phase change material molecules. For example, assume that the intermolecular forces and motions follow a certain classical mechanics model to describe how heat is transferred between phase change material molecules at the microscopic scale and the influence of molecular state changes during the phase change process on heat transfer.

[0115] At the mesoscopic level, according to the characteristics of 3D printed lattice structures, analyze the filling distribution of phase change materials inside the lattice. For example, consider the pore size, shape of the lattice, and the contact situation between the phase change material and the lattice wall, construct the corresponding heat conduction model, and simulate the heat transfer path and efficiency through the lattice structure and phase change materials at the mesoscopic scale.

[0116] At the macroscopic level, regard the entire wall as a whole, considering the size, position, orientation of the wall, and the heat exchange boundary conditions with the indoor and outdoor environments. For example, set the convective heat transfer coefficient between the outer side of the wall and the external environment, the heat exchange method between the inner side and the indoor air, etc., establish a macroscopic heat transfer equation to describe the overall heat income and expenditure and temperature change trend of the wall. By integrating the model architectures of these three levels, form a complete model framework that can describe the wall heat transfer process from different scales, laying a foundation for subsequent model calculations and analyses.

[0117] Step S303, according to the corresponding relationship between the building thermal management scenario and the physical field coupling method and intensity setting scheme, analyze and determine the physical field coupling method and intensity of the current building thermal management scenario as the starting state of model training.

[0118] Building thermal management scenario: It refers to the heat-related situations of buildings under different usage conditions and environmental conditions, covering various factors such as building types (residential, office buildings, shopping malls, etc.), usage time (day, night, weekdays, holidays), seasons (spring, summer, autumn, winter), and indoor and outdoor environmental parameters (temperature, humidity, light, etc.). The heat demands and heat transfer processes of buildings vary significantly under different scenarios.

[0119] Physical field coupling mode: In the process of building heat transfer, there are various interactions of physical fields, such as heat conduction, convection, radiation, and the phase change process of phase change materials. The physical field coupling mode describes the specific forms of the interconnection and influence between these physical fields. For example, the coupling of heat conduction and convection means that while heat is conducted in a solid material, heat exchange occurs through convection with the surrounding fluid; the coupling of heat conduction and radiation involves heat transfer through radiation on the surface of an object while heat conduction occurs inside the material.

[0120] Physical field coupling strength: An index to measure the degree of interaction between different physical fields. For example, in the coupling of heat conduction and convection, if the convection effect is strong and heat transfer is mainly dominated by convection, the coupling strength is high; conversely, if heat conduction plays a major role, the coupling strength is relatively low. The coupling strength determines the relative importance of each physical field in the heat transfer process.

[0121] Corresponding relationship: The association rule between the building heat management scenario and the physical field coupling mode and strength setting scheme obtained through theoretical research, experimental testing, and summary of practical project experience. It provides a basis for determining the appropriate physical field coupling mode and strength according to the specific building heat management scenario.

[0122] Step S304, incorporate the non-equilibrium heat flux equation into the model heat transfer calculation system.

[0123] Among them, the non-equilibrium heat flux equation is , where, is the heat flux density, is the thermal conductivity, is the temperature gradient, is the pressure gradient, is the mass concentration gradient, is the heat transfer coefficient related to pressure, is the heat transfer coefficient related to mass concentration.

[0124] Among them, the non-equilibrium heat flux equation: A mathematical expression used to describe the heat transfer law in a non-equilibrium state. In the building heat transfer scenario, the actual situation is often not in an ideal equilibrium state, and this equation can more accurately reflect the heat transfer process. In the given equation, the heat flux density is affected by various factors such as material properties, temperature change rate, pressure gradient, and mass concentration gradient. Compared with the traditional equilibrium heat transfer equation, it takes into account more dynamic factors' effects on heat flux.

[0125] Model heat transfer calculation system: It is a collection of a series of rules, formulas, and algorithms used to calculate the heat transfer process in a multi-physical-field coupled heat transfer model. It integrates the basic equations of different physical fields (such as heat conduction, convection, radiation, etc.), and conducts comprehensive calculations according to the coupling method and intensity of the physical fields to simulate the heat transfer behavior of structures such as building walls. Incorporating the non-equilibrium heat flux equation aims to improve this calculation system and enhance the simulation accuracy of the model for complex heat transfer processes.

[0126] The general process is as follows:

[0127] Suppose a multi-physical-field coupled model is constructed to simulate the heat transfer of a new type of phase change material wall. First, researchers obtain a suitable non-equilibrium heat flux equation from professional heat transfer literature, which fully considers the influence of temperature, pressure, and substance concentration changes on heat flux during the phase change process of the phase change material. Then, analyze the established model heat transfer calculation system to determine the position and method of integrating the equation. For example, in the macroscopic-level heat transfer calculation, the traditional equilibrium heat conduction equation was originally used to calculate the overall heat conduction of the wall. Now, integrate the non-equilibrium heat flux equation with the original equations of heat conduction, convection, radiation, etc. For a tiny unit in the wall, when calculating its heat flux density, introduce the parameters in the non-equilibrium heat flux equation. For example, determine the relevant coefficients according to the material properties, and use sensors to monitor data such as the temperature change rate, pressure gradient, and substance concentration gradient in real time, and substitute them into the equation for calculation. In this way, the original calculation system that only considered steady-state heat transfer, due to the incorporation of the non-equilibrium heat flux equation, can more accurately simulate the non-equilibrium heat transfer process of the wall caused by external environmental changes and the dynamic characteristics of the phase change material during actual use, thereby enhancing the accuracy and reliability of the model's simulation of the wall's heat transfer.

[0128] Step S305, select a deep neural network with a multi-layer perceptron structure having 3 hidden layers and 100 neurons in each layer, and integrate it into the model using a residual connection fusion strategy. The input layer of the deep neural network receives data on temperature, humidity, and building structure parameters. The hidden layer uses the ReLU function to extract features, and the output layer predicts heat transfer-related parameters.

[0129] Specifically, the deep neural network with a multi-layer perceptron (MLP) structure: a feedforward neural network composed of an input layer, multiple hidden layers, and an output layer. In this step, it has 3 hidden layers with 100 neurons in each layer, and the neurons are connected by weights. It can perform non-linear transformations on the input data and realize the prediction of heat transfer-related parameters by learning patterns in a large amount of data. Neurons in different layers can learn features at different levels of the data, gradually abstracting from the original input features to more advanced and representative features for the final prediction task.

[0130] Residual connection fusion strategy: A connection method adopted to address the problems of vanishing gradients or exploding gradients that may occur during the training of deep neural networks and to improve the network learning efficiency. Residual connections allow the network to directly learn the residuals between the input and output, that is, to make the learning objective of the network become the difference between "output - input" instead of directly learning the output itself. In this step, different layers of the multi-layer perceptron are fused through residual connections, enabling information to flow more efficiently in the network, helping the model converge faster, and improving the model's ability to model complex heat transfer processes.

[0131] ReLU function: That is, the Rectified Linear Unit, which is a commonly used activation function. Using the ReLU function in the hidden layer can introduce non-linearity into the neural network. Since the expressive power of a linear model is limited, through the ReLU function, when the input is greater than 0, the neuron outputs the input value normally; when the input is less than 0, the output is 0, enabling the neural network to learn more complex functional relationships and better extract data features.

[0132] Heat transfer related parameters: Various physical quantities closely related to the heat transfer process of building walls, such as heat flux density, temperature distribution, and changes in the thermal conductivity of materials. These parameters are the targets that need to be accurately predicted by the multi-physics field coupled heat transfer model. Through the analysis and processing of input data by the deep neural network, the predicted results of these parameters are output, providing key information for the model to comprehensively simulate the heat transfer process.

[0133] In step S306, the preprocessed data is input into the model that has been constructed with an architecture, determined the coupling method, incorporated the non-equilibrium equation, and integrated the deep neural network.

[0134] Model with a constructed architecture: The model architecture at the microscopic, mesoscopic, and macroscopic levels completed through step S302, which comprehensively describes the heat transfer mechanism from the atomic and molecular scale, the mesoscopic structure of materials to the overall building wall, providing a basic framework for the model.

[0135] Model with a determined coupling method: According to step S303, based on the corresponding relationship between the building thermal management scenario and the physical field coupling method and intensity setting scheme, determine the coupling method between physical fields such as heat conduction, convection, and radiation in the current scenario, such as how heat conduction and convection interact with each other, and the association form between heat conduction and radiation, enabling the model to reflect the physical field relationship in the actual heat transfer process.

[0136] Model incorporated with the non-equilibrium equation: In step S304, the non-equilibrium heat flux equation is incorporated into the model's heat transfer calculation system. This equation considers the effects of factors such as material properties, temperature change rate, pressure gradient, and substance concentration gradient on heat flux density, improving the simulation accuracy of the model for complex non-equilibrium heat transfer processes.

[0137] Model integrating a deep neural network: In step S305, a deep neural network with a multi-layer perceptron structure having 3 hidden layers with 100 neurons in each layer is integrated into the model using a residual connection fusion strategy. This deep neural network inputs data such as temperature, humidity, and building structure parameters, extracts features through the ReLU function in the hidden layers, and predicts heat transfer-related parameters in the output layer, enhancing the model's prediction and analysis capabilities.

[0138] In step S307, during training, the mean squared error loss function is used to calculate the deviation between the model prediction value and the true value. The stochastic gradient descent method is employed to iteratively update the model parameters at a learning rate of 0.01, causing the value of the loss function to decrease.

[0139] Mean squared error loss function: A commonly used function for measuring the difference between the model prediction value and the true value. It reflects the degree of deviation by calculating the average of the squares of the differences between the prediction value and the true value. In the training of the multi-physics field coupled heat transfer model, it can quantify the accuracy of the model's prediction of heat transfer-related parameters (such as heat flux density and temperature distribution). The smaller the loss value, the closer the model prediction is to the true value.

[0140] Model prediction value: After step S306, when the preprocessed data is input into the constructed multi-physics field coupled heat transfer model, the output result of the model for heat transfer-related parameters, such as the predicted value of the heat flux density of a wall at a certain moment.

[0141] True value: The accurate numerical values of the heat transfer-related parameters of the building wall obtained through actual measurement, experiment, or authoritative data sources, such as the temperature and heat flux density data obtained by installing high-precision sensors on the wall for real-time monitoring, which are used as the reference standard for model training.

[0142] Stochastic gradient descent method: An optimization algorithm used to update parameters during model training to minimize the value of the loss function. It randomly selects a small batch of data samples from the training dataset each time to calculate the gradient, and then updates the model parameters based on the gradient and the set learning rate. It has high computational efficiency and can quickly converge to better parameter values on large-scale datasets.

[0143] Learning rate: A hyperparameter in the stochastic gradient descent method that controls the step size of each parameter update. In this step, it is set to 0.01, that is, each time the parameters are updated, the model parameters move a distance of 0.01 times the gradient value in the opposite direction of the gradient. If the learning rate is too large, the model training may skip the optimal solution and fail to converge; if it is too small, the training is slow, consuming a large amount of time and computational resources.

[0144] The general process and an example are as follows: Taking the training of a heat transfer simulation model for the wall of a commercial building as an example. First, obtain a batch of real data on the heat transfer of the building wall from the data acquisition system, covering parameter values such as temperature and heat flux density at different positions. At the same time, run the constructed model, input the corresponding preprocessed data, and obtain the predicted values of the model for these heat transfer parameters. Then, use the mean square error loss function to calculate the deviation between the predicted values and the real values. Subsequently, use the stochastic gradient descent method to update the model parameters at a learning rate of 0.01. By continuously repeating this process, each time randomly select a small batch of data from the training dataset to calculate the loss and update the parameters, gradually reducing the value of the loss function, making the model's predicted values closer to the real values, and improving the accuracy of the model's simulation of the heat transfer process of the commercial building wall.

[0145] Step S308: Every time 10 parameter updates are completed, based on the model prediction results and real data under the latest parameters, calculate the influence weights of the coupling strengths of each physical field on the loss function. According to the weights, adjust the coupling strength coefficient by 10% to reduce the value of the loss function.

[0146] Parameter update: During the model training process, an operation of adjusting the model parameters (such as weights, biases, etc.) according to the gradients calculated by the loss function based on an optimization algorithm (such as the stochastic gradient descent method in step S307). Each parameter update aims to make the model prediction results closer to the real values and reduce the value of the loss function.

[0147] Model prediction results: The predicted values of heat transfer-related parameters (such as heat flux density, temperature distribution) output by the model after operating on the input data (such as environmental data related to building thermal management, building structure data, etc.) based on the current parameters.

[0148] Real data: Data obtained through actual measurement, experiment, or reliable historical data records, which accurately reflects the actual situation of the heat transfer of the building wall and is used to evaluate the accuracy of the model prediction.

[0149] Physical field coupling strength: In a multi-physical field coupled heat transfer model, an index describing the degree of interaction between different physical fields (such as heat conduction, convection, radiation, and phase change processes). In different building thermal management scenarios, the coupling strengths of each physical field are different, and they have an important impact on the accuracy of the model prediction.

[0150] Influence weight: A numerical value representing the degree of influence of the coupling strength of each physical field on the loss function. The larger the weight, the more significant the change in the value of the loss function caused by the change in the coupling strength of this physical field.

[0151] Coupling strength coefficient: A parameter used to quantify the coupling strength of physical fields. By adjusting this coefficient, the coupling strength between physical fields can be changed. In the model, the coupling strength coefficient is associated with the relevant equations or calculation processes of each physical field. Changing the coefficient value can adjust the degree of interaction between physical fields.

[0152] Step S309: According to the dynamic change results of the scenario within each time interval range, change the physical field coupling strength through a preset coupling strength adjustment function.

[0153] Time interval range: In the process of model training and application, a time interval set to capture the changes of the building thermal management scenario over time. For example, every 1 hour, 3 hours, etc. are used as a time interval range, and scenario data is collected and analyzed within this range.

[0154] Dynamic change results of the scenario: Refers to the changes of various factors in the building thermal management scenario within each time interval range. It includes environmental factors (such as changes in temperature, humidity, and sunlight intensity), user behavior factors (such as equipment usage and personnel activity changes), and building self-state factors (such as the rise or fall of wall temperature over time). These changes will affect the coupling situation of physical fields.

[0155] Preset coupling strength adjustment function: A mathematical function predefined when constructing a multi-physical-field coupled heat transfer model. This function calculates the corresponding physical field coupling strength adjustment value according to the parameters in the dynamic change results of the scenario (such as temperature change amount, equipment power change, etc.) to meet the heat transfer simulation requirements under different scenarios.

[0156] Physical field coupling strength: A quantitative index describing the degree of interaction between multi-physical fields (such as heat conduction, convection, radiation, and phase change processes). Under different scenarios, the contributions of each physical field to heat transfer are different, and the coupling strength will change accordingly, thereby affecting the simulation accuracy of the model for the heat transfer process.

[0157] Step S30A: Continuously monitor the mean square error. When the change value of the mean square error is less than the preset change value in consecutive preset rounds of training, it is determined that the training is completed.

[0158] Consecutive preset rounds of training: The preset number of consecutive training times. Each round of training includes a series of operations such as inputting a batch of data into the model, performing forward propagation to calculate the prediction result, calculating the loss function (here it is the mean square error), and updating the model parameters through backpropagation. In this step, a consecutive number of rounds of training is used as the observation period to determine whether the model training has reached a stable state.

[0159] Mean Square Error Variation Value: The difference between the mean square error obtained in the later round of training and the mean square error in the previous round during two adjacent rounds of training. This value reflects the change in the prediction accuracy of the model after one round of training. If the variation value is small, it indicates that the improvement degree of the model in this round of training is limited.

[0160] Preset Variation Value: A threshold set based on experience and expectations for the model performance before the start of model training. When the mean square error variation values in consecutive preset rounds of training are all less than this preset variation value, it is considered that the model has converged and achieved a good training effect, and the training can be stopped.

[0161] The preset coupling strength adjustment function is as follows:

[0162] ;

[0163] At time the adjusted coupling strength coefficient between the physical field and the physical field ;

[0164] is the number of scenario factors affecting the coupling strength;

[0165] is the weight of the k-th scenario factor, reflecting the relative importance of this factor in adjusting the coupling strength;

[0166] is the k-th scenario factor at time the influence function of which is used to quantify the influence degree of the change of this factor on the coupling strength;

[0167] is the specific state or value of the k-th scenario factor at time ;

[0168] is the basic coupling strength coefficient between the physical field and the physical field ;

[0169] The multi-physical-field coupled heat transfer model calculates the heat transfer process of the fine elements of the wall structure based on the finite element analysis method. By solving the equations of multi-physical-field coupling, the analysis results of the heat transfer model include:

[0170] Step S310, based on the preset fine element division basis, make a fine element division of the wall structure based on the finite element analysis method.

[0171] Basis for preset fine unit division: It is determined by the model developer according to the building wall design documents, heat transfer principles, and the construction experience of previous similar models. The building design documents contain information such as the material information and geometric dimensions of the wall, which are important bases for formulating the division basis. At the same time, the developer refers to the research results on heat transfer characteristics under different materials and boundary conditions in heat transfer theory, and adjusts and improves the division basis in combination with the actual engineering requirements. Finally, the determined fine unit division basis is stored in the form of a text file or a configuration file and read during the model construction process.

[0172] Finite element analysis method: It is implemented by using professional finite element analysis software or calling finite element analysis-related libraries in programming languages. For example, commercial finite element analysis software such as ANSYS and ABAQUS provides rich functions and tools, facilitating users to perform operations such as model construction, unit division, and solution settings. In development based on programming languages (such as Python), finite element analysis libraries like FEniCS can be used to apply the finite element analysis method by writing code. These software and libraries can be downloaded from the official websites, and some commercial software requires purchasing licenses.

[0173] Step S320: For each fine unit, construct a heat transfer equation based on the physical laws of heat conduction, convection, and radiation.

[0174] Heat conduction: The way heat transfers from the part with a higher temperature of an object along the object to the part with a lower temperature, which is the main way of heat transfer in solids. It follows Fourier's law, that is, the heat passing through a certain cross-section per unit time is proportional to the temperature gradient at that cross-section and is related to the thermal conductivity of the material.

[0175] Convection: The heat transfer process caused by the relative displacement of different parts with different temperatures in a fluid (gas or liquid). In the heat transfer of building walls, it mainly involves the convective heat transfer between the wall surface and the surrounding air. Convective heat transfer follows Newton's law of cooling, that is, the convective heat transfer quantity is related to the temperature difference between the fluid and the solid surface, the convective heat transfer coefficient, and the heat transfer area.

[0176] Radiation: The way an object transfers energy through electromagnetic waves. Thermal radiation does not require any medium and can also occur in a vacuum. The thermal radiation ability of an object is related to factors such as the temperature of the object and the surface emissivity, and follows the Stefan-Boltzmann law, that is, the radiative heat flux density per unit surface area of the object is proportional to the fourth power of the absolute temperature of the object.

[0177] Heat transfer equation: A mathematical equation that comprehensively considers three heat transfer modes, namely heat conduction, convection, and radiation, and is used to describe the heat transfer law within fine units. By constructing the heat transfer equation, the heat transfer process of each fine unit under different boundary conditions and material properties can be quantitatively analyzed, providing a basis for subsequent multi-physics field coupling analysis.

[0178] Step S330: After constructing the heat transfer equation for each tiny unit, based on the mutual influence between physical fields, construct an inter-field coupling term and add the coupling term to the unit equation.

[0179] Physical field: Here, it mainly refers to three physical fields of heat conduction, convection, and radiation, which respectively represent different heat transfer modes.

[0180] Inter-field coupling term: A mathematical expression that reflects the mutual influence between physical fields. Since in the actual heat transfer process, heat conduction, convection, and radiation do not exist in isolation but interact and influence each other, the inter-field coupling term is used to quantify this mutual influence.

[0181] Unit equation: That is, the heat transfer equation constructed in step S320. After adding the inter-field coupling term, it can more accurately describe the heat transfer process of the tiny unit under the interaction of multiple physical fields.

[0182] Step S340: Assemble the unit equation after adding the coupling term to form a multi-physics field coupling equation set for the entire wall structure.

[0183] Unit equation after adding the coupling term: Refers to the equation obtained after step S330, which is based on the heat transfer equation constructed for each tiny unit based on heat conduction, convection, and radiation and adding the inter-field coupling term. These equations describe the heat transfer situation of each tiny unit under the interaction of multiple physical fields.

[0184] Assembly: The process of combining the equations of each tiny unit according to certain rules and sequences. In finite element analysis, this involves considering factors such as the connection relationship and boundary conditions between units, so that the assembled equation set can reflect the heat transfer characteristics of the entire wall structure.

[0185] Multi-physics field coupling equation set: Assembled from the equations of multiple tiny units, comprehensively considering heat conduction, convection, radiation, and their coupling effects, and used to comprehensively describe the heat transfer process of the entire wall structure under the joint influence of multiple physical fields. By solving this equation set, heat transfer-related parameters such as the temperature distribution and heat flux density of each part of the wall can be obtained.

[0186] The general process is described as follows: 1. Read equations: Obtain the heat transfer equations of the micro-elements with added coupling terms from the storage location. These equations describe the heat transfer under the interaction of multiple physical fields within the elements. 2. Read structural information: Read the wall structure mesh file to clarify the connection relationships and boundary information between elements, such as which elements are connected through nodes. 3. Equation assembly: Integrate the element equations according to the finite element assembly rules. For elements sharing nodes, combine the relevant terms involving the same variables (such as temperature variables) at that node. 4. Form a system of equations: After completing the assembly of all element equations, organize them into a mathematical form suitable for solution, such as the commonly used KX = F matrix form in finite element analysis.

[0187] In step S350, use a finite element solution method based on the preconditioned conjugate gradient method to iteratively solve this system of equations to obtain the analysis results of the heat transfer model.

[0188] Preconditioned conjugate gradient method: An iterative algorithm for solving linear systems of equations. In a multi-physical-field coupled heat transfer model, since the assembled multi-physical-field coupled system of equations is usually a large-scale linear system of equations, direct solution has low efficiency. The preconditioned conjugate gradient method improves the condition number of the original system of equations by introducing a preconditioning matrix, thereby accelerating the iterative convergence speed and improving the solution efficiency. This method calculates the conjugate direction and search step size in each iteration to gradually approach the exact solution of the system of equations.

[0189] Finite element solution method: A method based on finite element analysis theory for discretizing continuous physical problems into a finite number of elements for solution. In a heat transfer model, by dividing the wall structure into micro-elements, constructing element equations and assembling them into a system of equations, the finite element solution method is used to obtain the numerical solution of the equations, and then the heat transfer related parameters of each part of the wall, such as temperature distribution and heat flux density, are obtained.

[0190] Iterative solution: For complex systems of equations, the exact solution cannot be obtained through direct calculation, but an iterative approach is adopted to gradually approach the exact solution. In each iteration, the solution vector is updated according to the result of the previous iteration through a specific algorithm until a certain convergence condition is met, and it is considered that a sufficiently accurate solution has been obtained.

[0191] Analysis results of the heat transfer model: The various parameters and information about the heat transfer process of the wall obtained by solving the multi-physical-field coupled system of equations, such as the temperature values at different positions of the wall, the heat transfer path and density distribution of the heat flux in the wall, etc. These results can help analyze the thermal performance of the wall and provide a basis for building thermal management.

[0192] Set weights for each input data according to different building types, usage scenarios, and seasons, including:

[0193] Step S510, aggregate the preprocessed multi-source data, and extract the basic features directly related to building type, usage scenario, and season from the multi-source data.

[0194] Feature extraction: Carefully analyze the multi-source data and identify the data parts directly related to building type, usage scenario, and season. For example, building type may be associated with information such as building structure form, wall materials, etc.; usage scenario may involve data such as equipment operating status, personnel activity frequency, etc.; season is related to data such as outdoor temperature, sunshine duration, etc. Precisely extract these directly related data from the aggregated data as basic features.

[0195] Step S520, use a convolutional neural network to process the season-related data to obtain a season feature vector, and at the same time use a long short-term memory network to analyze the usage scenario data to capture the features of the usage scenario and obtain a scenario feature vector.

[0196] Among them, Convolutional Neural Network (CNN): A deep learning neural network designed specifically for processing data with grid structures (such as images, time series data). It automatically extracts features from the input data through components such as convolutional layers, pooling layers, and fully connected layers. When processing season-related data, the convolutional neural network can capture local patterns and features in the data, such as the periodic change features of data such as temperature and sunshine duration caused by seasonal changes.

[0197] Season-related data: Various data closely related to seasons, such as environmental data such as average temperature, humidity, sunshine duration, precipitation, etc. in different seasons, as well as energy consumption pattern data related to seasons. These data reflect the impact of seasonal changes on the building environment and usage.

[0198] Season feature vector: After being processed by a convolutional neural network, it is a numerical vector representation converted from season-related data. This vector condenses the key feature information in the season data, facilitating subsequent fusion and analysis with other features. Each element in the vector represents different aspects of season features, and its numerical size reflects the strength of the feature.

[0199] Long Short-Term Memory (LSTM): A special type of Recurrent Neural Network (RNN) that can effectively handle long-term dependence problems in time series data. When analyzing usage scenario data, LSTM can remember past information and capture long-term features and trends in the usage scenario based on the current input and past memories, such as the usage patterns of building equipment and the activity rules of personnel over time in different time periods.

[0200] Usage scenario data: Describes the data of a building under different usage conditions, including the operating status of equipment in the building (such as the start time and power consumption of equipment like air conditioners, lighting, elevators, etc.), the activities of personnel (such as the number of people, activity areas, activity times), and data related to the purpose of building use (such as the passenger flow in a shopping mall, the workload in an office space), etc.

[0201] Scenario feature vector: A numerical vector obtained by analyzing usage scenario data through a long short-term memory network. It summarizes the key features in the usage scenario data, reflects the characteristics and laws of the usage scenario, and can be used for subsequent comprehensive feature analysis and weight generation.

[0202] Step S530, fuse the seasonal feature vector and the scenario feature vector to form a comprehensive feature.

[0203] Comprehensive feature: A new feature formed by combining the seasonal feature vector and the scenario feature vector in a certain way. The comprehensive feature integrates the key information of both seasons and usage scenarios, can more comprehensively describe the complex environment where the building is located, and provides richer and more representative information for subsequent input into the dynamic weight generation model to determine the weights of various input data.

[0204] Step S540, input the comprehensive feature into the dynamic weight generation model based on the attention mechanism neural network, and output the set weights for each input data.

[0205] Attention mechanism neural network: A special neural network architecture that enables the model to automatically focus on the importance of different parts of the input information when processing the input. In this step, it will analyze the importance of each input data (such as building structure parameters, environmental parameters, usage scenario parameters, etc.) for the final result (such as the accuracy of the heat transfer model, the accuracy of energy consumption prediction, etc.) according to the input comprehensive feature, and assign different attention weights to different input data.

[0206] Dynamic weight generation model: A model built based on the attention mechanism neural network, whose function is to generate dynamic weights for each input data according to the input comprehensive feature. These weights are not fixed, but will be dynamically adjusted with the changes of building types, usage scenarios, and seasons to adapt to the differences in the influence degrees of various input data on the results in different situations.

[0207] Set weights for each input data: The result output by the model, which are the weight values assigned to each input data (such as wall material parameters, indoor and outdoor temperature data, equipment operating power data, etc.) in multi-physics field coupled heat transfer models or energy management models, etc. The larger the weight, the more significant the influence of the input data on the model result under the current building type, usage scenario, and season.

[0208] According to the mapping relationship between the range of the sum of input data and the level, analyze and determine that the levels corresponding to the thermal conditions include:

[0209] Step S610: Input the obtained sum of input data, building type, usage scenario, season, and comprehensive features into the trained reinforcement learning model. The reinforcement learning model selects an action according to the learned policy to adjust the current range of the sum of input data.

[0210] Reinforcement learning model: A machine learning model that continuously learns the optimal policy by interacting with the environment and according to the reward signals feedback by the environment. In this step, the reinforcement learning model determines how to adjust the range of the sum of input data by learning the relationship between the sum of input data and the thermal conditions under different building types, usage scenarios, seasons, and comprehensive features, so as to more accurately judge the thermal condition level.

[0211] Action: The decision made by the reinforcement learning model in the current state (i.e., the state composed of the sum of input data, building type, usage scenario, season, and comprehensive features), such as expanding or shrinking the range of the sum of input data, or performing operations such as translating the range. The model selects an action that it believes is most conducive to accurately judging the thermal condition level according to the learned policy.

[0212] Suppose a thermal condition analysis is carried out on an office building.

[0213] First, query and obtain the current sum of input data from the database after data preprocessing. This sum is obtained by summing up various thermal-related data such as indoor and outdoor temperatures, operating power of office equipment, and lighting equipment power. At the same time, read from the building design document that the office building belongs to the commercial office building type, learn from the property management system that the current usage scenario is during the day on a working day, determine the current season as summer through the meteorological data interface, and read the comprehensive feature data generated and saved in step S530 from the local file system. This data integrates the seasonal characteristics of summer and the characteristics of the usage scenario of the office building during the day on a working day.

[0214] Then, load the pre-trained reinforcement learning model stored in a file (assuming PyTorch is used). Provide the sum of the obtained input data, building type, usage scenario, season, and comprehensive features as input information to the reinforcement learning model. The reinforcement learning model analyzes the state represented by the current input information according to the policy learned from previous training. For example, based on past experience, the model knows that during the day on weekdays in summer, there is frequent personnel activity and a lot of equipment operation in office buildings, resulting in a relatively large heat load and often a relatively high sum of input data. Based on this, the model selects an action, such as expanding the entire range of the current sum of input data upward by a certain proportion to better adapt to the characteristics of this heat condition and prepare for accurately judging the heat condition level later.

[0215] Step S620: Determine the corresponding heat condition level as the preliminary judgment result according to the adjusted range.

[0216] Adjusted range: In step S610, the new range obtained after the reinforcement learning model expands, shrinks, or translates the original range of the sum of input data according to the input information and the learned policy. This range better conforms to the actual heat condition under the current building type, usage scenario, and season.

[0217] Heat condition level: A series of pre-set classification levels for building heat conditions for the convenience of evaluating and managing building heat conditions. For example, the heat condition level can be divided into comfortable, relatively comfortable, uncomfortable, etc. Each level corresponds to a certain range of indoor and outdoor environmental parameters, energy consumption levels, and human thermal sensation degrees, etc. Different heat condition levels reflect the quality of the building heat environment and its impact on energy utilization efficiency.

[0218] Preliminary judgment result: The corresponding heat condition level found in the pre-established mapping relationship between the range and the heat condition level according to the adjusted range. This result is obtained based on the preliminary analysis of the model and will be further optimized and confirmed through steps such as defuzzification and fuzzy reasoning later.

[0219] Step S630: Perform defuzzification processing on the preliminary judgment result, the sum of input data, environmental data, and user behavior data together.

[0220] Defuzzification processing: The process of converting precise numerical values or clear categories (such as the preliminary judgment result, specific environmental data values) into fuzzy sets. In a fuzzy set, the membership degree of an element to the set is not simply "belonging" or "not belonging", but is represented by a numerical value between 0 and 1. Through defuzzification processing, the uncertainty and fuzziness in the data can be better handled, providing a suitable data form for subsequent fuzzy rule-based reasoning.

[0221] Step S640: Perform inference based on a pre - constructed fuzzy rule base.

[0222] Pre - constructed fuzzy rule base: A set of rules established based on professional knowledge, practical experience, and a large amount of experimental data in the field of building thermodynamics before the thermal condition analysis. These rules are presented in the form of "if... then...", for example, "if the outdoor temperature is high and there is frequent human activity, then the thermal condition may be relatively hot". The fuzzy rule base associates the fuzzified input data (fuzzy membership information of the preliminary judgment result, the sum of input data, environmental data, and user behavior data) with possible thermal condition levels, providing a basis for inference.

[0223] Inference: The process of analyzing and processing the fuzzified input data using the rules in the fuzzy rule base to obtain a more accurate judgment of the thermal condition level. The inference process is not a simple logical judgment, but rather comprehensively considers multiple fuzzy input data and their membership degrees, and determines the membership degree distribution of the final thermal condition level through rule matching and calculation in the rule base.

[0224] Step S650: Use a defuzzification method to convert the fuzzy judgment result into a clear thermal condition level.

[0225] Defuzzification method: A means of converting the fuzzy membership degree distribution of the thermal condition level obtained through fuzzy inference into a clear and practical - decision - making - applicable thermal condition level. Since the fuzzy inference result is a distribution with different membership degrees for multiple thermal condition levels, defuzzification extracts a single and definite thermal condition level from this fuzzy information to clearly evaluate the building's thermal condition.

[0226] Fuzzy judgment result: The result obtained after performing inference on the fuzzified input data (preliminary judgment result, sum of input data, environmental data, user behavior data) based on the fuzzy rule base in Step S640. It is represented by the membership degree values corresponding to different thermal condition levels, reflecting the possibility degree of each thermal condition level under the current input data conditions.

[0227] Clear thermal condition level: A single thermal condition level determined after defuzzification processing, such as "comfortable", "uncomfortable", "overheated", etc. This level is a clear and definite description of the current thermal condition state of the building, which can be directly used to guide subsequent building thermal management decisions, such as whether to adjust the operation parameters of the air - conditioning system, whether to strengthen ventilation, etc.

[0228] Assume that continuing with the example of the thermal condition analysis of an office building, the fuzzy judgment result obtained after Step S640 is {'comfortable': 0.48, 'warm': 0.35, 'uncomfortable': 0.17}.

[0229] If the maximum membership degree method is used for defuzzification. The maximum membership degree method is to select the thermal condition level with the largest membership degree as the finally determined thermal condition level. Among the above fuzzy judgment results, the membership degree of "comfortable" is the largest, which is 0.48. Therefore, after defuzzification, it is determined that the currently determined thermal condition level of this office building is "comfortable".

[0230] If the centroid method is adopted, it is first necessary to set a quantization value for each thermal condition level. Assume that the quantization value of "comfortable" is 1, the quantization value of "relatively hot" is 2, and the quantization value of "uncomfortable" is 3. According to the centroid method formula, the numerator is calculated as: 0.48×1 + 0.35×2 + 0.17×3 = 0.48 + 0.7 + 0.51 = 1.69, and the denominator is: 0.48 + 0.35 + 0.17 = 1. Then the centroid value is 1.69÷1 = 1.69. According to the corresponding relationship between the pre-set quantization values and the thermal condition levels, it is judged that 1.69 is closer to the quantization value 1 corresponding to "comfortable". Therefore, it is determined that the currently determined thermal condition level of this office building is "comfortable". Through such a defuzzification process, the membership degree distribution of the thermal condition level obtained by fuzzy reasoning is transformed into a definite thermal condition level, providing a clear decision-making basis for building thermal management.

[0231] The reinforcement learning model selects an action according to the learned policy to adjust the current input data total range, including:

[0232] Step S611, the reinforcement learning model adopts a multi-layer perceptron structure. The input layer receives the state vector integrated by the input data total, building type, usage scenario, season, and the previously extracted comprehensive features. After non-linear transformation through multiple hidden layers, the output layer calculates the corresponding Q values for various predefined actions. Among them, various predefined actions include translation, scaling, splitting and merging.

[0233] Hidden layer: The neural network layer located between the input layer and the output layer, which transforms the input information through a non-linear activation function, extracts higher-level features, and enhances the expression ability of the model.

[0234] Non-linear transformation: The operation of transforming the neuron input using a non-linear activation function (such as ReLU, Sigmoid, etc.), enabling the neural network to learn complex non-linear relationships and improving the model's data processing ability.

[0235] Output layer: The last layer of the multi-layer perceptron, which calculates the Q values corresponding to various predefined actions according to the information output by the hidden layer.

[0236] Pre-defined actions: Operations set in advance for the range of the sum of input data, including translation (moving the entire range by a certain value), scaling (enlarging or reducing the range proportionally), splitting and merging (splitting the range into multiple smaller ranges or merging multiple ranges). The model evaluates the advantages and disadvantages of these actions by calculating the Q-value.

[0237] Q-value: An estimated value used to measure the long-term cumulative reward of taking a certain pre-defined action in the current state. The higher the Q-value, the more likely it is to obtain a better long-term result by taking the corresponding action in that state. The model selects the optimal action based on the Q-value.

[0238] Step S612, use The policy to determine the action to be taken.

[0239] Among them, is the ε-greedy policy. To balance exploration and exploitation, the ε-greedy policy randomly selects an action with a probability of ε and selects the action with the highest Q-value with a probability of 1 - ε. Among them, ε is a parameter between 0 and 1, called the exploration rate. For example, when ε = 0.1, the model has a 10% probability of randomly selecting an action and a 90% probability of selecting the action with the highest Q-value. In this way, the model can not only use the learned knowledge to select the current optimal action but also explore new actions with a certain probability to avoid falling into local optima.

[0240] Suppose a thermal condition analysis of a hotel is carried out. Step S611 has been completed, and the Q-values for pre-defined actions such as translation, scaling, splitting and merging are obtained, which are 0.7 (translation), 0.5 (scaling), and 0.3 (splitting and merging) respectively.

[0241] Load the reinforcement learning model stored in a file (assuming PyTorch is used). The policy adopted by this model is the ε-greedy policy, and ε = 0.2.

[0242] The model first generates a random number between 0 and 1. Suppose the generated random number is 0.15. Since this random number is less than the exploration rate ε (0.2), the model will select an action randomly. Randomly select one from the pre-defined actions (translation, scaling, splitting and merging). Suppose the scaling action is randomly selected.

[0243] If the generated random number is greater than 0.2, for example, 0.8, then the model will select the action with the highest Q-value, that is, the translation action, according to the greedy policy. By using the policy in this way to determine the action to be taken, it prepares for adjusting the range of the sum of input data according to the action later to analyze the thermal condition of the hotel more accurately.

[0244] Step S613: Determine the interval adjustment strategy according to the mapping relationship between the determined action and the interval adjustment strategy, and apply it to the original interval to complete the adjustment of the range of the sum interval of the input data.

[0245] Interval adjustment strategy: Specific adjustment methods and rules preset for each predefined action. For example, for the translation action, the interval adjustment strategy may stipulate moving the entire interval to the left or right by a certain value; for the scaling action, it may stipulate expanding or shrinking the interval range by a certain scale factor; for the splitting / merging action, it will clarify how to split the original interval into multiple small intervals or how to merge multiple small intervals into a new interval.

[0246] Mapping relationship: The corresponding connection between the action and the interval adjustment strategy. It associates the action determined in step S612 with the corresponding specific adjustment method, enabling the model to find the corresponding interval adjustment strategy based on the selected action, thereby adjusting the range of the sum interval of the input data. This mapping relationship is determined and stored during the model construction and training phases.

[0247] Original interval: The range of the sum of the input data determined before step S611, which is the basis for adjustment. This interval range is set according to the initial input data and certain rules, and needs to be adjusted according to the learning and judgment of the model as the heat condition analysis process progresses.

[0248] Suppose a heat condition analysis is performed on a hotel, and the action determined in step S612 is "scaling". First, read the mapping relationship between the action and the interval adjustment strategy from the file. Search for the interval adjustment strategy corresponding to the "scaling" action in the file. Suppose the found description is "shrink the original interval range by a scale factor of 0.8". Then, read the original interval range from the file. Suppose the original interval is [100, 200]. According to the determined interval adjustment strategy, perform a scaling operation on the original interval. The lower limit value is adjusted to 100×0.8 = 80, and the upper limit value is adjusted to 200×0.8 = 160, resulting in the adjusted interval [80, 160]. Through such a process, based on the action determined in step S612, find the corresponding interval adjustment strategy and apply it to the original interval to complete the adjustment of the range of the sum interval of the input data, providing a data basis for more accurate analysis of the hotel's heat condition in the future. For example, when judging the heat condition level, the adjusted interval range can better reflect the actual heat condition of the current hotel.

[0249] Furthermore, the intelligent control method for building energy-saving walls also includes a distributed control method for building energy-saving walls based on intelligent edge computing, specifically as follows:

[0250] Distributed nodes are deployed on each floor or functional area of the building. Each node collects parameters related to the environment (temperature, humidity, light, etc.), user behavior (personnel activities, equipment status, etc.), and heat transfer (wall temperature, heat flux density, etc.) in real time through a local sensor network. Each edge computing node independently conducts heat demand analysis and heat condition level judgment based on the local training model and real-time data, and determines a preliminary regulation plan from the local policy library, such as regulating phase change materials on this floor and fine-tuning the local lattice structure.

[0251] Every 15 minutes (preset interval), each distributed node sends local environmental data, heat condition analysis results, and preliminary regulation decisions to adjacent nodes and the central control platform.

[0252] The central control platform and nodes complete the detection of regulation decision conflicts within 1 hour after receiving data based on a pre-established rule library and optimization algorithm. If a conflict is detected, the central control platform uses a multi-objective optimization algorithm (such as setting the number of iterations of the genetic algorithm to 50 times) within 15 minutes to generate a collaborative regulation plan by integrating objectives such as energy conservation and comfort. After the central control platform generates the plan, it sends it to each distributed node within 3 minutes, and the node completes the adjustment of the local decision within 2 minutes and feeds back the execution preparation status.

[0253] After each distributed node receives the collaborative regulation plan, it sends a control instruction to the actuator within 5 minutes, and the actuator completes the adjustment of parameters such as phase change materials, lattice structures, and heat exchange fluids within 10 minutes.

[0254] After the adjustment is completed, the sensor collects the regulated data in real time and feeds back the environmental and heat transfer-related data to the central control platform and nodes every 30 minutes. After the node and the platform receive the feedback data, they complete the evaluation of the regulation effect within 15 minutes according to the preset evaluation indicators (the energy consumption reduction rate is accurate to two decimal places, the indoor temperature fluctuation range is accurate to 0.1 °C, etc.). If it does not meet the standard, the conflict detection and subsequent processes are repeated.

[0255] This application also provides a lattice structure wall. The lattice structure wall includes a lattice structure frame and a phase change material filled in the internal space of the lattice structure frame. The lattice structure wall also includes a macroscopic encapsulation structure that encapsulates the phase change material inside the lattice structure frame and concrete that is poured into a template embedded with the lattice structure frame to tightly combine the lattice with the concrete. The phase change material is paraffin or hydrated salt. The hydrated salt phase change material has a higher latent heat of phase change, a larger heat storage density, and better thermal conductivity, thereby further enhancing the heat storage and heat regulation capabilities of the wall. At the same time, compared with paraffin, the hydrated salt phase change material is more environmentally friendly and has a lower cost, making it more suitable for wide application in the field of building energy conservation. The macroscopic encapsulation process simplifies the filling and encapsulation steps, avoids the complex manufacturing process in microcapsule technology, and reduces production costs.

[0256] When the lattice structure wall is a single-layer lattice structure, its composition can be referred to Figure 2 as shown. The lattice framework is manufactured by 3D printing technology and forms a structure with high space utilization through precise geometric design. The interior of the lattice structure is filled with phase change materials (such as paraffin or hydrated salts). The phase change materials are injected through a specially designed inlet (upper side) and the discharge or replacement of the materials is achieved through an independent outlet (lower side), forming an efficient dynamic regulation loop. The arrangement of the inlet and outlet ensures that the phase change materials can be evenly distributed in the lattice structure, providing stable heat storage and release capabilities.

[0257] Figure 3 Shows a partial enlarged view of the inlet of the single-layer lattice structure. The lattice interior is filled with phase change materials (such as paraffin or hydrated salts) for efficient heat storage and release. The lattice provides a stable filling space for the phase change materials and increases the heat contact area with the outside world through an optimized geometric shape, thereby improving the heat storage and heat release efficiency. The design of the lattice simplifies the packaging process of the phase change materials and ensures their stability and efficiency in thermal management, suitable for various building energy-saving scenarios.

[0258] Furthermore, the lattice structure wall can also be a double-layer lattice structure. In this case, the lattice structure framework is a double-layer lattice structure framework. The inner layer lattice provides a stable filling space for the phase change materials, and the outer layer lattice is arranged around the inner layer and is provided with a heat exchange fluid circulation channel filled with a heat exchange fluid. The heat exchange fluid is water or ethylene glycol solution, etc. Specifically, it can be referred to Figure 4 , Figure 4 which shows the composition and functional design of the double-layer lattice structure.

[0259] Figure 5 Shows a partial enlarged view of the inlet of the double-layer lattice structure. The lattice is divided into an inner layer and an outer layer. The inner layer is filled with phase change materials (such as paraffin or hydrated salts); the outer layer is filled with a heat exchange medium (such as water), and the rapid conduction and dissipation of heat are achieved through the circulation of the fluid. The function of the inner layer lattice is the same as that of the single-layer lattice structure, and the outer layer lattice is designed as a fluid channel for accommodating and guiding the flow of the heat exchange fluid to ensure that heat can be quickly transferred to the external environment or internal space of the wall.

[0260] The embodiments of this specific implementation manner are all preferred embodiments of this application, and do not limit the protection scope of this application accordingly. Therefore, all equivalent changes made according to the structure, shape, and principle of this application should be covered within the protection scope of this application.

Claims

1. A method for intelligent management and control of building energy-saving walls, characterized in that: include: Obtain environmental data and user behavior data, and perform data preprocessing; Input environmental data and user behavior data into the pre-built heat demand prediction model to output building heat demand information; Collect and obtain real-time monitoring data of heat transfer related parameters, and input the real-time monitoring data into the pre-built multi-physics field coupled heat transfer model. The multi-physics field coupled heat transfer model calculates the heat transfer process of the fine units of the wall structure based on the finite element analysis method, and obtains the heat transfer model analysis results by solving the multi-physics field coupling equations. The heat transfer model analysis results include the temperature distribution at different locations of the wall at different times, the heat transfer rate, and the phase change state of the phase change material; Standardize the data included in the building heat demand information and the data included in the heat transfer model analysis results to make them uniform in scope; According to different building types, usage scenarios and seasons, weights are set for each input data, and the total input data is calculated using the weighted average method; According to the mapping relationship between the interval range that the sum of input data falls into and the level, the level corresponding to the thermal condition is analyzed and determined; According to the mapping relationship between the thermal condition level and the overall control plan, the overall control plan is analyzed and determined, and the determined overall control plan is executed. The overall control plan includes a phase change material control plan, a lattice structure control plan, and a heat exchange fluid control plan. Among them, the phase change material control plan: adjustments are made around the phase change material filled in the wall lattice structure. The phase change material can absorb or release a large amount of latent heat during the solid-liquid phase change process, which has a significant impact on the thermal performance of the wall. This plan chooses to replace phase change materials with different melting points according to seasonal changes, external temperature changes, and the thermal needs of the building; lattice structure control Solution: Focus on optimizing and adjusting the lattice structure of 3D printing. The lattice structure provides a stable filling space for the phase change material. Its geometric shape and structural parameters directly affect the thermal contact area between the phase change material and the outside world and the heat conduction efficiency. The control scheme includes changing the pore size of the lattice and the arrangement of the lattice units; Heat exchange fluid control scheme: Precisely control the fluid in the heat exchange fluid circulation channel of the outer lattice. The fluid circulates in the channel and plays an important role in transferring heat. This scheme achieves efficient transfer and dissipation of wall heat by adjusting the circulation rate and temperature of the fluid; The multi-physics coupled heat transfer model calculates the heat transfer process of the fine units of the wall structure based on the finite element analysis method. By solving the multi-physics coupled equations, the heat transfer model analysis results include: According to the preset fine unit division basis, the wall structure is divided into fine units based on the finite element analysis method; For each fine unit, the heat transfer equation is constructed based on the physical laws of heat conduction, convection, and radiation; After completing the construction of the heat transfer equation for each tiny unit, the field coupling terms are constructed based on the mutual influence between the physical fields, and the coupling terms are added to the unit equation; Assemble the unit equations after adding coupling terms to form a multi-physics field coupling equation group of the entire wall structure; The finite element solution method based on the preconditioned conjugate gradient method is used to iteratively solve the equations and obtain the analysis results of the heat transfer model.

2. The intelligent management and control method for energy-saving building walls according to claim 1, characterized in that: The steps to build the heat demand forecasting model are as follows: Collect environmental data and user behavior data, and perform data preprocessing; Build a Transformer model based on spatiotemporal decoupling, randomly initialize the model parameters, and determine the model hyperparameters; Divide the preprocessed data into a training set and a validation set according to a preset ratio; The training set data is input into the Transformer model based on spatiotemporal decoupling, and the prediction results are calculated through forward propagation. The weighted mean square error loss function is used to calculate the difference between the prediction results and the actual heat demand value. In addition, a Dropout operation is performed during each forward propagation to obtain the prediction results under different parameter samples. Based on the results of the loss function, the gradient is calculated through the back-propagation algorithm to update the model parameters; After each training cycle is completed, the validation set loss is calculated. If the change in validation set loss for consecutive preset cycles is less than the set threshold, the training is considered completed.

3. The intelligent management and control method for energy-saving building walls according to claim 1, characterized in that: The steps to build a multiphysics coupled heat transfer model are as follows: Collect and obtain multi-source data and perform data preprocessing; Build a model architecture at the micro, meso and macro levels; According to the correspondence between the building thermal management scenario and the physical field coupling mode and intensity setting scheme, the physical field coupling mode and intensity of the current building thermal management scenario are analyzed and determined as the starting state of the model training; The nonequilibrium heat flow equation Integrate into the model heat transfer calculation system, where is the heat flux density, is the thermal conductivity, is the temperature gradient, is the pressure gradient, is the substance concentration gradient, is the pressure-dependent heat transfer coefficient, is the heat transfer coefficient related to the species concentration; A deep neural network with a multilayer perceptron structure of 3 hidden layers and 100 neurons in each layer was selected and integrated into the model using a residual connection fusion strategy. The input layer of the deep neural network received temperature, humidity, and building structure parameter data, the hidden layer used the ReLU function to extract features, and the output layer predicted heat transfer related parameters. Input the preprocessed data into a model that has a well-built architecture, a determined coupling method, incorporates non-equilibrium equations, and integrates a deep neural network; During training, the mean square error loss function is used to calculate the deviation between the model prediction value and the true value. The stochastic gradient descent method is used to iteratively update the model parameters at a learning rate of 0.01 to reduce the loss function value. After every 10 parameter updates, the weight of the influence of the coupling strength of each physical field on the loss function is calculated based on the model prediction results and real data under the latest parameters. According to the weight, the coupling strength coefficient is adjusted by 10% to reduce the loss function value. According to the dynamic change results of the scene within each interval time range, the physical field coupling strength is changed through a preset coupling strength adjustment function; The mean square error is continuously monitored. When the mean square error change value is less than the preset change value in consecutive preset rounds of training, the training is considered complete.

4. A building energy-saving wall intelligent management and control method according to claim 3, characterized in that: The preset coupling strength adjustment function is as follows: ； For Moment, physical field With physical field Adjusted coupling strength coefficient; is the number of scenario factors that affect the coupling strength; is the weight of the kth scenario factor, reflecting the relative importance of this factor in adjusting the coupling strength; is the kth scene factor At the moment The influence function is used to quantify the influence of the change of this factor on the coupling strength; is the kth scene factor at time The specific state or value of For physical field With physical field The basic coupling strength coefficient.

5. A method for intelligent management and control of building energy-saving walls according to claim 4, characterized in that: According to different building types, usage scenarios and seasons, weights are set for various input data, including: Aggregate the preprocessed multivariate data and extract basic features directly related to building type, usage scenario, and season from the multivariate data; The convolutional neural network is used to process the season-related data to obtain the season feature vector. At the same time, the long short-term memory network is used to analyze the usage scenario data, capture the characteristics of the usage scenario, and obtain the scenario feature vector. The seasonal feature vector and the scene feature vector are fused to form a comprehensive feature; The comprehensive features are input into the dynamic weight generation model based on the attention mechanism neural network, and the weights set for each input data are output.

6. A building energy-saving wall intelligent management and control method according to claim 5, characterized in that: According to the mapping relationship between the range in which the sum of the input data falls and the level, the levels corresponding to the thermal conditions are analyzed and determined to include: The acquired input data sum, building type, usage scenario, season and comprehensive features are input into the trained reinforcement learning model. The reinforcement learning model selects an action based on the learned strategy and adjusts the current input data sum range. According to the adjusted interval range, determine the corresponding thermal condition level as the preliminary judgment result; The preliminary judgment result is fuzzified together with the sum of input data, environmental data, and user behavior data; Reasoning based on a pre-built fuzzy rule base; The defuzzification method is used to transform the fuzzy judgment results into clear thermal condition levels.

7. A method for intelligent management and control of building energy-saving walls according to claim 6, characterized in that: The reinforcement learning model selects an action based on the learned strategy and adjusts the current input data sum range including: The reinforcement learning model adopts a multi-layer perceptron structure. The input layer receives the state vector integrated by the sum of input data, building type, usage scenario, season, and comprehensive features extracted in the early stage. After nonlinear transformation of multiple hidden layers, the output layer calculates the corresponding Q value for various predefined actions, including translation, scaling, splitting and merging. use The strategy determines the actions to be taken; According to the mapping relationship between the determined action and the interval adaptation strategy, the interval adaptation strategy is determined and applied to the original interval to complete the adjustment of the interval range of the sum of the input data.

8. A lattice structure wall, characterized in that: According to the method described in any one of claims 1 to 7, the lattice structure wall comprises: Lattice structure framework; Phase change material, which is filled in the internal space of the lattice structure frame and is paraffin or hydrated salt, is used to store and release heat; A macro packaging structure for packaging the phase change material within a lattice structure framework; Concrete is poured into the formwork embedded with the lattice structure frame, so that the lattice and concrete are tightly combined.

9. The lattice structure wall according to claim 8, characterized in that: The lattice structure frame is a double-layer lattice structure frame, the inner lattice provides a stable filling space for the phase change material, the outer lattice is arranged around the inner layer, and a heat exchange fluid circulation channel is provided, and the channel is filled with heat exchange fluid.

Images