A method, system, device and medium for adaptive calculation of building cooling and heating loads

Abstract

The present invention provides a method, system, device and medium for adaptive calculation of building cooling and heating loads, belonging to the technical field of load prediction. The method comprises: collecting image data and power consumption data of a building, extracting time series features from the power consumption data after preprocessing; extracting image features from the image data using edge detection, feature point recognition and semantic segmentation technology; generating a three-dimensional point cloud using image features through multi-view geometry technology and surface reconstruction method, and converting the three-dimensional model into a BIM model using Revit software; calculating solar radiation and internal heat sources based on power consumption data and the power of internal equipment of the building, and training a cooling and heating load prediction model using random forest and neural network algorithms; dynamically adjusting and optimizing the cooling and heating load prediction model using online learning and grid search technology, inputting real-time power consumption data and environmental parameters into the cooling and heating load prediction model, and generating a cooling and heating load prediction result for the building.

Description

Technical Field

[0001] The present invention belongs to the technical field of load forecasting, and more specifically relates to a method, system, equipment and medium for adaptively calculating cooling and heating loads of a building. Background Art

[0002] At present, building energy management is becoming increasingly important as a key link in promoting sustainable development. Accurate calculation of building cooling and heating loads is not only the cornerstone of ensuring energy efficiency and comfort during the building design stage, but also the key to achieving energy conservation and emission reduction and optimizing energy utilization during building operation. However, traditional cooling and heating load calculation methods are unable to cope with the complexity and dynamism of modern buildings.

[0003] The first problem is data acquisition. Traditional methods rely heavily on detailed architectural design drawings and precise parameters, but in actual operations, especially for buildings with a long history or undergoing renovation, these data are often incomplete or difficult to trace, which greatly limits the accuracy and practicality of calculations. This not only prolongs the project cycle, but also increases costs.

[0004] The problem of calculation accuracy cannot be ignored either. Empirical formulas and simplified models are not able to predict the complex geometric structures, changing building materials and increasingly complex indoor environments in modern buildings. The results of these simplified processes often deviate from reality and cannot provide strong data support for architectural design, nor can they meet the current urgent need for high-precision energy-saving design.

[0005] More importantly, traditional methods are difficult to adapt to the dynamic changes in building cooling and heating loads. The cooling and heating demand of a building is a complex system affected by multiple factors, including but not limited to outdoor temperature, humidity, wind speed, solar radiation, as well as indoor personnel activities and equipment usage. The changes in these factors are real-time and unpredictable, and traditional methods are based on static assumptions and lack flexible adjustment mechanisms, resulting in significant deviations between the predicted results and actual needs, which not only affects the comfort of the indoor environment, but also causes energy waste. Therefore, exploring more advanced, efficient, and adaptable cooling and heating load calculation methods has become an important issue that needs to be urgently addressed in the construction field. Summary of the invention

[0006] In view of the above problems, the purpose of the present invention is to provide a method, system, equipment and medium for adaptive calculation of building cooling and heating loads. By comprehensively using computer vision, building information modeling and machine learning technology, high-precision and adaptive calculation of building cooling and heating loads is achieved, which is suitable for scenarios such as building design, operation management and energy-saving optimization.

[0007] In order to achieve the above object, the present invention is implemented through the following technical solutions:

[0008] In a first aspect, an embodiment of the present application provides a method for adaptively calculating cooling and heating loads of a building, comprising:

[0009] Collect image data and electricity consumption data of buildings, and extract time series features from the electricity consumption data after preprocessing;

[0010] Extract image features from image data using edge detection, feature point recognition and semantic segmentation techniques;

[0011] The extracted image features are used to generate a 3D point cloud through multi-view geometry technology and surface reconstruction methods, and the 3D model is converted into a BIM model using Revit software;

[0012] Based on the electricity consumption data and the power of the equipment inside the building, the PySolar library is used to calculate the solar radiation and internal heat source, and the random forest and neural network algorithms are used to train the cooling and heating load prediction model;

[0013] Online learning and grid search technology are used to dynamically adjust and optimize the cooling and heating load prediction model. Real-time electricity consumption data and environmental parameters are input into the cooling and heating load prediction model to generate building cooling and heating load prediction results.

[0014] In an optional embodiment, the image data and electricity consumption data of the building are collected, and after preprocessing, time series features are extracted from the electricity consumption data, including:

[0015] Use drones and handheld cameras to collect photos of key parts of the building's interior and exterior from multiple angles to generate image data;

[0016] Use the OpenCV library to perform image enhancement and geometric correction on the image data, and use the LabelMe tool to use rectangular boxes or polygonal boxes to mark key parts in the image;

[0017] Save the annotation results as a JSON file.

[0018] In an optional embodiment, the collecting of image data and electricity consumption data of the building, extracting time series features from the electricity consumption data after preprocessing, further includes:

[0019] Use existing energy consumption systems or smart meters to collect the building's electricity consumption data for the past year, including air conditioning power consumption data, lighting power consumption data, and equipment power consumption data; organize the collected data into power consumption data and store it in a cloud server or local database;

[0020] Use the Z-score method or the isolation forest algorithm to remove outliers in the electricity consumption data, and use mean filling or interpolation to handle missing values ​​in the electricity consumption data;

[0021] Based on the electricity consumption data, the average daily electricity consumption, seasonal changes in electricity consumption and peak load are calculated to generate time series characteristics.

[0022] In an optional embodiment, the extracting image features from the image data using edge detection, feature point recognition and semantic segmentation technology includes:

[0023] According to the image data, the Canny algorithm is used to detect the edge information of the image through Gaussian filtering, gradient calculation, non-maximum suppression and double threshold detection in sequence;

[0024] The SIFT algorithm is used to construct the scale space and differential Gaussian pyramid based on the image data, detect extreme points and calculate descriptors, identify the key feature points of the image data, and determine the location and descriptors of the key feature points;

[0025] Using the Mask R-CNN model, high-level features of image data are extracted and regions are generated through convolutional neural networks and region proposal networks. Classification and bounding box regression are performed on each candidate region to generate a segmentation mask for each candidate region.

[0026] In an optional embodiment, the method of generating a 3D point cloud using the extracted image features by using multi-view geometry technology and a surface reconstruction method, and converting the 3D model into a BIM model using Revit software, comprises:

[0027] Based on the edge information of the image, the location and descriptor of key feature points, and the segmentation mask of each candidate area, a 3D point cloud is constructed by matching feature points under different perspectives using camera calibration parameters and triangulation method, and then the 3D point cloud is converted into a 3D model through surface reconstruction method.

[0028] Convert the 3D model into BIM format, and use Revit software to mark wall materials, window types, and equipment locations to generate a BIM model.

[0029] In an optional embodiment, the method of calculating solar radiation and internal heat source based on electricity consumption data and the power of internal equipment in the building in combination with the PySolar library, and training the cold and hot load prediction model using random forest and neural network algorithms includes:

[0030] According to the orientation and geographical location of the building, the PySolar library is used to calculate the solar radiation intensity I on the building surface according to the following formula:

[0031] I = Ibcos(θ) + Id(1+cos(β)) / 2

[0032] Where, Ib is the direct radiation intensity, Id is the scattered radiation intensity, θ is the incident angle, and β is the inclination angle of the building surface;

[0033] Based on the electricity consumption data and the power of the equipment inside the building, the heat Q generated by the internal heat source is calculated according to the formula Q=P×t, and the heat Qp generated by the personnel is calculated according to the formula Qp=n×q;

[0034] Where P is the power of the equipment, t is the operating time of the equipment, n is the number of people, and q is the heat generated per person per hour;

[0035] Random forest and neural network algorithms are used to combine image features and electricity consumption data to generate and train the cold and hot load prediction model, and cross-validation and grid search techniques are used to optimize the parameters of the cold and hot load prediction model.

[0036] In an optional implementation, the online learning and grid search technology is used to dynamically adjust and optimize the cooling and heating load prediction model, and real-time power consumption data and environmental parameters are input into the cooling and heating load prediction model to generate the building cooling and heating load prediction results, including:

[0037] Obtain the building's electricity consumption data and environmental parameters in real time and pre-process the data;

[0038] The pre-processed electricity consumption data and environmental parameters are input into the cooling and heating load prediction model, and the model parameters are dynamically adjusted using incremental learning or online learning algorithms;

[0039] Define a search space for the key parameters of the cooling and heating load prediction model. In the search space, use the grid search technique to try different parameter combinations, and use the cross-validation technique to evaluate the performance of each parameter combination to determine the optimal parameter combination as the final parameters of the cooling and heating load prediction model.

[0040] Input the real-time power consumption data and environmental parameters into the cooling and heating load prediction model to generate the cooling and heating load prediction results of the building;

[0041] Collect actual electricity consumption data from the building, perform error analysis based on the building cooling and heating load prediction results, and adjust the model parameters or structure based on the analysis results.

[0042] In a second aspect, the embodiment of the present application further provides a building cooling and heating load adaptive calculation system, comprising:

[0043] The data collection and preprocessing module is used to collect the image data and electricity consumption data of the building, and after preprocessing, extract the time series features from the electricity consumption data;

[0044] An image feature extraction module is used to extract image features from image data using edge detection, feature point recognition and semantic segmentation techniques;

[0045] 3D reconstruction and BIM model generation module, which is used to generate 3D point cloud using extracted image features through multi-view geometry technology and surface reconstruction method, and convert the 3D model into BIM model using Revit software;

[0046] Model building module, which is used to calculate solar radiation and internal heat sources based on electricity consumption data and the power of equipment inside the building, combined with the PySolar library, and train the cooling and heating load prediction model using random forest and neural network algorithms;

[0047] The prediction module is used to dynamically adjust and optimize the cooling and heating load prediction model using online learning and grid search technology, input real-time power consumption data and environmental parameters into the cooling and heating load prediction model, and generate the cooling and heating load prediction results of the building.

[0048] In a third aspect, an embodiment of the present application further provides an electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein when the processor executes the program, the steps of the method for adaptively calculating the cooling and heating loads of a building as described in any one of the above items are implemented.

[0049] In a fourth aspect, an embodiment of the present application further provides a storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the method for adaptively calculating cooling and heating loads of a building as described in any one of the above items are implemented.

[0050] It can be seen from the above technical solutions that the present invention has the following advantages:

[0051] In the adaptive calculation method for building cooling and heating loads provided in this application, accurate prediction and dynamic optimization of building cooling and heating loads are achieved through data fusion, integration of three-dimensional models and BIM technology, accurate calculation of solar radiation and internal heat sources, advanced prediction models and optimization technology, and real-time prediction and feedback optimization cycles. This not only improves the efficiency and accuracy of building energy consumption management, but also provides strong support for the sustainable development of buildings and energy conservation and emission reduction.

[0052] This application uses edge detection, feature point recognition and semantic segmentation technology to extract image features from image data, and uses time series analysis technology to extract time series features from electricity consumption data. By fusing the image data and electricity consumption data of the building, it can fully capture the physical characteristics and operating conditions of the building. Image data provides spatial structure and material information of the building, while electricity consumption data reflects the actual energy consumption of the building. This data fusion makes the cold and hot load forecast more accurate and comprehensive.

[0053] This application generates a 3D point cloud using multi-view geometry technology and surface reconstruction methods, and converts it into a BIM model using Revit software, providing accurate building space information for cooling and heating load prediction. The BIM model not only contains the spatial structure of the building, but also contains detailed information such as wall materials, window types, equipment locations, etc., which are essential for accurately calculating cooling and heating loads.

[0054] This application combines the PySolar library to calculate solar radiation and calculate internal heat sources based on power consumption data and equipment power, which can accurately assess the external and internal heat loads on the building. This is of great significance for improving the accuracy and robustness of cooling and heating load prediction.

[0055] This application uses random forest and neural network algorithms to train the cooling and heating load prediction model, and uses online learning and grid search technology for dynamic adjustment and optimization, which can ensure that the prediction model is always in the best state. This dynamic optimization mechanism enables the prediction model to quickly adapt to changes in building usage patterns and external environmental conditions.

[0056] This application obtains the building's electricity consumption data and environmental parameters in real time, and inputs the pre-processed data into the cooling and heating load prediction model. This solution can generate the building's cooling and heating load prediction results in real time. At the same time, by collecting actual electricity consumption data for error analysis and adjusting the model parameters or structure according to the analysis results, the prediction model is continuously optimized and improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solution of the present invention, the accompanying drawings required for use in the description will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0058] Figure 1 A schematic diagram of the flow chart of the adaptive calculation method for building cooling and heating loads provided in this application.

[0059] Figure 2 A schematic diagram of the process of image feature extraction provided in this application.

[0060] Figure 3 A schematic diagram of the process flow of the 3D reconstruction and BIM model generation method provided in this application.

[0061] Figure 4 This is a schematic diagram of the structure of the building cooling and heating load adaptive calculation system provided in this application.

[0062] Figure 5 A schematic diagram of the structure of the electronic device provided in this application. DETAILED DESCRIPTION

[0063] In the specific steps of the building cooling and heating load adaptive calculation method described in detail below, various embodiments of the present disclosure will be described more comprehensively. The present disclosure may have various embodiments, and adjustments and changes may be made therein. However, it should be understood that there is no intention to limit the various embodiments of the present disclosure to the specific embodiments disclosed herein, but the present disclosure should be understood to cover all adjustments, equivalents and / or alternatives that fall within the spirit and scope of the various embodiments of the present disclosure.

[0064] The core of the present invention is to provide a method for adaptively calculating the cooling and heating loads of a building. In the related art, the traditional cooling and heating load calculation method is highly dependent on detailed architectural design drawings and precise parameters, and has calculation accuracy problems, making it difficult to adapt to the dynamic changes of the cooling and heating loads of a building.

[0065] The adaptive calculation method for building cooling and heating loads provided by the present invention realizes the optimization of the whole process from data collection, feature extraction, model construction to adaptive calculation by comprehensively using computer vision, building information modeling (BIM) and machine learning technology, thereby realizing the accurate prediction and dynamic optimization of building cooling and heating loads. This not only improves the efficiency and accuracy of building energy consumption management, but also provides strong support for the sustainable development and energy conservation and emission reduction of buildings.

[0066] Hereinafter, the terms "include" or "may include" used in various embodiments of the present disclosure indicate the presence of disclosed functions, operations, or elements, and do not limit the addition of one or more functions, operations, or elements. In addition, as used in various embodiments of the present disclosure, the terms "include", "have", and their cognates are intended only to indicate specific features, numbers, steps, operations, elements, components, or a combination of the foregoing, and should not be understood as first excluding the presence of one or more other features, numbers, steps, operations, elements, components, or a combination of the foregoing or the possibility of adding one or more features, numbers, steps, operations, elements, components, or a combination of the foregoing.

[0067] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0068] See also Figure 1 The figure is a flow chart of a method for adaptively calculating building cooling and heating loads in a specific embodiment, the method comprising:

[0069] S1: Collect image data and electricity consumption data of buildings, and extract time series features from the electricity consumption data after preprocessing.

[0070] First, we need to collect image data and electricity data of the building. Image data includes photos of the exterior and interior of the building, which can be taken by drones, handheld cameras and other equipment to ensure that all facades and major interior spaces of the building are covered. Electricity data is collected through smart meters or manual records, covering the building's itemized electricity data for the past year, including electricity consumption of various parts such as air conditioning, lighting, and equipment. These data are the basis for subsequent calculations, and their integrity and accuracy must be guaranteed.

[0071] S2: Use edge detection, feature point recognition and semantic segmentation techniques to extract image features from image data.

[0072] Edge detection is an important step in image processing, which is used to extract the outline information of buildings. The Canny edge detection algorithm is used, which can accurately detect the edges in the image through multi-level filtering and non-maximum suppression technology. The edge detection results provide a basis for subsequent feature point recognition and semantic segmentation.

[0073] Feature point recognition is an important means of extracting key features of buildings. Using the SIFT (Scale Invariant Feature Transform) feature extraction algorithm, key feature points in images can be identified. These feature points include door and window locations, wall thickness, etc.

[0074] This step uses the deep learning model Mask R-CNN for semantic segmentation. The model uses convolutional neural network (CNN) and region proposal network (RPN) technology to accurately identify and segment different parts of the image.

[0075] S3: The extracted image features are used to generate a 3D point cloud through multi-view geometry technology and surface reconstruction methods, and the 3D model is converted into a BIM model using Revit software.

[0076] Among them, the multi-view geometry method can construct a three-dimensional point cloud by matching feature points under different perspectives, and then generate a three-dimensional model of the building.

[0077] S4: Based on the electricity consumption data and the power of the equipment inside the building, the PySolar library is used to calculate the solar radiation and internal heat source, and the random forest and neural network algorithms are used to train the cooling and heating load prediction model.

[0078] In this step, solar radiation and internal heat sources are calculated first, and the cold and hot load prediction model is trained by combining image features and electricity consumption data through machine learning algorithms to achieve the goal of accurately predicting the cold and hot load of the building under different conditions. Through the above method, combined with the public library data of heat transfer coefficients of various insulation materials in the building, a high-performance cold and hot load prediction model can be trained. The model can not only accurately predict the cold and hot load of the building under different conditions, but also provide optimization suggestions to help managers formulate reasonable energy management strategies.

[0079] S5: Use online learning and grid search technology to dynamically adjust and optimize the cooling and heating load prediction model, input real-time power consumption data and environmental parameters into the cooling and heating load prediction model, and generate the cooling and heating load prediction results of the building.

[0080] For example, firstly, the power consumption data and environmental parameters of the building are obtained in real time, and the data is preprocessed; then the preprocessed power consumption data and environmental parameters are input into the cold and hot load prediction model, and the incremental learning or online learning algorithm is used to dynamically adjust the model parameters; at this time, the search space is defined for the key parameters of the cold and hot load prediction model, and different parameter combinations are tried in the search space using the grid search technology, and the performance of each parameter combination is evaluated by the cross-validation technology to determine the optimal parameter combination as the final parameter of the cold and hot load prediction model. Furthermore, the real-time power consumption data and environmental parameters are input into the cold and hot load prediction model to generate the cold and hot load prediction results of the building. Finally, the actual power consumption data is collected from the building, and the error analysis is performed based on the cold and hot load prediction results of the building, and the model parameters or structure are adjusted according to the analysis results.

[0081] In an exemplary embodiment, this step can implement adaptive calculation and feedback optimization:

[0082] Adaptive computing dynamically adjusts the cold and hot load forecasting model based on real-time power consumption data and environmental parameters to ensure the real-time and accuracy of the forecast results. First, by installing sensors and smart meters, the building's power consumption data and environmental parameters, such as outdoor temperature and humidity, are monitored in real time, and these data are transmitted to the central server for processing. Next, the collected data is preprocessed, including removing outliers and filling missing values ​​to ensure data quality. Then, the model parameters are updated using this real-time data so that the model can respond quickly to new information. An online learning method (incremental learning) is used to gradually adjust the model parameters to ensure that the model can promptly reflect the latest power consumption patterns and environmental changes to maintain its accuracy and robustness.

[0083] Feedback optimization continuously improves the cold and hot load prediction model through feedback from actual operation data, thereby improving its prediction accuracy and robustness. Specifically, the actual operation data of the building, including actual power consumption, indoor temperature, indoor humidity, etc., is collected and compared with the prediction results of the model. The performance of the model is evaluated by calculating the prediction error (mean square error MSE, mean absolute error MAE). According to the error analysis results, the model parameters are adjusted and the model structure is optimized. Grid search technology is used to try different parameter combinations and select the optimal parameters. Through continuous feedback and optimization, the model can more accurately predict the cold and hot loads of the building under different conditions, providing reliable decision support for energy management and equipment operation. This method not only improves the prediction accuracy of the model, but also enhances its ability to adapt to complex environmental changes, ensuring the stability and reliability of the model in practical applications.

[0084] In this embodiment, by comprehensively using computer vision, building information modeling and machine learning technology, high-precision and adaptive calculation of building cooling and heating loads is achieved, which is suitable for scenarios such as building design, operation management and energy-saving optimization.

[0085] In an embodiment of the present invention, based on step S1, a possible embodiment is given below to illustrate its specific implementation scheme in a non-limiting manner.

[0086] S101: Image data collection, preprocessing and storage. This includes: using drones and handheld cameras to collect photos of key parts inside and outside the building from multiple angles to generate image data; using the OpenCV library to perform image enhancement and geometric correction on the image data in turn, and using the LabelMe tool to use rectangular boxes or polygonal boxes to mark key parts in the image; and saving the annotation results as a JSON file.

[0087] As an example, image data collection can be achieved in the following ways:

[0088] Collect images - photos of the building exterior: Use a drone or handheld camera to take panoramic photos of the building exterior from multiple angles and heights; the shooting range should cover all facades of the building, including the main facade, secondary facade, roof and ground floor. Pay special attention to shooting key parts such as doors, windows, walls, balconies, etc., which are crucial for subsequent feature extraction and 3D reconstruction; select multiple shooting angles to ensure that each facade has enough photo coverage, and shoot from different heights and distances for subsequent multi-view geometry reconstruction; choose sunny weather to shoot, avoid direct strong light or heavy shadows, and ensure clear image quality.

[0089] Collect images - photos of the interior of the building: Use a handheld camera to shoot, and make sure the camera resolution is high enough to capture the details of the interior of the building; shoot the main spaces inside the building, including rooms, corridors, stairs, elevators, etc. Pay special attention to shooting key locations such as air conditioning equipment, lighting equipment, windows, doors, etc.; choose multiple shooting angles to ensure that each space has enough photos to cover. It is recommended to shoot from different heights and positions to facilitate subsequent feature extraction and 3D reconstruction; choose a time period with sufficient natural light to shoot, avoid using flash, and ensure natural image quality.

[0090] As an example, image data preprocessing can be achieved in the following way:

[0091] First, use image enhancement technology to improve the quality of the image, making it more suitable for subsequent feature extraction and processing. Image enhancement technology mainly includes contrast adjustment and denoising. Contrast adjustment can be achieved through histogram equalization and gamma correction. Histogram equalization adjusts the grayscale histogram of the image to make the brightness distribution of the image more uniform. This can be achieved using the `cv2.equalizeHist` function in the OpenCV library. Gamma correction changes the brightness and contrast of the image by adjusting the gamma value of the image. When the gamma value is greater than 1, the image becomes darker; when the gamma value is less than 1, the image becomes brighter. Gamma correction can be achieved using the `cv2.LUT` function in the OpenCV library. Denoising technology is used to reduce noise in the image and improve the clarity of the image. Common denoising methods include Gaussian filtering and median filtering. Gaussian filtering convolves the image with a Gaussian kernel to smooth the noise in the image. Gaussian filtering can be achieved using the `cv2.GaussianBlur` function in the OpenCV library. Median filtering replaces the pixel values ​​in the image through a median filter to remove salt and pepper noise. Median filtering can be implemented using the `cv2.medianBlur` function in the OpenCV library.

[0092] Next, the geometric distortion in the image is corrected by geometric correction technology to ensure that the straight lines and angles of the image are correct. Geometric correction technology mainly includes perspective transformation and distortion correction. Perspective transformation is used to correct the perspective distortion in the image so that the straight lines in the image remain straight. The specific steps are to first find four corner points in the image manually or automatically. These corner points are used to define the perspective transformation area of ​​the image. The `cv2.findContours` function in the OpenCV library can automatically detect the corner points. Then the perspective transformation matrix is ​​calculated based on the four corner points. The `cv2.getPerspectiveTransform` function in the OpenCV library can be used to calculate the transformation matrix. Finally, the perspective transformation matrix is ​​applied to the image to generate a corrected image. The `cv2.warpPerspective` function in the OpenCV library can be used to implement perspective transformation. Distortion correction is used to correct the distortion caused by the lens so that the straight lines in the image remain straight. The specific steps are to obtain the distortion parameters through camera calibration, including radial distortion and tangential distortion. The `cv2.calibrateCamera` function in the OpenCV library can be used for camera calibration. Then use the distortion parameters to correct the image and eliminate the distortion. The `cv2.undistort` function in the OpenCV library can be used to achieve distortion correction.

[0093] Finally, annotate the image and mark the key parts of the building, such as doors, windows, walls, etc., for subsequent feature extraction and semantic segmentation. Select the labeling tool LabelMe, mark all the door and window locations in the image, and use rectangular boxes or polygonal boxes for labeling. Mark the boundaries of the wall and use polygonal boxes for labeling. If needed, you can also annotate other key parts, such as balconies, stairs, etc. Save the labeling results in a standard format, JSON file, for subsequent feature extraction and semantic segmentation. Use the export function provided by the labeling tool to easily save the labeling results.

[0094] S102: Collection, preprocessing and extraction of time series features of electricity consumption data. This includes: first, using the existing energy consumption system or smart meter to collect the building's electricity consumption data for the past year, including air conditioning electricity consumption data, lighting electricity consumption data and equipment electricity consumption data; organizing the collected data into electricity consumption data and storing it in a cloud server or local database; then, using the Z-score method or the isolation forest algorithm to remove outliers in the electricity consumption data, and using mean filling or interpolation to process missing values ​​in the electricity consumption data. Finally, based on the electricity consumption data, calculate the average daily electricity consumption, seasonal changes in electricity consumption and peak load, and generate time series features.

[0095] As an example, the collection of electricity consumption data can be achieved in the following ways:

[0096] First, if the building is equipped with an energy management system, detailed electricity consumption data for one year can be directly exported from the system. This data should include but is not limited to the electricity consumption of air conditioning, lighting, equipment, etc. Ensure that the exported data format is standardized and the content is complete for subsequent data processing and analysis.

[0097] If the building does not have an energy management system, it is recommended to replace it with a smart meter to ensure that electricity consumption data can be recorded and transmitted in real time. Smart meters should have a sub-item metering function and be able to record the electricity consumption of each part such as air conditioning, lighting, and equipment separately. Collect electricity consumption data regularly through the communication interface of the smart meter (such as Wi-Fi, Bluetooth, wired connection, etc.). It is recommended to collect data once a day or every hour to ensure the continuity and integrity of the data. Store the collected data in a cloud server or local database to ensure the security and accessibility of the data.

[0098] If it is not possible to install a smart meter, you can collect electricity usage data manually. Use a paper record sheet or spreadsheet to record electricity usage data. The record sheet should include information such as date, time, electricity usage, and power-consuming equipment. It is recommended to record data every day or every few hours to ensure the continuity and integrity of the data. For important equipment and high-energy-consuming equipment, the recording frequency can be increased to more accurately reflect their electricity usage. Enter the manually recorded data into a spreadsheet or database in a timely manner to facilitate subsequent data processing and analysis.

[0099] As an example, the preprocessing and feature extraction of electricity consumption data can be achieved in the following ways:

[0100] The preprocessing of electricity consumption data is a key step to ensure the accuracy and reliability of the subsequent cooling and heating load calculation model. The preprocessing mainly includes two steps: data cleaning and feature extraction. The specific steps are as follows:

[0101] First, data cleaning refers to removing outliers and missing values ​​to ensure the accuracy and completeness of the data. Outliers refer to values ​​in the data that are significantly deviated from the normal range. These values ​​may be caused by equipment failure, recording errors, etc., which will affect the accuracy of subsequent analysis. Methods for removing outliers include statistical methods and machine learning methods. In statistical methods, the Z-score method can be used to detect outliers. The specific steps are to calculate the Z-score value of each data point. Data points with a Z-score value greater than 3 or less than -3 are considered outliers and removed from the data set. In machine learning methods, the Isolation Forest algorithm can be used to detect outliers. Isolation Forest isolates data points by constructing multiple decision trees. Outliers usually require fewer splits to be isolated. The specific steps are to use Python's `sklearn.ensemble.IsolationForest` class to train the Isolation Forest model, predict the outlier labels of data points, and data points with a label of -1 are considered outliers and removed from the data set.

[0102] There are two methods for dealing with missing values: deletion and filling. Deletion is to directly delete the data rows or columns containing missing values. This method is simple and direct, but it may lead to a reduction in the amount of data and affect the representativeness of the analysis results. Filling is to fill missing values ​​with reasonable values. Common filling methods include mean filling, median filling, interpolation, and forward or backward filling. Mean filling is to fill missing values ​​with the mean of the same column of data; median filling is to fill missing values ​​with the median of the same column of data; interpolation is to fill missing values ​​with methods such as linear interpolation or polynomial interpolation; forward or backward filling is to fill missing values ​​with the value of the previous or next data point.

[0103] Next, feature extraction is to extract useful time series features from the cleaned data, which will be used in the subsequent cooling and heating load calculation model. Daily average electricity consumption refers to the average electricity consumption per day, which can reflect the daily electricity consumption pattern of the building. The specific steps are to group the data by day, calculate the total electricity consumption per day, and then find the average. Seasonal changes refer to the changing trend of electricity consumption with the seasons, which can reflect the electricity consumption pattern of the building in different seasons. The specific steps are to group the data by month or quarter, calculate the total electricity consumption for each month or quarter, then find the average, and draw a trend chart of seasonal changes to observe the changing trend of electricity consumption with the seasons. Peak load refers to the maximum electricity consumption per day or month, which can reflect the peak electricity consumption period of the building. The specific steps are to group the data by day or month, calculate the maximum electricity consumption per day or month, and draw a change chart of the peak load to observe the distribution of peak electricity consumption periods.

[0104] In an embodiment of the present invention, based on step S2, a possible embodiment is given below to illustrate its specific implementation scheme in a non-limiting manner.

[0105] refer to Figure 2 As shown, this embodiment discloses an image feature extraction method, which specifically includes the following steps:

[0106] S201: Based on the image data, the Canny algorithm is used to detect the edge information of the image through Gaussian filtering, gradient calculation, non-maximum suppression and double threshold detection in sequence.

[0107] As an example, edge detection is an important step in image processing, which is used to extract the outline information of buildings. The Canny edge detection algorithm is used, which can accurately detect the edges in the image through multi-level filtering and non-maximum suppression technology. The edge detection results provide a basis for subsequent feature point recognition and semantic segmentation. The specific steps are as follows:

[0108] 1. Gaussian filtering: To reduce the noise in the image, first perform a Gaussian filter on the image. Gaussian filtering convolves the image with a Gaussian kernel to smooth out the noise in the image. The specific steps are to select an appropriate Gaussian kernel size and standard deviation, and use the `cv2.GaussianBlur` function in the OpenCV library to perform a Gaussian filter on the image. You can use a 5x5 Gaussian kernel and a Gaussian filter with a standard deviation of 1.0.

[0109] 2. Gradient calculation: Calculate the gradient magnitude and direction of the image. The Canny algorithm uses the Sobel operator to calculate the gradient of the image in the horizontal and vertical directions respectively. The specific steps are to use the `cv2.Sobel` function in the OpenCV library to calculate the gradient in the horizontal and vertical directions, and then calculate the gradient magnitude through the Pythagorean theorem and calculate the gradient direction through the inverse tangent function.

[0110] 3. Non-maximum suppression: The non-maximum suppression (NMS) technique is used to remove pixels that are not the local maximum in the gradient direction. The specific steps are to traverse each pixel and check its neighboring pixels in the gradient direction. If the gradient amplitude of the current pixel is not the local maximum, it is set to 0. This can retain the local maximum on the edge and remove other non-edge points.

[0111] 4. Double Threshold Detection: Determine the strength of the edge through the double threshold detection (Double Thresholding) technology. The specific steps are to set two thresholds, a high threshold and a low threshold. Pixels with gradient amplitudes greater than the high threshold are considered strong edge points, and pixels with gradient amplitudes less than the low threshold are discarded. Pixels between the two are determined to be edge points through the edge connection algorithm (Hysteresis Thresholding). The specific steps are to use the `cv2.Canny` function in the OpenCV library to implement double threshold detection.

[0112] Through the above steps, Gaussian filtering reduces the noise in the image, gradient calculation provides the direction and strength information of the edge, non-maximum suppression retains the local maximum on the edge, and dual threshold detection determines the strength of the edge. These steps together ensure the accuracy and reliability of edge detection and provide high-quality edge information for subsequent image processing and analysis.

[0113] S202: Using the SIFT algorithm, a scale space and a differential Gaussian pyramid are constructed according to the image data, extreme points are detected and descriptors are calculated, key feature points of the image data are identified, and the positions and descriptors of the key feature points are determined.

[0114] As an example, feature point recognition is an important means to extract key features of buildings. Use the SIFT (Scale Invariant Feature Transform) feature extraction algorithm to identify key feature points in the image. These feature points include door and window locations, wall thickness, etc. The specific steps are as follows:

[0115] 1. Scale space construction: The scale space is constructed by Gaussian pyramid, and the images are convolved using Gaussian kernels of different scales to generate a series of images of different scales. These images constitute the scale space.

[0116] 2. Construction of differential Gaussian pyramid: The differential Gaussian pyramid is generated by subtracting Gaussian images of adjacent scales, and the differential images of each scale are calculated. These images are used to detect extreme points.

[0117] 3. Extreme point detection: Detect extreme points in the differential Gaussian pyramid. In the differential image at each scale, check each pixel and its neighborhood pixels. If the pixel is a maximum or minimum value in the neighborhood, it is considered an extreme point.

[0118] 4. Key point descriptor calculation: Calculate the descriptor technology for each detected key point, calculate the gradient direction histogram around the key point, and generate a descriptor vector.

[0119] Through the above steps, scale space construction and differential Gaussian pyramid construction ensure the scale invariance of feature points, extreme point detection and key point descriptor calculation ensure the robustness and uniqueness of feature points. These together ensure the accuracy and reliability of feature point recognition and provide high-quality feature point information for subsequent image processing and analysis.

[0120] S203: Use the Mask R-CNN model to extract high-level features of image data and generate regions through convolutional neural networks and region proposal networks, classify and regress bounding boxes for each candidate region, and generate segmentation masks for each candidate region.

[0121] As an example, the deep learning model Mask R-CNN is used for semantic segmentation. This model can accurately identify and segment different parts in an image through convolutional neural network (CNN) and region proposal network (RPN) technology.

[0122] First, ensure that data preparation and data preprocessing have been completed. On this basis, use a pre-trained convolutional neural network (such as ResNet50) as the backbone network to extract high-level features of the image. The region proposal network (RPN) is responsible for generating candidate regions, which are regions that may contain objects. RPN generates multiple candidate regions on the feature map by sliding windows, and generates an object score and bounding box regression value for each candidate region. For each candidate region, the RoI Pooling layer is used to map it to a fixed-size feature map to ensure that candidate regions of different sizes can be processed uniformly for subsequent classification and segmentation. Each candidate region is classified and bounding box regressed through a fully connected layer. The classification layer predicts which category the candidate region belongs to, and the regression layer adjusts the bounding box position of the candidate region to make it surround the object more accurately. In Mask R-CNN, a segmentation branch is added to generate a segmentation mask for each candidate region through a small fully convolutional network (FCN). Each mask corresponds to a candidate region, indicating which category each pixel in the region belongs to. The segmentation branch runs in parallel with the classification and regression branches, sharing the feature map of the backbone network.

[0123] The trained Mask R-CNN model can be applied to new images to generate detailed semantic segmentation results. The specific process is that the image to be processed is input into the model, the model extracts high-level features of the image through the backbone network, and then RPN generates candidate regions. Next, each candidate region is classified and bounding box regressed, and the segmentation branch generates a segmentation mask for each candidate region. Finally, the segmentation results are visualized to display the category label of each pixel. The semantic segmentation results provide detailed building information for subsequent 3D reconstruction and BIM model generation.

[0124] In an embodiment of the present invention, based on step S3, a possible embodiment is given below to illustrate its specific implementation scheme in a non-limiting manner.

[0125] refer to Figure 3 As shown, this embodiment discloses a 3D reconstruction and BIM model generation method, which specifically includes the following steps:

[0126] S301: Based on the edge information of the image, the location and descriptor of the key feature points, and the segmentation mask of each candidate area, a 3D point cloud is constructed by matching the feature points under different viewpoints using camera calibration parameters and triangulation method, and the 3D point cloud is converted into a 3D model through a surface reconstruction method.

[0127] As an example, this step uses the multi-view geometry method to match feature points under different viewing angles to construct a 3D point cloud and then generate a 3D model of the building. Specifically, it includes:

[0128] First of all, the core of multi-view geometry technology is to extract and match feature points from multiple images from different perspectives, construct a three-dimensional point cloud through these feature points, and finally generate a three-dimensional model of the building.

[0129] After the feature point matching is completed, the camera's intrinsic parameters (such as focal length, principal point coordinates) and extrinsic parameters (such as rotation matrix, translation vector) are obtained through camera calibration. Camera calibration can be performed using a checkerboard calibration plate and implemented through tools such as OpenCV. Using the matched feature points and camera parameters, the three-dimensional coordinates of each feature point are calculated through triangulation to construct a three-dimensional point cloud. The triangulation method calculates the spatial position of the feature point through the feature point position and camera parameters under two or more perspectives. It is converted into a three-dimensional model through surface reconstruction methods (Poisson surface reconstruction, Delaunay triangulation). These methods generate continuous three-dimensional surfaces by fitting point cloud data, providing a basis for subsequent BIM model generation.

[0130] S302: Convert the 3D model into BIM format, and use Revit software to mark the wall materials, window types, and equipment locations to generate a BIM model.

[0131] As an example, the 3D reconstructed model is converted into BIM format, and the professional software Revit is used to annotate the model and mark various properties of the building, such as wall materials, window types, equipment locations, etc.

[0132] Specific annotation contents include wall materials, window types, equipment locations, etc. For example, mark the wall materials, such as concrete, masonry, glass, etc.; mark the window types, such as single-layer glass, double-layer glass, blinds, etc.; mark the locations of various equipment in the building, such as air conditioners, lighting equipment, fire-fighting equipment, etc.

[0133] It should be noted that the properties of each component can be edited in the BIM software to ensure the integrity and accuracy of the information. Property editing includes physical properties and functional properties. Physical properties include the size, weight, thermal conductivity, etc. of the component; functional properties include the function of the room, the power of the equipment, etc.; maintenance information includes the maintenance cycle and warranty period of the equipment, etc. The generated BIM model is verified to ensure the accuracy of the model's geometric structure and property information.

[0134] In an embodiment of the present invention, based on step S4, a possible embodiment is given below to illustrate its specific implementation scheme in a non-limiting manner.

[0135] Solar radiation and internal heat sources are calculated as follows:

[0136] According to the orientation and geographical location of the building, the solar radiation intensity on the building surface is calculated using the solar radiation calculation model PySolar library. The calculation formula is:

[0137] I = Ibcos(θ) + Id(1+cos(β)) / 2

[0138] Where I is the total solar radiation intensity, Ib is the direct radiation intensity, Id is the diffuse radiation intensity, θ is the incident angle, and β is the inclination angle of the building surface.

[0139] Internal heat sources include heat generated by equipment inside the building, human activities, etc. The contribution of internal heat sources is calculated based on electricity consumption data and the power of equipment inside the building. The calculation results of internal heat sources are used in the cooling and heating load calculation model to improve the calculation accuracy. The calculation formula is:

[0140] Q=P×t

[0141] Where Q is the heat generated, P is the power of the device, and t is the operating time of the device.

[0142] The heat generated by human activities also needs to be considered. Based on the usage of the building, the number of people and the intensity of activity can be estimated, and then the heat generated by people can be calculated. The calculation formula is:

[0143] Qp=n×q

[0144] Where Qp is the heat generated by the person, n is the number of people, and q is the heat generated per person per hour (usually around 100W).

[0145] Furthermore, random forest and neural network algorithms are used to generate and train the cold and hot load prediction model in combination with image features and electricity consumption data, and cross-validation and grid search techniques are used to optimize the parameters of the cold and hot load prediction model.

[0146] By combining image features and electricity consumption data through machine learning algorithms, the cooling and heating loads of buildings under different conditions can be accurately predicted. The following are the specific model training methods and technical routes:

[0147] First, two main machine learning algorithms were used: random forest and neural network. Random forest improves the generalization and robustness of the model by integrating multiple decision trees. Each decision tree is trained based on a different subset of features, and the final result is obtained by voting. This method can effectively handle high-dimensional data and reduce the risk of overfitting. The neural network uses a convolutional neural network (CNN) to capture complex relationships in the data through multiple layers of nonlinear transformations. Neural networks have strong expressive power and can handle complex input data and extract deep features.

[0148] During the model training process, the cross-validation technique is used to evaluate the performance of the model. Specifically, the K-fold cross-validation method is used to divide the data set into K subsets. Each time, K-1 subsets are used as training sets, and the remaining subset is used as a validation set. This process is repeated K times to ensure that each subset is used as a validation set once. In this way, the performance of the model on different data sets can be comprehensively evaluated and the model with the best performance can be selected. Cross-validation not only improves the generalization ability of the model, but also reduces the deviation caused by data partitioning.

[0149] To further optimize the model parameters, a grid search technique is used. Grid search exhaustively tries all possible parameter combinations, evaluates the performance of each combination, and finally selects the parameter combination that optimizes the model performance. The specific steps include defining the parameter range, performing a grid search, evaluating the performance of each parameter combination, and selecting the best parameter combination. For example, for a random forest, you can adjust parameters such as the number of trees and the maximum depth; for a neural network, you can adjust the number of hidden layers, the number of neurons in each layer, the type of activation function, etc. Grid search ensures the optimal configuration of model parameters, thereby improving the prediction accuracy of the model.

[0150] After the model training is completed, the model is evaluated using performance indicators. Common performance indicators include mean square error (MSE), mean absolute error (MAE), etc. These indicators can quantify the prediction error of the model and help select the model with the best performance. After the evaluation is completed, the trained model is saved for subsequent use.

[0151] Through the above method, combined with the public library data of heat transfer coefficients of various building insulation materials, a high-performance cold and hot load prediction model can be trained. The model can not only accurately predict the cold and hot load of the building under different conditions, but also provide optimization suggestions to help managers formulate reasonable energy management strategies.

[0152] like Figure 4As shown, the following is an embodiment of the building cooling and heating load adaptive calculation system provided by the embodiment of the present disclosure. The system and the building cooling and heating load adaptive calculation method of the above-mentioned embodiments belong to the same inventive concept. For details not described in detail in the embodiment of the building cooling and heating load adaptive calculation system, reference can be made to the embodiment of the above-mentioned building cooling and heating load adaptive calculation method.

[0153] A building cooling and heating load adaptive calculation system comprises: a data collection and preprocessing module 1, an image feature extraction module 2, a three-dimensional reconstruction and BIM model generation module 3, a model construction module 4 and a prediction module 5.

[0154] The data collection and preprocessing module 1 is used to collect the image data and electricity consumption data of the building, and after preprocessing, extract the time series features from the electricity consumption data.

[0155] The image feature extraction module 2 is used to extract image features from image data using edge detection, feature point recognition and semantic segmentation technology.

[0156] The 3D reconstruction and BIM model generation module 3 is used to generate a 3D point cloud using the extracted image features through multi-view geometry technology and surface reconstruction methods, and convert the 3D model into a BIM model using Revit software.

[0157] Model building module 4 is used to calculate solar radiation and internal heat sources based on electricity consumption data and the power of equipment inside the building, combined with the PySolar library, and train the cooling and heating load prediction model using random forest and neural network algorithms.

[0158] The prediction module 5 is used to dynamically adjust and optimize the cooling and heating load prediction model by using online learning and grid search technology, input real-time power consumption data and environmental parameters into the cooling and heating load prediction model, and generate the cooling and heating load prediction results of the building.

[0159] The building cooling and heating load adaptive calculation system provided in this embodiment realizes accurate prediction and dynamic optimization of building cooling and heating loads through data fusion, integration of three-dimensional models and BIM technology, accurate calculation of solar radiation and internal heat sources, advanced prediction models and optimization technology, and real-time prediction and feedback optimization cycles. This not only improves the efficiency and accuracy of building energy consumption management, but also provides strong support for the sustainable development and energy conservation and emission reduction of buildings.

[0160] Figure 5 A schematic diagram of the hardware structure of an electronic device for implementing various embodiments of the present invention.

[0161] The adaptive calculation method for building cooling and heating loads provided in the embodiment of the present application can be applied to electronic devices. Those skilled in the art will appreciate that the electronic device structure involved in the embodiment of the present invention does not constitute a limitation on the electronic device, and the electronic device may include more or fewer components than shown, or combine certain components, or arrange components differently. In the embodiment of the present invention, the electronic device includes but is not limited to a laptop computer, a desktop computer, a workbench, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of the present application described and / or required herein.

[0162] The electronic device may include a processor, an external memory interface, an internal memory, a universal serial bus (USB) interface, a charging management module, a power management module, a battery, a wireless communication module, an audio module, a speaker, a microphone, a sensor module, buttons, a camera, a display, and a SIM card interface, etc.

[0163] The processor may include one or more processing units, for example, the processor may include a central processing unit (CPU), an application processor (AP), a modem processor, a graphics processing unit (GPU), an image signal processor (ISP), a controller, a memory, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural-network processing unit (NPU), etc. Among them, different processing units may be independent devices or integrated into one or more processors.

[0164] The processor can be the nerve center and command center of the electronic device. The controller can generate an operation control signal according to the instruction operation code and timing signal to complete the control of fetching and executing instructions.

[0165] A memory may also be provided in the processor for storing instructions and data. In some embodiments, the memory in the processor is a cache memory. The memory may store instructions or data that the processor has just used or is cyclically used. If the processor needs to use the instruction or data again, it may be directly called from the memory. This avoids repeated access, reduces the waiting time of the processor, and thus improves system efficiency.

[0166] The external memory interface can be used to connect an external memory card, such as a MicroSD card, to expand the storage capacity of the electronic device. The external memory card communicates with the processor through the external memory interface to implement data storage functions. For example, files such as music and videos can be saved in the external memory card.

[0167] The internal memory can be used to store computer executable program codes, which include instructions. The processor executes various functional applications and data processing of the electronic device by running the instructions stored in the internal memory. The internal memory may include a program storage area and a data storage area. The internal memory may include a high-speed random access memory and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, a universal flash storage (UFS), etc.

[0168] The wireless communication function of an electronic device can be realized through an antenna, a wireless communication module, a modem processor, and a baseband processor.

[0169] The wireless communication module can provide wireless communication solutions for electronic devices, including wireless local area networks (WLAN) (such as wireless fidelity (Wi-Fi) networks), Bluetooth (BT), global navigation satellite system (GNSS), frequency modulation (FM), near field communication (NFC), infrared technology (IR), etc.

[0170] Electronic devices can implement audio functions, etc. through audio modules, speakers, receivers, microphones, headphone jacks, and application processors.

[0171] Electronic devices can achieve shooting functions through ISP, camera, video codec, GPU, display and application processor.

[0172] Electronic devices can achieve display functions through GPU, display screen and application processor.

[0173] The GPU is a microprocessor for image processing that connects the display screen and the application processor. The GPU is used to perform mathematical and geometric calculations for graphics rendering. The processor may include one or more GPUs that execute program instructions to generate or change display information.

[0174] The display screen is used to display images, videos, etc. The display screen includes a display panel.

[0175] The above-mentioned electronic device realizes the adaptive calculation method of building cooling and heating loads of the present application by collecting and preprocessing the image and electricity consumption data of the building, extracting time series features and image features, and then generating a three-dimensional BIM model. Based on the electricity consumption data, equipment power, solar radiation and other environmental parameters, the machine learning algorithm is used to train and optimize the cooling and heating load prediction model, and finally realize the technical solution of real-time prediction of the cooling and heating loads of the building, achieving the beneficial effect of significantly improving the accuracy and real-time performance of the cooling and heating load prediction of the building.

[0176] The storage medium provided in the present application stores a program product that can implement a method for adaptively calculating cooling and heating loads of a building.

[0177] The adaptive calculation method for building cooling and heating loads includes: collecting image data and electricity consumption data of the building, extracting time series features from the electricity consumption data after preprocessing; extracting image features from the image data using edge detection, feature point recognition and semantic segmentation technology; generating a three-dimensional point cloud using the extracted image features through multi-view geometry technology and surface reconstruction methods, and converting the three-dimensional model into a BIM model using Revit software; calculating solar radiation and internal heat sources based on electricity consumption data and the power of internal equipment in the building, combining the PySolar library, and training the cooling and heating load prediction model using random forest and neural network algorithms; using online learning and grid search technology to dynamically adjust and optimize the cooling and heating load prediction model, inputting real-time electricity consumption data and environmental parameters into the cooling and heating load prediction model, and generating the building cooling and heating load prediction results.

[0178] In some possible implementations, the adaptive calculation method for building cooling and heating loads disclosed herein can be implemented in the form of a program product, which includes a program code. When the program product is run on a terminal device, the program code is used to enable the terminal device to execute the steps described in the above “Exemplary Method” section of this specification according to various exemplary implementations of the present disclosure.

[0179] The storage medium of the present disclosure can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can be, for example, but not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0180] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for adaptively calculating building cooling and heating loads, characterized in that: include: Collect image data and electricity consumption data of buildings, and extract time series features from the electricity consumption data after preprocessing; Extract image features from image data using edge detection, feature point recognition and semantic segmentation techniques; The extracted image features are used to generate a 3D point cloud through multi-view geometry technology and surface reconstruction methods, and the 3D model is converted into a BIM model using Revit software; Based on the electricity consumption data and the power of the equipment inside the building, the PySolar library is used to calculate the solar radiation and internal heat source, and the random forest and neural network algorithms are used to train the cooling and heating load prediction model; Adopt online learning and grid search technology to dynamically adjust and optimize the cooling and heating load prediction model, input real-time power consumption data and environmental parameters into the cooling and heating load prediction model, and generate the cooling and heating load prediction results of the building; The image data and electricity consumption data of the building are collected, and after preprocessing, time series features are extracted from the electricity consumption data, including: Use drones and handheld cameras to collect photos of key parts of the building's interior and exterior from multiple angles to generate image data; Use the OpenCV library to perform image enhancement and geometric correction on the image data, and use the LabelMe tool to use rectangular boxes or polygonal boxes to mark key parts in the image; Save the annotation results as a JSON file; The method of extracting image features from image data by using edge detection, feature point recognition and semantic segmentation technology includes: According to the image data, the Canny algorithm is used to detect the edge information of the image through Gaussian filtering, gradient calculation, non-maximum suppression and double threshold detection in sequence; The SIFT algorithm is used to construct the scale space and differential Gaussian pyramid based on the image data, detect extreme points and calculate descriptors, identify the key feature points of the image data, and determine the location and descriptors of the key feature points; Use the Mask R-CNN model to extract high-level features of image data and generate regions through convolutional neural networks and region proposal networks, classify and regress bounding boxes for each candidate region, and generate segmentation masks for each candidate region; The method uses multi-view geometry technology and surface reconstruction method to generate a 3D point cloud using the extracted image features, and uses Revit software to convert the 3D model into a BIM model, including: Based on the edge information of the image, the location and descriptor of the key feature points, and the segmentation mask of each candidate area, a 3D point cloud is constructed by matching the feature points under different viewpoints using camera calibration parameters and triangulation, and the 3D point cloud is converted into a 3D model through surface reconstruction method; Convert the 3D model into BIM format, and use Revit software to mark wall materials, window types, and equipment locations to generate a BIM model.

2. The building cooling and heating load adaptive calculation method according to claim 1 is characterized in that: The collecting of image data and electricity consumption data of the building, and extracting time series features from the electricity consumption data after preprocessing, also includes: Use existing energy consumption systems or smart meters to collect the building's electricity consumption data for the past year, including air conditioning power consumption data, lighting power consumption data, and equipment power consumption data; organize the collected data into power consumption data and store it in a cloud server or local database; Use the Z-score method or the isolation forest algorithm to remove outliers in the electricity consumption data, and use mean filling or interpolation to handle missing values ​​in the electricity consumption data; Based on the electricity consumption data, the average daily electricity consumption, seasonal changes in electricity consumption and peak load are calculated to generate time series characteristics.

3. The building cooling and heating load adaptive calculation method according to claim 1 is characterized in that: The method is based on electricity consumption data and the power of equipment inside the building, combined with the PySolar library to calculate solar radiation and internal heat sources, and uses random forest and neural network algorithms to train the cold and hot load prediction model, including: According to the orientation and geographical location of the building, the PySolar library is used to calculate the solar radiation intensity I on the building surface according to the following formula: I = Ibcos(θ) + Id(1+cos(β)) / 2 Where, Ib is the direct radiation intensity, Id is the scattered radiation intensity, θ is the incident angle, and β is the inclination angle of the building surface; Based on the electricity consumption data and the power of the equipment inside the building, the heat Q generated by the internal heat source is calculated according to the formula Q=P×t, and the heat Qp generated by the personnel is calculated according to the formula Qp=n×q; Where P is the power of the equipment, t is the operating time of the equipment, n is the number of people, and q is the heat generated per person per hour; Random forest and neural network algorithms are used to combine image features and electricity consumption data to generate and train the cold and hot load prediction model, and cross-validation and grid search techniques are used to optimize the parameters of the cold and hot load prediction model.

4. The building cooling and heating load adaptive calculation method according to claim 3 is characterized in that: The online learning and grid search technology is used to dynamically adjust and optimize the cooling and heating load prediction model, and real-time power consumption data and environmental parameters are input into the cooling and heating load prediction model to generate the cooling and heating load prediction results of the building, including: Obtain the building's electricity consumption data and environmental parameters in real time and pre-process the data; The pre-processed electricity consumption data and environmental parameters are input into the cooling and heating load prediction model, and the model parameters are dynamically adjusted using incremental learning or online learning algorithms; Define a search space for the key parameters of the cooling and heating load prediction model. In the search space, use the grid search technique to try different parameter combinations, and use the cross-validation technique to evaluate the performance of each parameter combination to determine the optimal parameter combination as the final parameters of the cooling and heating load prediction model. Input the real-time power consumption data and environmental parameters into the cooling and heating load prediction model to generate the cooling and heating load prediction results of the building; Collect actual electricity consumption data from the building, perform error analysis based on the building cooling and heating load prediction results, and adjust the model parameters or structure based on the analysis results.

5. A building cooling and heating load adaptive calculation system, characterized in that: include: The data collection and preprocessing module is used to collect the image data and electricity consumption data of the building, and after preprocessing, extract the time series features from the electricity consumption data; An image feature extraction module is used to extract image features from image data using edge detection, feature point recognition and semantic segmentation techniques; 3D reconstruction and BIM model generation module, which is used to generate 3D point cloud using extracted image features through multi-view geometry technology and surface reconstruction method, and convert the 3D model into BIM model using Revit software; Model building module, which is used to calculate solar radiation and internal heat sources based on electricity consumption data and the power of equipment inside the building, combined with the PySolar library, and train the cooling and heating load prediction model using random forest and neural network algorithms; The prediction module is used to dynamically adjust and optimize the cooling and heating load prediction model using online learning and grid search technology, input real-time power consumption data and environmental parameters into the cooling and heating load prediction model, and generate the cooling and heating load prediction results of the building; The image data and electricity consumption data of the building are collected, and after preprocessing, time series features are extracted from the electricity consumption data, including: Use drones and handheld cameras to collect photos of key parts of the building's interior and exterior from multiple angles to generate image data; Use the OpenCV library to perform image enhancement and geometric correction on the image data, and use the LabelMe tool to use rectangular boxes or polygonal boxes to mark key parts in the image; The annotation results are saved as a JSON file; the image features are extracted from the image data using edge detection, feature point recognition and semantic segmentation technology, including: According to the image data, the Canny algorithm is used to detect the edge information of the image through Gaussian filtering, gradient calculation, non-maximum suppression and double threshold detection in sequence; The SIFT algorithm is used to construct the scale space and differential Gaussian pyramid based on the image data, detect extreme points and calculate descriptors, identify the key feature points of the image data, and determine the location and descriptors of the key feature points; Use the Mask R-CNN model to extract high-level features of image data and generate regions through convolutional neural networks and region proposal networks, classify and regress bounding boxes for each candidate region, and generate segmentation masks for each candidate region; The method uses multi-view geometry technology and surface reconstruction method to generate a 3D point cloud using the extracted image features, and uses Revit software to convert the 3D model into a BIM model, including: Based on the edge information of the image, the location and descriptor of the key feature points, and the segmentation mask of each candidate area, a 3D point cloud is constructed by matching the feature points under different viewpoints using camera calibration parameters and triangulation, and the 3D point cloud is converted into a 3D model through surface reconstruction method; Convert the 3D model into BIM format, and use Revit software to mark wall materials, window types, and equipment locations to generate a BIM model.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the building cooling and heating load adaptive calculation method as described in any one of claims 1 to 4 are implemented.

7. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the building cooling and heating load adaptive calculation method as claimed in any one of claims 1 to 4 are implemented.

Images