Optimization method for transforming old city into green building based on multi-source data fusion

Abstract

The invention relates to a green building reconstruction optimization method based on multi-source data fusion. According to the method, oblique photography, laser scanning, thermal infrared imaging and a sensor network are used for collecting building data, and the data quality is ensured through preprocessing of time synchronization, space registration, data calibration and the like. And a dynamic thermotechnical three-dimensional model is constructed, and model parameters are updated in real time. And a multi-objective optimization algorithm is adopted to generate a green reconstruction scheme, and EnergyPlus is utilized to perform simulation verification. And energy consumption and environmental parameters are monitored in real time and compared with model prediction, so that closed-loop iterative optimization is realized. The existing building can be transformed intelligently and finely, the energy-saving benefit and the environment comfort degree are improved, and good application prospects are achieved.

Description

Technical Field

[0001] This invention relates to the fields of green building renovation and digital twin technology, specifically a method for optimizing green buildings in urban renewal based on multi-source data fusion. Background Technology

[0002] In recent years, green building renovation technologies have developed rapidly in the field of urban renewal, with multi-source data fusion becoming a core means to improve the accuracy of energy efficiency assessment. Existing technologies mainly rely on the static integration of Building Information Modeling (BIM) and Geographic Information System (GIS), combined with IoT sensor networks (such as temperature, humidity, and energy consumption monitoring) to collect real-time data. Deep learning algorithms (such as U-Net semantic segmentation) are used for building structure identification, while multi-objective optimization (NSGA-II) supports the generation of energy-saving solutions. The introduction of digital twin technology has enabled solution simulation verification, and some systems adjust renovation strategies through feedback mechanisms. However, the data layer still relies mainly on homogeneous datasets, resulting in limited dynamic environmental response capabilities, and the renovation closed-loop largely depends on manual intervention.

[0003] Current technologies suffer from three major shortcomings: First, insufficient depth of multi-source data fusion: While existing methods integrate BIM, GIS, and sensor data, they fail to address spatiotemporal registration errors caused by heterogeneous data and measurement deviations due to equipment drift. This results in low model input reliability. Second, disconnect between optimization decision-making and implementation: Traditional optimization algorithms (such as static genetic algorithms) lack dynamic environmental coupling mechanisms after generating solutions. During sudden weather changes, the model cannot update heat transfer parameters in real time, leading to high deviations in measured energy consumption. Furthermore, solution verification relies on fixed-scene simulations, failing to incorporate the randomness of residents' behaviors (such as air conditioning usage frequency), resulting in insufficient social acceptance. Third, lack of dynamic triggering for closed-loop feedback: Existing feedback systems mostly rely on periodic manual inspections, unable to respond promptly to renovation failure events (such as insulation layer delamination). Even with automatic monitoring, only local adjustments are triggered (such as sensor calibration), without establishing a full-chain iterative mechanism (calibration → model → solution linkage). Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by this invention is: existing methods for green building renovation in old cities suffer from insufficient accuracy in multi-source heterogeneous data fusion, leading to model distortion; lagging dynamic response during sudden environmental changes, causing energy efficiency deviations; lack of a closed-loop iterative mechanism between renovation schemes and implementation effects; and the problem of how to achieve dynamic self-optimization throughout the entire process from data collection to scheme optimization.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a green building optimization method for urban renewal based on multi-source data fusion, comprising collecting building geometric structure data, thermal performance data and environmental monitoring data as multi-source datasets, preprocessing the multi-source datasets, and outputting standardized data after preprocessing; A dynamic 3D model with thermal properties is constructed by integrating standardized data, and the parameters of the dynamic 3D model are dynamically updated based on real-time environmental monitoring data. Based on a dynamic 3D model, a green transformation scheme is generated through multi-objective optimization, and digital twin technology is used for simulation verification. The verified green transformation scheme is then output. Based on the discrepancy between environmental monitoring data and dynamic 3D model predictions after the implementation of the verified green transformation plan, the parameters for data calibration are dynamically adjusted, the dynamic 3D model is updated, and the green transformation plan is iterated.

[0007] As a preferred embodiment of the green building optimization method for urban renewal based on multi-source data fusion described in this invention, the geometric structure data is collected using three-dimensional scanning technology, which includes oblique photogrammetry and lidar scanning. Thermal infrared imaging technology was used to collect thermal performance data and obtain the temperature distribution on the building's exterior surface. The environmental monitoring data is collected using a distributed sensor network, which includes temperature and humidity sensors, light intensity sensors, and gas concentration sensors deployed on the interior and exterior facades of the building. The preprocessing includes: time synchronization, spatial registration and noise filtering, as well as data calibration and normalization.

[0008] This invention presents a method for optimizing green buildings in urban renewal based on multi-source data fusion. In a preferred embodiment, the oblique photogrammetry employs an unmanned aerial vehicle (UAV) equipped with an oblique camera to acquire multi-angle images; The lidar scanning uses a ground-based three-dimensional laser scanner; The thermal infrared imaging technology uses a thermal imager; The distributed sensor network achieves synchronous data transmission through a wireless self-organizing network protocol. The time synchronization uses the NTP protocol to uniformly calibrate the timestamps of multi-source data; the spatial registration uses a feature point matching algorithm to spatially register the laser point cloud with the oblique photogrammetric image. The noise filtering employs a combined wavelet transform and median filtering algorithm to separate and filter out impulse noise and Gaussian noise in the sensor data.

[0009] As a preferred embodiment of the green building optimization method for urban renewal based on multi-source data fusion described in this invention, the dynamic three-dimensional model construction includes: dividing standardized data into building structural units based on a semantic segmentation algorithm, wherein the building structural units include walls, roofs, doors and windows; Each building structural unit is assigned thermal properties, including thermal conductivity, heat capacity, and solar radiation absorptivity. By fusing 3D scanning data and thermal infrared images, a dynamic 3D model with thermal property labels is generated; the updating of dynamic 3D model parameters includes: real-time access to a meteorological data interface to obtain hourly meteorological parameters of the target area; Based on the meteorological parameters, the thermal parameters of the three-dimensional model are dynamically adjusted. Couple building energy consumption monitoring data and update the equivalent thermal resistance coefficient in the model's heat conduction equation based on the backpropagation algorithm.

[0010] As a preferred embodiment of the green building optimization method for urban renewal based on multi-source data fusion described in this invention, the generation of the green renovation scheme includes: determining optimization objectives, including energy saving rate, carbon emission reduction rate, and economic efficiency; Identify decision variables, including the thickness of the external wall insulation layer, the window-to-wall ratio, window type, shading measures, and photovoltaic coverage. A multi-objective optimization model is established, with the optimization objective as the objective function and the decision variables as independent variables, to perform multi-objective optimization; The simulation verification includes: applying the generated green transformation plan to a dynamic three-dimensional model for simulation; evaluating the energy saving rate, carbon emission reduction rate and economic indicators, and determining whether the preset constraints are met. If the simulation results do not meet the constraints, the decision variables are adjusted, and the optimization and verification are repeated.

[0011] As a preferred embodiment of the green building optimization method for urban renewal based on multi-source data fusion described in this invention, the method includes: real-time monitoring of environmental monitoring data of the renovated building, including temperature, humidity, and energy consumption; comparison of the actual environmental monitoring data with the predicted values ​​of the dynamic three-dimensional model; calculation of the difference; setting a difference threshold; and triggering iterative optimization when the difference exceeds the threshold. The iterative optimization includes: adjusting the multi-source data calibration parameters and dynamic 3D model parameters according to the type of difference; re-optimizing and verifying the green renovation scheme to generate a new green renovation scheme; deploying the new green renovation scheme to the actual building and continuing to monitor and iterate.

[0012] A green building optimization system for urban renewal based on multi-source data fusion, wherein: The data acquisition and preprocessing module is used to collect building geometry data, thermal performance data, and environmental monitoring data, and to preprocess multi-source datasets to output standardized data. The dynamic 3D model construction and update module is used to integrate standardized data to construct a dynamic 3D model with thermal properties, and to dynamically update the parameters of the dynamic 3D model based on real-time environmental monitoring data. The scheme generation and verification module is used to generate green transformation schemes based on dynamic 3D models through multi-objective optimization, and to verify them using digital twin technology, outputting the verified green transformation schemes. The closed-loop iterative optimization module is used to monitor environmental data after the transformation is implemented. Based on the difference between the actual monitoring data and the predicted values ​​of the 3D model, it triggers the adjustment of data calibration parameters, the updating of the 3D model, and the iteration of the green transformation plan.

[0013] A computer device includes: a memory and a processor; the memory stores a computer program, and the processor executes the computer program to implement the steps of the method described in any one of the present invention.

[0014] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in any one of the present invention.

[0015] The beneficial effects of this invention are as follows: Through the full-process design of "data acquisition - model building - scheme optimization - iterative feedback", multi-source data fusion and digital twin technology are deeply integrated. This not only solves the pain points of data fragmentation, model staticization and scheme verification lag in old city renovation, but also ensures the scientific nature and sustainability of green renovation effects through dynamic update and iteration mechanisms. It provides a systematic and feasible technical path for the low-carbon and ecological renovation of old city buildings. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 The first embodiment of the present invention provides an overall flowchart of a green building optimization method for urban renewal based on multi-source data fusion. Detailed Implementation

[0018] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0019] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for optimizing green buildings in urban renewal based on multi-source data fusion is provided, comprising: S1: Collect building geometry data, thermal performance data, and environmental monitoring data as a multi-source dataset, preprocess the multi-source dataset, and output standardized data after preprocessing.

[0020] Geometric data was acquired using 3D scanning technology, which includes oblique photogrammetry and laser scanning. LiDAR scanning and oblique photogrammetry utilize UAVs equipped with oblique cameras to acquire multi-angle images with an oblique angle ≥45° and a flight overlap rate ≥80%. LiDAR scanning employs a ground-based 3D laser scanner, achieving a point cloud density ≥10,000 points / square meter and a spatial resolution ≤5cm. This combined approach achieves complete coverage of the macroscopic form of the building complex while ensuring accurate acquisition of microscopic details of building components (such as doors, windows, and pipe interfaces). Compared to single scanning technologies, data integrity is improved by more than 60%.

[0021] Thermal performance data is collected using thermal infrared imaging technology to obtain the temperature distribution on the building's exterior surface. The thermal infrared imaging technology employs a thermal imager with a radiation temperature measurement accuracy of ≤±0.5℃ and a spatial registration error of <3 pixels. This technology captures areas of abnormal temperature on the building surface (such as thermal bridges in walls and heat leakage through door and window gaps), providing a visual basis for subsequent insulation modifications.

[0022] Environmental monitoring data is collected using a distributed sensor network, which includes temperature and humidity sensors, light intensity sensors, and gas concentration sensors deployed on the interior and exterior facades of the building. The distributed sensor network achieves synchronous data transmission through wireless self-organizing network protocols (such as Zigbee and LoRa).

[0023] Preprocessing includes time synchronization, spatial registration, noise filtering, and data calibration and normalization.

[0024] Time synchronization uses the NTP protocol to uniformly calibrate the timestamps of multi-source data. Considering that the operating temperature of the device may affect the clock accuracy, the following formula is introduced to compensate for the timestamps:

[0025] in, For the calibrated synchronization timestamp, This is the original timestamp. This refers to the time difference between the server and the device, calculated using the NTPv4 protocol. The temperature (°C) is the operating temperature of the sensing device. This method introduces a temperature compensation term. This enables more accurate time synchronization in both high and low temperature environments, improving the time consistency of multi-source data.

[0026] Spatial registration employs a feature point matching algorithm to spatially register the laser point cloud with the oblique photogrammetric image, achieving a registration residual of <0.1m. Noise removal utilizes a joint algorithm of wavelet transform and median filtering to separate and remove impulse noise and Gaussian noise from the sensor data. An optimization-based spatial registration method is employed to minimize the matching error between the laser point cloud and the thermal infrared image in a unified 3D space, while also considering the robustness of the initial transformation.

[0027] in, Let the coordinates of the laser point cloud be... The points corresponding to the thermal infrared image. Let be the transformation matrix to be solved. The initial transformation matrix is ​​obtained based on UAV POS data, and 0.3 is the regularization coefficient. This optimization objective improves the accuracy and stability of registration, especially in regions where features are not obvious, by adding a penalty term to the initial transformation matrix.

[0028] The noise filtering uses a joint algorithm of wavelet transform (db4 wavelet basis) and median filtering (window width k=2) to filter out noise from sensor data, effectively separating and suppressing impulse noise and Gaussian noise.

[0029] Data calibration involves establishing a sensor calibration model based on the sensor's inherent characteristics and environmental factors to calibrate the sensor data. The calibrated multi-source data is then transformed into a unified spatiotemporal coordinate system and normalized to generate standardized data that can be fused. Sensor calibration models include: a temperature sensor calibration model based on multiple linear regression, considering the influence of temperature, humidity, and light intensity on temperature sensor measurements; and a gas concentration sensor calibration model based on a BP neural network, used to calibrate gas concentration sensors to address the cross-influence of temperature and humidity.

[0030] For distributed sensor network data, a federated learning approach is used for data calibration, aggregating the calibration model parameters of each node (device) without uploading the original data.

[0031]

[0032] in, For global calibration of model parameters, For the first Local calibration model parameters at nodes. For the first The weight of a node.

[0033] The weight calculation takes into account the amount of data and the variance of the data:

[0034] in, For the first Local calibration model parameters at nodes. For the first The variance of the node data, and the variance weights here ( The core point is that it can suppress the influence of nodes with large variance (i.e., more noise) on the global model, thereby obtaining a more robust and accurate global calibration model.

[0035] Temperature sensor calibration: A multiple linear regression model is used to consider the effects of temperature, humidity, and light intensity on the temperature sensor measurements.

[0036] in, The calibrated temperature value. This is the original temperature measurement value. Relative humidity, Light intensity, This is the regression constant term (intercept). These are the regression coefficients (weights) for the original temperature measurement, relative humidity, and light intensity, respectively.

[0037] Standardized data output: All calibrated multi-source data are unified to the WGS-84 coordinate system. Continuous variables are normalized using the following formula:

[0038] in, The normalized variable values, The original variable value, The mean of this variable. Let be the standard deviation of this variable. Through the above processing, standardized data that can be fused is generated.

[0039] S2: Integrate standardized data to construct a dynamic 3D model with thermal properties, and dynamically update the parameters of the dynamic 3D model based on real-time environmental monitoring data.

[0040] The construction of the dynamic 3D model includes: dividing standardized data into building structural units based on semantic segmentation algorithms, including walls, roofs, doors and windows; assigning thermal properties to each building structural unit, including thermal conductivity, heat capacity and solar radiation absorptivity; and fusing 3D scanning data and thermal infrared images to generate a dynamic 3D model with thermal property labels.

[0041] Construction of dynamic 3D models: The dynamic 3D model consists of the following core elements: geometric information, including the building's geometry, size, and location, acquired through 3D scanning data; thermal properties, including the thermal conductivity, heat capacity, and solar radiation absorptivity of the building's structural units, acquired through material databases and thermal infrared images; and temporal information, including meteorological data and energy consumption data, used to describe the model's changes over time.

[0042] The model's data structure adopts an object-oriented data structure, modeling building structural units (walls, roofs, doors, windows, etc.) as objects. Each object contains a description of its geometry and dimensions; a description of its thermal performance; a description of its surface temperature; and a record of the time of change in its properties.

[0043] This object-oriented data structure makes it easy to manage and update the model.

[0044] Building structural unit division: Based on the Mask R-CNN semantic segmentation algorithm, standardized data is divided into building structural units, including walls, roofs, doors and windows, providing a basis for subsequent thermal property assignment.

[0045] Dynamic Assignment of Thermal Properties: A dynamic thermal property assignment method that can adapt to material aging is adopted. The formula for calculating solar radiation absorptivity is as follows:

[0046] in, The effective solar radiation absorptivity is dynamically adjusted. The initial solar radiation absorptivity of the material (can be found in the material handbook or obtained through experimental measurement). The temperature gradient in the thermal infrared image reflects the degree of uneven heat distribution on the building surface. This represents the ambient temperature gradient, reflecting the temperature changes in the environment in which the building is located. The calibration coefficient is set to 0.15, and is determined through on-site measurements and data analysis of existing buildings. This method can solve the problem that traditional static assignment cannot adapt to material aging, thus improving the accuracy of the model.

[0047] 3D scan data and thermal infrared images are fused to generate a dynamic 3D model with thermal property labels. Using graphics libraries such as OpenGL, the dynamic 3D model is rendered into a 3D image, and information such as the model's temperature distribution is displayed in real time. Color mapping is used to map different temperature ranges to different colors, thus intuitively displaying the temperature distribution on the building surface.

[0048] Updating the dynamic 3D model parameters includes: real-time access to a meteorological data interface to obtain hourly meteorological parameters (temperature, humidity, wind speed, solar radiation) for the target area; dynamically adjusting the thermal parameters of the 3D model based on these meteorological parameters; and coupling building energy consumption monitoring data to update the equivalent thermal resistance coefficient in the model's heat conduction equation based on a backpropagation algorithm. ).

[0049] Adaptive Step Size Thermal Resistance Update: Introducing Adaptive Step Size Based on model prediction error Adjusting the update step size based on the magnitude of the parameter size improves the efficiency and stability of model parameter calibration.

[0050] The formula for updating the thermal resistivity is as follows:

[0051] in, This is the current thermal resistance coefficient. For the updated thermal resistivity, To adaptively update the step size, This is the partial derivative (i.e., gradient) of the prediction error with respect to the thermal resistance coefficient. This represents the model prediction error.

[0052] Formula for calculating prediction error:

[0053] in, The temperature value predicted by the model. This is the actual measured temperature value. Step size. The rules for determining the value: When When >5, set =0.1; when When ≤5, set =0.05. This mechanism allows the model to converge faster when the prediction error is large and to be finely adjusted when the error is small.

[0054] Meteorological abrupt change response: Establish a response mechanism for extreme weather or rapid changes, such as when the instantaneous wind speed increase is >40% or the temperature drops sharply by >8℃ / hour, the model can adjust parameters in real time to reflect dynamic changes.

[0055] S3: Based on a dynamic 3D model, a green transformation scheme is generated through multi-objective optimization, and digital twin technology is used for simulation verification. The verified green transformation scheme is then output.

[0056] Generating green retrofit plans includes: determining optimization objectives and decision variables. Based on building type and actual needs, optimization objectives (such as energy saving rate, carbon emission reduction rate, investment payback period, indoor comfort, etc.) and decision variables (such as external wall insulation layer thickness, window-to-wall ratio, window type, shading measures, photovoltaic coverage, etc.) are determined. A multi-objective optimization model is established, with the optimization objectives as the objective function and the decision variables as independent variables.

[0057] Creative Constraint Repair Strategy and Monte Carlo Verification: The constraint repair strategy addresses situations where the optimized solution fails to meet hard constraints such as economic efficiency. It employs linear interpolation or scaling to repair the decision variables, ensuring the feasibility of the solution. The repair formula is as follows:

[0058] in, These are the original decision variable values. For the corrected decision variable values, This is the upper limit of cost (e.g., 2000 yuan / ㎡). This is the lower limit of cost (e.g., 800 yuan / ㎡). This represents the total cost of the current solution. The strategy aims to proportionally adjust excess costs back into the budget while maintaining the relative relationships of other performance parameters as much as possible.

[0059] Monte Carlo Validation: To evaluate the robustness of the solution under uncertainties, this embodiment introduces Monte Carlo simulation technology and constructs a scenario-adaptive robustness evaluation framework. This framework, targeting urban redevelopment schemes, identifies and models key uncertainties, and achieves a more accurate robustness evaluation through a scenario-adaptive weight allocation mechanism and weighted statistical analysis. The specific implementation process is as follows: Uncertainty factor modeling: Mathematical modeling of key uncertainty factors affecting the performance of old city renovation plan: Resident behavior - randomness of air conditioner use: The number of times air conditioners are used per unit time follows the characteristics of Poisson distribution. By setting the average number of times air conditioners are turned on, the randomness of air conditioner use is simulated. Material performance degradation: The process of material performance degradation follows a normal distribution law. A mathematical model is constructed to describe it based on the average value and standard deviation of the degradation. Scene-Adaptive Weight Allocation: The core innovation of this embodiment lies in the scene-adaptive weight allocation mechanism, which can quantify the differences in the impact of various uncertainties on the solution performance. Defining scene weight coefficients. This is used to measure the impact of various scenarios on the objective function (such as energy saving rate). Taking energy saving rate as an example, its calculation formula is as follows:

[0060] in, For the first The weighting coefficients for each scenario. Indicates the first The change in energy saving rate of each scenario relative to the baseline scenario; This represents the total number of scenarios. Using this formula, the weight of each scenario will be dynamically adjusted based on its actual impact on energy efficiency.

[0061] Monte Carlo simulation steps Set simulation parameters: Determine the relevant parameters for the Poisson and normal distributions, and set the number of iterations. To ensure the reliability of the simulation results; Loop iteration: Execution The loop iterates through the following steps in each iteration: Random sampling: randomly selecting values ​​from a predefined distribution to simulate the actual changes in uncertain factors.

[0062] Modify model parameters: Update the relevant parameters in the dynamic 3D model based on the sampling results.

[0063] Conduct simulations: Perform simulation experiments on the modified model to obtain performance data under the corresponding scenarios.

[0064] Scene weighting: Based on the above scene weighting coefficient formula, calculate the weight of the current scene to provide a basis for subsequent statistical analysis. Weighted statistical analysis: for The simulation results are weighted statistically, and by calculating statistical measures such as the weighted average and weighted standard deviation of performance indicators, the robustness of the solution can be accurately evaluated. Taking the weighted average energy saving rate as an example, its calculation formula is as follows:

[0065] in, The weighted average energy saving rate; For the first Weighting coefficients for each scenario; For the first The energy-saving rate for each scenario is calculated using this method. This fully reflects the differences in the contribution of different scenarios to the final evaluation results, effectively improving the accuracy of robustness assessment.

[0066] Simulation Verification: The optimized and improved green retrofit plan is applied to a dynamic 3D model and simulated using building energy consumption simulation software (such as EnergyPlus). The results are evaluated to determine if the preset optimization objectives are met (e.g., energy saving rate ≥30%, carbon emission reduction rate ≥25%, investment payback period ≤8 years, etc.). If the simulation results do not meet the constraints, the decision variables are adjusted, and the optimization and verification are repeated.

[0067] S4: Based on the difference between the environmental monitoring data and the predicted values ​​of the dynamic 3D model after the implementation of the verified green transformation plan, dynamically trigger the parameter adjustment of data calibration, the update of the dynamic 3D model, and the iteration of the green transformation plan.

[0068] Real-time monitoring and difference calculation: After implementing green renovation, environmental monitoring data (temperature, humidity, energy consumption, etc.) of the building are monitored in real time. The measured data are compared with the predicted values ​​of the dynamic 3D model, the difference is calculated, and a difference threshold is set.

[0069] Difference threshold definition: Energy consumption deviation exceeding 5% of the predicted energy consumption for 24 hours.

[0070] Temperature deviation: The difference between the actual indoor temperature and the predicted temperature exceeds 2℃.

[0071] Three-link collaborative iterative optimization: Energy consumption anomaly trigger: When an energy consumption deviation is detected to exceed a threshold, it means that there is a systematic bias in the model's prediction of the building's actual energy consumption. At this time, the mechanism to update the S1 calibration parameters is triggered, possibly by recalibrating the sensor data or fine-tuning the sensor model parameters in S1 to improve the accuracy of data input.

[0072] Temperature Anomaly Trigger: When a temperature deviation exceeding a threshold is detected, it indicates that the model's simulation of the building's thermal behavior is inadequate. At this point, a mechanism to update the S2 model parameters is triggered, such as using the previously mentioned adaptive step-size thermal resistance update algorithm to calibrate key thermal parameters in the model (e.g., thermal resistance, heat transfer coefficient).

[0073] Solution Failure Trigger: If, after adjusting the parameters of S1 and S2, the model's prediction accuracy still fails to meet requirements, or the transformation effect is far below expectations, it may mean that the initial green transformation solution itself has fundamental flaws or has failed. In this case, the S3 optimization process will be restarted to regenerate and validate a new green transformation solution based on the latest model and data.

[0074] Cyclic execution: Deploy new (or optimized) green renovation solutions into actual buildings and continue to monitor and iterate, forming a closed loop of continuous optimization.

[0075] Example 2, an embodiment of the present invention, provides a green building optimization method for urban renewal based on multi-source data fusion. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiment.

[0076] This example selects a typical office building built in 1995 in downtown Shanghai as the subject of renovation. The building has a concrete exterior, ordinary single-pane windows, an outdated air conditioning system, and high overall energy consumption. The renovation aims to achieve the following objectives: significantly reduce building operating energy consumption, reducing annual comprehensive energy consumption by more than 30%; improve indoor environmental comfort; reduce carbon emissions; and ensure that the investment payback period does not exceed 8 years. First, data acquisition and preprocessing were performed. High-resolution images of the building's exterior were acquired using a DJI Phantom 4 Pro drone. Precise 3D point cloud data was collected using a FARO Focus S70 terrestrial 3D laser scanner for 3D building modeling. Temperature distribution data of the building facade was acquired using a FLIR T650sc thermal imager. Temperature and humidity sensors, light intensity sensors, and CO2 concentration sensors were installed on typical floors and the inner side of the exterior walls to construct a sensor network, which was then transmitted to the server via the LoRa wireless ad hoc network protocol. The timestamps of all sensor and image data were calibrated using the NTP protocol, and after temperature compensation, the timestamp accuracy was improved to within 0.05 seconds. Spatial registration was performed between the oblique photogrammetric images and the laser point cloud based on the SIFT feature matching algorithm, controlling the registration residual to within 0.1 meters. A joint algorithm combining wavelet transform and median filtering was used to filter noise from the sensor data. A calibration model for the temperature sensor was established based on multiple linear regression, comprehensively considering the influence of temperature, humidity, and light intensity on the measured values. After calibration, the measurement accuracy of the temperature sensor was improved from ±0.5℃ to ±0.2℃. A BP neural network was used to calibrate the gas concentration sensor to address the cross-influence of temperature and humidity, and federated learning was employed for data calibration. This improved the measurement accuracy of the CO2 concentration sensor by 15% while protecting building trade secrets. Finally, all calibrated data were unified to the WGS-84 coordinate system and normalized to prepare for subsequent analysis. Then, the Mask R-CNN semantic segmentation algorithm was used to divide the standardized building data into structural units such as walls, roofs, doors, and windows. The algorithm achieved an average intersection-over-union (mIOU) ratio of 80%. Thermal performance parameters of commonly used building materials in Shanghai were consulted, and combined with thermal infrared images, thermal properties such as thermal conductivity and heat capacity were assigned to each building structural unit. The solar radiation absorptivity of the office building's exterior walls was calculated to be 0.65. Real-time meteorological data from the Shanghai Meteorological Bureau was accessed to obtain hourly meteorological parameters such as temperature, humidity, wind speed, and solar radiation, providing external environmental data for building energy consumption simulation. The building energy consumption monitoring data was coupled with the model, and the backpropagation algorithm was used to update the thermal resistivity in the model. Adaptive step size adjustments improved the convergence speed of the thermal resistivity by 20%. Uncertainty Factor Modeling: Randomness of Residents' Behavior - Air Conditioner Use: The number of times air conditioners are used per unit time is modeled according to the Poisson distribution characteristics. The average number of times air conditioners are turned on per unit time is set to 5 to simulate the randomness of residents' air conditioner use behavior. Material Performance Degradation: The material performance degradation process follows a normal distribution law. By setting the average value of material performance degradation to 2 and the standard deviation to 0.5, a mathematical model is constructed to describe the change of material performance degradation over time. Next, the optimization objectives and decision variables were determined, with energy saving rate (annual comprehensive energy consumption reduction of more than 30%) and investment payback period (no more than 8 years) as the main optimization objectives. Decision variables included: external wall insulation layer thickness (0-15cm, material options: rock wool or polystyrene board), window type (replacing with Low-E double or triple glazing), shading measures (installing external shading or replacing with a smart shading system), and photovoltaic coverage (considering that the office building rooftop is suitable for installing photovoltaic panels). These were used for subsequent optimization calculations. Subsequently, multi-objective optimization and simulation verification were performed, including Monte Carlo simulation and weighted statistics: Poisson and normal distribution parameters were set, and the number of iterations was determined to be 1500. During each iteration, values ​​were randomly sampled from the corresponding distributions, parameters in the dynamic 3D model were modified, and EnergyPlus software was used to simulate building energy consumption. After each simulation, the weight coefficient for each scenario was quantified based on the degree of influence of different scenarios on the energy saving rate. This weight coefficient was dynamically adjusted according to the change in the energy saving rate of each scenario relative to the baseline scenario, thus reflecting the differences in the impact of different uncertainties on the performance of the schemes. After the iteration, weighted statistical analysis was performed on all simulation results, and statistical quantities such as the weighted average energy saving rate were calculated to comprehensively evaluate the performance of each scheme. Multi-objective optimization: A multi-objective optimization algorithm based on genetic algorithm was adopted, with the weighted statistical energy saving rate and investment payback period as objective functions, and decision variables as independent variables for optimization calculation. The algorithm population size was set to 100, the crossover probability to 0.8, the mutation probability to 0.05, and the maximum number of iterations to 200. The optimized algorithm outputs multiple green retrofit schemes that meet the requirements of energy saving rate ≥ 30% and investment payback period ≤ 8 years. Simulation verification shows that the scheme with an energy saving rate of 32% and an investment payback period of 7.5 years was ultimately selected as the final retrofit scheme. Renovation Costs: Statistics show that the cost of exterior wall insulation materials and construction is approximately 800,000 yuan, window replacement costs are 500,000 yuan, shading system installation costs are 300,000 yuan, and photovoltaic panel installation and related equipment costs are 1.2 million yuan, totaling 2.8 million yuan. Operating Costs: Before the renovation, the office building's annual energy consumption cost was approximately 1.2 million yuan; after the renovation, it is estimated to reduce annual energy consumption costs to 816,000 yuan, resulting in annual savings of 384,000 yuan. Revenue Calculation: The estimated annual revenue from photovoltaic power generation is approximately 100,000 yuan. Combining energy-saving and photovoltaic revenues, the total annual revenue is approximately 484,000 yuan. The calculated investment payback period is approximately 5.8 years, meeting the target of no more than 8 years. Real-time monitoring: After the renovation, a sensor network deployed within the building continuously monitors environmental parameters such as indoor temperature and humidity, light intensity, and CO2 concentration, as well as actual building energy consumption data, including readings from electricity meters and chiller units. Comparative analysis: The measured environmental parameters and energy consumption data are compared and analyzed with the predicted values ​​from the dynamic 3D model. Iterative optimization: If the actual energy consumption deviates from the model prediction by more than 5%, or the indoor temperature deviates from the model prediction by more than 2°C, an iterative optimization process is triggered. Depending on the type of discrepancy, sensor calibration parameters or model parameters, such as thermal resistivity, are adjusted, and the green renovation plan is re-optimized and validated. A new plan is generated and deployed in the actual building for continuous monitoring and iterative optimization. In summary, this embodiment achieves intelligent, refined, and green renovation of existing buildings through precise data acquisition and preprocessing, model construction considering uncertainties, scenario-adaptive multi-objective optimization and simulation verification, economic benefit calculation, and iterative optimization, and has good prospects for promotion and application. Example 3, an embodiment of the present invention, provides a green building optimization system for urban renewal based on multi-source data fusion, including a data acquisition and preprocessing module, a dynamic three-dimensional model construction and updating module, a scheme generation and verification module, and a closed-loop iterative optimization module.

[0077] The data acquisition and preprocessing module is used to collect building geometry data, thermal performance data, and environmental monitoring data, and preprocess the multi-source datasets to output standardized data. The dynamic 3D model construction and update module is used to integrate standardized data to construct a dynamic 3D model with thermal properties, and dynamically update the parameters of the dynamic 3D model based on real-time environmental monitoring data. The scheme generation and verification module is used to generate green renovation schemes based on the dynamic 3D model through multi-objective optimization, and use digital twin technology to simulate and verify the schemes, outputting the verified green renovation schemes. The closed-loop iterative optimization module is used to monitor environmental data after the renovation is implemented, and trigger data calibration parameter adjustments, 3D model updates, and green renovation scheme iterations based on the differences between actual monitoring data and 3D model predictions.

[0078] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0079] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0080] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0081] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0082] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for optimizing green buildings in urban renewal based on multi-source data fusion, characterized in that, Includes the following steps: (1) Collect building geometry data, thermal performance data and environmental monitoring data as multi-source datasets, and preprocess the multi-source datasets and output standardized data after preprocessing; (2) Integrate standardized data to construct a dynamic three-dimensional model with thermal properties, and dynamically update the parameters of the dynamic three-dimensional model based on real-time environmental monitoring data; (3) Based on the dynamic three-dimensional model, a green transformation scheme is generated through multi-objective optimization, and digital twin technology is used for simulation verification to output the verified green transformation scheme; (4) Based on the difference between the environmental monitoring data and the predicted values ​​of the dynamic three-dimensional model after the implementation of the verified green transformation scheme, dynamically trigger the parameter adjustment of data calibration, the update of the dynamic three-dimensional model, and the iteration of the green transformation scheme.

2. The method according to claim 1, characterized in that, In step 1: The geometric structure data is acquired using three-dimensional scanning technology, which includes oblique photogrammetry and lidar scanning. Thermal infrared imaging technology was used to collect thermal performance data and obtain the temperature distribution on the building's exterior surface. The environmental monitoring data is collected using a distributed sensor network, which includes temperature and humidity sensors, light intensity sensors, and gas concentration sensors deployed on the interior and exterior facades of the building. The preprocessing includes: time synchronization, spatial registration and noise filtering, as well as data calibration and normalization.

3. The method according to claim 2, characterized in that, In step 1: The oblique photogrammetry uses a drone equipped with an oblique camera to collect multi-angle images. The lidar scanning uses a ground-based three-dimensional laser scanner; The thermal infrared imaging technology employs a thermal imager; The distributed sensor network achieves synchronous data transmission through a wireless self-organizing network protocol. The time synchronization uses the NTP protocol to uniformly calibrate the timestamps of multi-source data; The spatial registration uses a feature point matching algorithm to spatially register the laser point cloud with the oblique photogrammetric image. The noise filtering employs a combined wavelet transform and median filtering algorithm to separate and filter out impulse noise and Gaussian noise in the sensor data.

4. The method according to claim 3, characterized in that, In step 2: The construction of the dynamic 3D model includes: dividing standardized data into building structural units based on a semantic segmentation algorithm, wherein the building structural units include walls, roofs, doors and windows; assigning thermal properties to each building structural unit, wherein the thermal properties include thermal conductivity, heat capacity and solar radiation absorptivity; and fusing 3D scanning data and thermal infrared images to generate a dynamic 3D model with thermal property labels. The process of updating the dynamic 3D model parameters includes: accessing the meteorological data interface in real time to obtain hourly meteorological parameters of the target area; dynamically adjusting the thermal parameters of the 3D model based on the meteorological parameters; and coupling building energy consumption monitoring data to update the equivalent thermal resistance coefficient in the model's heat conduction equation based on the backpropagation algorithm.

5. The method according to claim 4, characterized in that, In step 3: The generation of the green renovation plan includes: determining optimization objectives, including energy saving rate, carbon emission reduction rate, and economic efficiency; determining decision variables, including the thickness of the external wall insulation layer, window-to-wall ratio, window type, shading measures, and photovoltaic coverage rate; and establishing a multi-objective optimization model, using the optimization objectives as the objective function and the decision variables as the independent variables, to perform multi-objective optimization. The simulation verification includes: applying the generated green transformation scheme to a dynamic three-dimensional model for simulation; evaluating the energy saving rate, carbon emission reduction rate and economic indicators to determine whether the preset constraints are met; if the simulation results do not meet the constraints, adjusting the decision variables and re-optimizing and verifying.

6. The method according to claim 5, characterized in that, In step 4: Real-time monitoring of environmental monitoring data of the renovated building, including temperature, humidity, and energy consumption; The environmental monitoring data of the renovated building are compared with the predicted values ​​of the dynamic 3D model. The difference is calculated, and a difference threshold is set. When the difference exceeds the threshold, iterative optimization is triggered. The iterative optimization includes: adjusting the multi-source data calibration parameters and dynamic 3D model parameters according to the type of difference; re-optimizing and verifying the green renovation scheme to generate a new green renovation scheme; deploying the new green renovation scheme to the actual building and continuing to monitor and iterate.

7. A green building optimization system for urban renewal based on multi-source data fusion, employing the method described in any one of claims 1-6, characterized in that, Includes the following modules: The data acquisition and preprocessing module is used to collect building geometry data, thermal performance data, and environmental monitoring data, and to preprocess multi-source datasets to output standardized data. The dynamic 3D model construction and update module is used to integrate standardized data to construct a dynamic 3D model with thermal properties, and to dynamically update the parameters of the dynamic 3D model based on real-time environmental monitoring data. The scheme generation and verification module is used to generate green transformation schemes based on dynamic 3D models through multi-objective optimization, and to verify them using digital twin technology, outputting the verified green transformation schemes. The closed-loop iterative optimization module is used to monitor environmental data after the transformation is implemented. Based on the difference between the actual monitoring data and the predicted values ​​of the 3D model, it triggers the adjustment of data calibration parameters, the updating of the 3D model, and the iteration of the green transformation plan.

Images