BIM-based building light environment neural network control method

Abstract

The application discloses a kind of based on BIM's building light environment neural network control method, comprising the following steps: step one, building light environment information collection;Step two, BIM environment creation;Step three, utilize revit to establish BIM model, utilize city thermal environment system software Envi-met, the surrounding environment information collected is input into software, simulates the influence of the change of surrounding environment data to building;Step four, according to light environment analysis, determine light environment scheme mode, fine control.The application is simulated by using BIM model, finds the best scheme of the lowest energy consumption under the condition of meeting the light environment comfort requirement.Through the adoption of building light environment neural network technology, the building light environment neural network system can reserve customizable entry, and reserve differentiated customizable options to ensure different functional spaces have different needs for light environment.

Description

Technical Field

[0001] This invention relates to a method for controlling building lighting environment, specifically a neural network control method for building lighting environment based on BIM, belonging to the field of intelligent building technology. Background Technology

[0002] Among the design, construction, and operation and maintenance (O&M) phases of BIM (Building Information Modeling) technology, O&M applications are the latest to be implemented and currently face the greatest limitations. At present, BIM technology's application in O&M mainly focuses on information storage and the visualization of operation and maintenance. A large amount of information in the BIM model is either filtered out during lightweighting processes or left idle because no application scenarios have been identified. This technology leverages BIM technology's utilization of lighting and energy consumption in O&M, thereby increasing the application of BIM technology in the O&M phase.

[0003] Meanwhile, there is almost no existing research on the application of BIM technology to building lighting environment neural networks. Current intelligent building management only makes intelligent judgments and executes based on the inherent physical characteristics, technical features, and usage characteristics of the building's interior. It can only be applied to extreme situations such as fire alarms, security monitoring, and air quality monitoring within the building, lacking the adjustment and control of the daily user environment and the real-time adjustment and control of changes in the building's external lighting, thermal, wind, acoustic, and air environments. This invention combines BIM technology, energy consumption calculation software, urban thermal environment system software, and neural network technology and applies them to buildings, enabling BIM technology to not only be used in extreme situations but also to adjust and control the daily user environment. Summary of the Invention

[0004] There is virtually no existing research on the application of BIM technology to building lighting environment neural networks. Current intelligent building management only addresses the inherent physical characteristics of the building's interior, limiting its application to extreme situations such as fire alarms. This invention provides a BIM-based building lighting environment neural network control method, attempting to combine and apply BIM technology, energy consumption calculation software, urban thermal environment system software, and neural network technology to buildings. This enables BIM technology, beyond its applications in extreme situations, to regulate and control the daily usage environment, and to detect and respond in real-time to changes in the building's external lighting, thermal, wind, acoustic, and air environments.

[0005] This invention is implemented as follows:

[0006] A BIM-based neural network control method for building lighting environment includes the following steps:

[0007] Step 1: Building lighting environment information collection;

[0008] Step 2: BIM Environment Creation;

[0009] Step 3: Use Revit to create a BIM model, and use the urban thermal environment system software Envis-met to input the collected surrounding environmental information into the software to simulate the impact of changes in the surrounding environmental data on the building.

[0010] Step 4: Based on the light environment analysis, determine the light environment scheme and implement refined control.

[0011] This invention uses BIM technology to simulate and compare environmental and building-related information of different buildings around the site of an existing building, obtains the lighting environment measures of the building with the lowest energy consumption, and then operates the building intelligently, thereby improving the actual building lighting environment utilization efficiency and helping to reduce the overall building energy consumption.

[0012] A further step is:

[0013] Step one specifically includes:

[0014] Determine the light environment sampling area. Divide the light environment sampling area into a 3m*3m grid within a 1km radius of the building. Collect the daylight factor and illuminance in each sampling area and analyze the light comfort.

[0015] The complexity of building lighting environment information collection stems from the intricate and diverse needs for indoor environmental comfort and building heating and air conditioning energy consumption control. Factors such as the mutual influence between different building groups, the mutual influence between different individual buildings within a building group, the different orientations of different rooms within an individual building, and the different exterior envelope forms and materials of rooms facing the same orientation all contribute to the regionalization, complexity, and diversity of lighting environment information collection needs. Furthermore, as the refinement of regional divisions increases, this complexity and diversity also intensifies. Therefore, the collection of lighting environment coefficients employs a relatively precise 3m*3m grid.

[0016] The device for acquiring information about the building's external light environment is called a light environment information collector (hereinafter referred to as the collector). The external light environment acquisition system is composed of collectors: the collector consists of sensors and protective shielding structures. The sensors are mainly sensors that sense the intensity of light, supplemented by sensors that monitor radiation energy within a certain spectral range.

[0017] The collector is designed with a reasonable angle of light reception and a suitable installation environment to collect only direct sunlight and energy radiation, avoiding or significantly reducing the influence of local ambient light and reflected light.

[0018] External factors such as direct sunlight shading and ambient light interference can affect the accuracy of data acquisition. The sensors in the data acquisition unit employ multiple measures to eliminate any external factors that could affect the accuracy and continuity of information acquisition. Specific measures are as follows:

[0019] To eliminate the obstruction caused by falling debris or birds landing, two data collectors are used at each collection point, maintaining a certain distance (3m) between them and allowing for comparison. If the data shows significant differences more than three times, a third collection point needs to be activated for remeasurement to determine the discrepancy. If differences reappear, manual intervention is required, and the data collector is alerted and instructed to be maintained.

[0020] In real life, it is impossible to avoid the obstruction of light by surrounding buildings and ambient light when placing a data collector (for example, it is impossible to avoid the lack of obstruction during the day on the coldest day). This is also the significance of actually using a data collector.

[0021] In areas with snow accumulation, customized collector setups can be implemented, and snow melting devices and snow melting structures can be used on the collectors to prevent snow accumulation.

[0022] Due to the limited number of sensors and their location at the highest point, it is impossible to collect light environment information for every part of the building. Therefore, the sensors are limited to collecting direct sunlight. The impact of surrounding building curtain walls on the light environment is collected using BIM simulation. Regarding light pollution from surrounding building curtain walls, since the incident direction of primary reflection light pollution from curtain walls is exactly opposite to the direction of direct sunlight, and the duration and intensity of the pollution are fixed, the data acquisition device is specially designed. Each data acquisition device uses a combination of multiple sensors with partial blocking. Each sensor is responsible for a specific angle of illumination, and the multiple sensors are combined to form a complete angle of illumination. (See appendix) Figure 5 and 6 .

[0023] The computer identifies, filters, and combines algorithms to collect data from each sensor in order to eliminate or reduce the impact of environmental light pollution sources.

[0024] The layout of the data collector corresponds to the service scope and target of the neural network. This ensures the accuracy of the collected information, the precision of sunlight environment information, and global coverage.

[0025] The service area can be formed by the combined integration of multiple data collection areas, ranging from a large urban area, a street block, a residential community, to a single building. Multiple building neural networks can share a single service area. The data collector is placed at a high point within the service area.

[0026] A further step is:

[0027] Step two specifically includes:

[0028] With the assistance of BIM technology, building lighting environment acquisition systems can become more streamlined and comprehensive. The BIM environment includes a series of objects with fixed characteristics, such as buildings, structures, ground and roads, green spaces, and landscapes. The BIM environment is the result of virtualizing the building's neural network service object and its environment using BIM technology. It consists of the building itself and various BIM models that affect the building's external lighting environment. By using Revit to create a BIM model and Envi-met, an urban thermal environment system software, and inputting collected surrounding environmental information into the software, changes in the surrounding environmental data can be simulated to impact the building, achieving a true digital twin.

[0029] A further step is:

[0030] Step three specifically includes:

[0031] A BIM model is built using Revit, and the urban thermal environment system software EnVi-Met is used to input collected surrounding environmental information, simulating the impact of changes in surrounding environmental data on the building. After the building's external lighting environment acquisition system and BIM environment are built, the construction of the building's BIM model's lighting environment data neural network needs to be deepened and supplemented according to the building's lighting environment simulation requirements. Combined with various actuators, intelligent adjustment can be achieved.

[0032] It mainly includes buildings, structures, ground, roads, landscapes, and greenery near the service body, as well as objects at a distance that may cause light pollution to the service body.

[0033] When the service environment covers a large area, the service environment model can acquire basic information from existing regional GIS systems, select potential impact objects (such as a high-rise glass curtain wall building several kilometers away from the service entity), finely adjust its lighting environment data, and include it in the service environment BIM model creation scope. Outside a 1-kilometer range, a function-based gradient mesh is used within the software, linking the Bézier curve mesh with the gradient formula to increase the mesh spacing.

[0034] The accuracy of the service ontology is far greater than that of the service environment. It should be determined by comprehensively considering all building operation and maintenance needs. Generally, the model accuracy is ≥ LOD500, corresponding to the completion stage. The model can be used for final settlement and as a central database integrated into the building operation and maintenance system. Creating the building ontology model (including interior decoration) is the most important part of this step, focusing on shading and light-blocking components such as doors and windows, requiring the establishment of window adjustability. In addition to material information, it also includes physical information such as the thermal resistance and heat storage coefficient of the building envelope. It should also be noted that the service ontology is also one of the influencing factors of the service environment.

[0035] The obtained data was linked to the BIM model in real time, and Design Builder software was used to perform indoor thermal energy simulation calculations on the structure including the surrounding landscape and building lighting environment to obtain simulated lighting environment energy consumption data. SPSS data analysis software was used to analyze the lighting environment simulation data to obtain comparative lighting environment results.

[0036]

[0037] In nature, LSG reaches a maximum value of 2.13 and a minimum value of 0.77.

[0038] This study compares the heat conduction, solar heat gain, and total heat gain of windows through the light-transmitting building envelope. The thermal insulation performance of windows in winter and their light-transmitting and shading performance in summer are evaluated using the heat transfer coefficient K and light-to-heat ratio (LSG). Simultaneously, data from BIM simulations of specific models are compared to obtain energy consumption comparison results. Data analysis reveals a coupling relationship between energy consumption and shading requirements; for example, lower energy consumption often results in higher shading and lower light dispersion. Using BIM energy consumption simulations of windows, the energy consumption simulation data of the three optimal light environment scenarios are compared.

[0039] The performance of the building envelope can be comprehensively evaluated using two parameters: winter thermal insulation performance (K) and summer light transmission and shading performance (LSG). The value of K can be used to adjust whether windows are opened and their opening angle, whether curtains are semi-transparent or fully transparent, whether air conditioning is needed, and to set up smart home systems. This ensures sufficient indoor light, and if insufficient, automatically turns on local light sources for intelligent control, thus guaranteeing indoor lighting needs while reducing energy consumption.

[0040] Different climate zones require different approaches. Taking a cold-winter, hot-summer region as an example, in the simulation model, the optimal option is a scheme with an adjustable heat transfer coefficient and light-to-heat ratio. Based on building sensitivity, lighting environment strategies and combinations are established. For example, in high-sensitivity areas, a comprehensive lighting environment scheme is implemented through superposition and recombination; artificial and natural lighting environments are considered holistically; and the overall lighting environment of a space is considered uniformly. Correspondingly, low-sensitivity areas use a basic lighting environment scheme.

[0041] (4) Based on the prediction and analysis of the light environment, determine the light environment scheme and continuously adjust it.

[0042] A fuzzy neural network control system is constructed, comprising a neural network prediction model and a fuzzy controller. The neural network prediction model uses the actual light intensity from the data acquisition device, the light intensity simulated by BIM, visible light transmittance, and heat transfer coefficient as input vectors, and the actual light intensity after implementing appropriate shading measures as the output vector, thereby predicting the effectiveness of the shading measures. The fuzzy controller uses the deviation Lux between the actual light intensity after implementing shading measures and the expected light intensity (referring to expert experience and national illuminance standards) and the rate of change of deviation Lux. c As inputs, the shading area difference Δs and the indoor temperature difference Δt are used as outputs to obtain the shading area deviation value Δs. This deviation value Δs is added to the shading area s to obtain the final shading area s+Δs, which is then used as the shading area value input to the next cycle of the neural network prediction model. The fuzzy controller uses a comprehensive generation method based on the linguistic fuzzy system inference method to obtain the fuzzy result.

[0043] The neural network prediction model uses the backpropagation (BP) algorithm for the learning process and has a three-layer structure. The first layer is the input layer with four nodes, which are the actual light intensity of the data acquisition device, the light intensity simulated by BIM, the visible light transmittance, and the heat transfer coefficient. The second layer is the hidden layer with four nodes. The third layer is the output layer with one node, which is the actual light intensity after taking the corresponding shading measures.

[0044] The hidden layer transfer function of the neural network prediction model uses the sigmoid transfer function logsig; the training function is triggered by the batch gradient descent method (TRAINGDM) with momentum; the training algorithm uses the quasi-Newton method; and the maximum number of training iterations is set to 10. 3 Training times; training target is 10. -2 The optimization parameters for the quasi-Newton training direction are calculated first, and then an appropriate learning rate is found.

[0045] The BP algorithm uses the squared network error as the objective function and employs gradient descent to calculate the minimum value of the objective function.

[0046] A database is established to store sensor data acquired by sensors placed on building rooftops. The neural network prediction model selects representative data from this database and continuously learns autonomously.

[0047] Based on the optimal scheme analyzed and evaluated in (3), real-time control is implemented, and corresponding lighting environment measures are taken. The lighting environment effect is fed back to the building BIM model in real time through sensors. Several lighting environment combination schemes are determined according to the lighting environment comfort requirements. The obtained data is linked to the BIM model in real time, and the impact of micro-environment solar intensity changes on building energy consumption is simulated using BIM. The energy consumption simulation data of different lighting environment combination schemes are compared, and the energy consumption comparison results are obtained. Under the condition of simultaneously meeting the lighting environment comfort requirements and minimizing energy consumption, the final lighting environment scheme is determined and transmitted to the existing building for real-time control, and corresponding lighting environment measures are taken. The lighting environment strategy and method are adjusted at any time through positive and negative feedback.

[0048] In the subsequent architecture of computer deep learning programming models, by using deep learning and adding influencing factors, the building lighting environment neural network system can achieve a high degree of intelligence by combining comprehensive information such as seasonal factors, weather forecasts, and other types of sensor devices.

[0049] Scenario 1: The distribution of snow accumulation in the service environment after snowfall is analyzed, and the reflectance and color information of the surface material in that area are intelligently adjusted. The energy consumption simulation data of the light environment is the distribution of snow accumulation. The comparison results of the light environment show that the reflectance information of the surface material is different in areas with more snow and areas with less snow. The light environment solution is determined by the illuminance of outdoor lighting fixtures at night after snowfall, which is controlled by connecting to the operation platform.

[0050] Scenario 2: The environmental model of deciduous trees changes with the seasons, and the system intelligently adjusts the light transmittance and color information of deciduous trees in winter and summer. The simulated data of light environment energy consumption shows the light transmittance under different leaf conditions in winter and summer. The comparison result of the light environment is the difference in light transmittance between winter and summer. The system determines the light environment plan, which includes a winter supplementary lighting plan and the adjustment of different illuminance of lamps in winter and summer. The system is controlled by connecting to the public operation platform.

[0051] Scenario 3 provides information for special cases where the external light environment changes frequently and significantly under cloudy weather conditions. Applicable execution plans can be made according to different needs and application scenarios. The light environment energy consumption simulation data is the distribution of snow cover. The light environment comparison results show that the surface material has different color reflection information in areas with more snow and less snow. The light environment plan is determined by the outdoor illuminance after snowfall at night, which is controlled by connecting to the operation platform.

[0052] Scenario 4 can intelligently exclude situations where the building lighting environment neural network system is unsuitable: for example, when the data collector is only installed on the top of a super high-rise building, there is a special case where the external lighting environment differs greatly between the upper half of the building and the lower half, which are inside the cloud layer. This can be solved by adding distributed auxiliary data collectors. The lighting environment energy consumption simulation data compares the indoor lighting environment distribution below the cloud layer. The comparison result determines whether the indoor light below the cloud layer meets the lighting environment requirements. The lighting environment solution is determined by using local lighting and supplemental lighting, controlled through a connection to the operation platform.

[0053] Scenario 5: Unpredictable light environment factors, such as light pollution from moving or parked vehicle windows incident on the data collector, can be excluded when setting the model boundary conditions. The light environment energy consumption simulation data excludes incident light from moving or parked vehicle windows. The light environment comparison result is a stable light environment result after removing these unstable factors. The determined light environment solution is the indoor light environment treatment method, such as opening windows or curtains, and is controlled through connection to the operation platform.

[0054] In extreme cases, intelligent judgment and emergency measures should be in place to address the possibility of abnormalities in the data acquisition system. Filtered and processed execution information should be output based on the application characteristics of different execution terminals to protect the execution mechanism and terminals.

[0055] The present invention has at least the following outstanding beneficial effects:

[0056] By placing a data collector at the highest point of the building, the light environment can be collected more accurately, leading to better control. Through simulation using a BIM model, the optimal solution with the lowest energy consumption while meeting light environment comfort requirements can be found, applicable to the following areas:

[0057] 1) Room light environment adjustment: Without the need for independent room sensors, it can automatically turn on or off or adjust the indoor artificial lighting in real time according to changes in the external light halo; adjust the shading system to improve glare, adjust the natural light intensity and light characteristics; and automatically block regular light pollution.

[0058] 2) Building thermal performance regulation: Adjustable shading measures to form a building adaptive shading system aimed at improving the thermal performance of the building envelope in summer; participate in the adjustment of the heat collector wall to achieve the purpose of intelligent adjustment of the heat collector wall.

[0059] By adopting architectural light environment neural network technology, the architectural light environment neural network system can reserve customizable entry points and differentiated customizable options. Different implementation schemes can be formulated for the characteristics of intelligent public buildings and intelligent residential buildings to ensure that different functional spaces have different light environment needs.

[0060] It can output information in a standard format, meeting the needs of users in each independent functional space to build more sophisticated neural networks and actuators.

[0061] The building lighting environment neural network system can record user-customized data as well as data from user-manual interventions and adjustments during use, and grasp the patterns within these data to gradually reflect user habits and reduce the frequency of user-manual adjustments and interventions.

[0062] The intelligent BIM-based building lighting environment control system, which combines neural network control and fuzzy control, achieves the adaptability and autonomous learning of the lighting environment control system, laying a solid foundation for the subsequent integration of intelligent lighting environment furniture and intelligent manufacturing, and providing a new approach to intelligent lighting environment control. Attached Figure Description

[0063] Figure 1 This is a schematic diagram of the invention;

[0064] Figure 2 A diagram illustrating how to set a "daylight path" in Autodesk Revit software;

[0065] Figure 3 A diagram illustrating how to set "Building Space Type" in the Autodesk Revit software ribbon;

[0066] Figure 4 This is a diagram illustrating how to select "Schedule / Quantity" in Autodesk Revit software.

[0067] Figure 5 A diagram illustrating the process of a data collector gathering sunlight.

[0068] Figure 6 This is a schematic diagram of the data collector. Detailed Implementation

[0069] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0070] As attached Figure 1 As shown, a BIM-based neural network control method for building lighting environment has the following specific steps: (1) collecting building lighting environment information; (2) creating BIM environment; (3) using Revit to build a BIM model, and using the urban thermal environment system software Envis-met to input the collected surrounding environment information into the software to simulate the impact of changes in surrounding environment data on the building; (4) determining the lighting environment scheme based on lighting environment prediction and deduction analysis, and continuously feeding back and adjusting.

[0071] (1) Building Lighting Environment Information Collection: Planning the scope of lighting environment collection. Within a 1-kilometer radius, a 3*3 grid is used to divide the lighting environment collection area, and lighting environment information collectors are distributed on the rooftops of the building clusters to collect daylighting coefficients and illuminance.

[0072] (2) BIM environment creation.

[0073] A. Use Autodesk Revit to create a basic architectural model.

[0074] Obtaining a BIM model of a building cluster allows access to its planning information. Using Revit software as the platform for building information modeling, a BIM building model can be created based on this planning information, or it can be obtained from Revit forward design results. The building cluster model includes multiple surrounding buildings situated within a certain distance. The spacing between the surrounding buildings and the target building directly affects the building's lighting performance.

[0075] Using Autodesk Revit, the project environment is created, including models of surrounding buildings, structures, ground and roads, model greenery and landscape, and other objects with fixed characteristics; each model is also numbered to distinguish it.

[0076] There are specific distinctions and requirements regarding model accuracy, as follows:

[0077] The shape accuracy of the service environment model: Models are created based on the principle that higher accuracy is achieved when the model is closer to the service entity; generally, only the surface model is created. For transparent structural environments, an internal model must be created.

[0078] Information accuracy of the service environment model: In addition to the light occlusion information generated by the model shape, the necessary information is added to the building model that determines potential objects that affect the external light environment of the service body, including: surface material information of the shape, including the linear reflection characteristics and color characteristics of light and heat radiation, and material information of the transmissive material.

[0079] It mainly includes buildings, structures, ground, roads, landscapes, and greenery near the service body, as well as objects at a distance that may cause light pollution to the service body.

[0080] When the service environment covers a large area, the service environment model can acquire basic information from existing regional GIS systems, select potential impact objects (such as a high-rise glass curtain wall building several kilometers away from the service entity), finely adjust its lighting environment data, and include it in the service environment BIM model creation scope. Outside a 1-kilometer range, a function-based gradient mesh is used within the software, linking the Bézier curve mesh with the gradient formula to increase the mesh spacing.

[0081] B. Use the daily data collected by the Autodesk Revit data collector to build a solar radiation model.

[0082] Accurately locate the building's geographical information, and in Autodesk Revit, select "Daylight Settings" under "Daylight Path" to simulate the daylight path. Set the "Daylight Path" to simulate the changes in sunlight throughout the day, as shown in the attached image. Figure 2 As shown.

[0083] Define building types based on different building functions and properties, and determine the lighting load of main rooms. For example, if a residential project is a residential building, fill in the residential building parameters in the lighting-related settings according to the requirements of GB50034 "Standard for Lighting Design of Buildings". Later, modify the settings for auxiliary rooms, machine rooms, and other areas separately to save on repetitive definition work. Set these settings in the Revit model's "Manage" - "MEP Settings" - Building Space Type. Through Revit's space functions, automatically obtain information about different rooms in the building: area, volume, etc., as shown in the attached image. Figure 3 As shown.

[0084] According to the requirements of the standard "Standard for Lighting Design of Buildings", illuminance requirements are specified for different functional areas within the building, as shown in the attached document. Figure 4 As shown, in the "Analysis" - "Schedule / Quantity" properties dialog box, select the space type and required lighting level in the "Fields" section. Based on the actual project situation, add rows to the schedule, entering the space type and corresponding illuminance required in the project.

[0085] After completing the parameter settings and specifying the illuminance requirements, the lighting fixtures are arranged by creating a spatial illuminance analysis detail table.

[0086] 3. Using the microclimate modeling software Envi-met, the collected surrounding environmental information (especially vegetation) is input into the software to simulate the impact of changes in surrounding environmental data on the temperature, humidity, wind speed, and other environmental factors inside the building.

[0087] This system comprehensively considers various forms of shading from buildings, surfaces, and multiple reflections from structures and vegetation to calculate shortwave and longwave radiation fluxes. Advanced modeling of radiation processes within the plant canopy, including scattering and diffuse reflection, is performed. Evapotranspiration and sensible heat fluxes from plants are determined, including comprehensive simulations of all plant physical parameters (such as photosynthetic rate). Feedback processes between soil moisture content and plant water are simulated. Water and heat exchange within the soil system is simulated. Three-dimensional heat transfer simulations are conducted based on ground materials and moisture content. Advanced calculations include hydraulic water exchange in the soil, encompassing root water uptake and plant water supply, and a study of the microclimate of the building environment.

[0088] Model building. Obtain building geospatial information and convert the acquired images into a recognizable .BMP format base map. Based on this base map, build a simulation model of the study area in ENVI-met software. Since the maximum grid size of the provided computational area is 250×250×30, sufficient space is left in the horizontal direction, and the height in the vertical direction is greater than or equal to twice the height of the tallest building in the simulation area.

[0089] In addition to modeling architectural space information, ENVI-met modeling also requires setting up and modeling vegetation, underlying surfaces, etc. Then, calculations are performed to view parameters such as wind speed, pressure, temperature, humidity, and PM2.5.

[0090] The obtained ENVI-met outdoor microenvironment data is linked to the BIM model in real time. Design Builder software is used to perform indoor thermal energy simulation calculations on the structure including the surrounding landscape and building lighting environment, yielding simulated lighting environment energy consumption data. SPSS data analysis software is then used to analyze the simulated lighting environment data, resulting in comparative lighting environment analysis.

[0091] The study compares the heat conduction, solar heat gain, and heat gain through the light-transmitting building envelope, and evaluates the thermal insulation performance of windows in winter and their light-transmitting and sun-shading performance in summer using two indicators: heat transfer coefficient and light-to-heat ratio.

[0092] Establish illuminance requirements and specify illuminance requirements for different functional areas within the building in accordance with the requirements of the standard "Standard for Lighting Design of Buildings".

[0093] After setting all parameters and specifying illuminance requirements, spatial lighting analysis allows for the simultaneous comparison of data simulated in BIM using a specific model. This provides energy consumption comparison results, which are then analyzed. Since energy consumption comparison and shading requirements are coupled, for example, low energy consumption may result in poor lighting comfort. Therefore, BIM energy consumption simulations of windows are used to compare the optimal energy balance schemes for three lighting environments.

[0094] The performance of the building envelope can be comprehensively evaluated using two parameters: winter thermal insulation performance (K) and summer light transmission and shading performance (LSG). The value of K can be used to adjust whether windows are opened and their opening angle, whether curtains are semi-transparent or fully transparent, whether air conditioning is needed, and to set up smart home systems. This ensures sufficient indoor light, and if insufficient, automatically turns on local light sources for intelligent control, thus guaranteeing indoor lighting needs while reducing energy consumption.

[0095] Different climate zones require different approaches. Taking a cold-winter, hot-summer region as an example, in the simulation model, the optimal option is a scheme with an adjustable heat transfer coefficient and light-to-heat ratio. Based on building sensitivity, lighting environment strategies and combinations are established. For example, in high-sensitivity areas, a comprehensive lighting environment scheme is implemented through superposition and recombination; artificial and natural lighting environments are considered holistically; and the overall lighting environment of a space is considered uniformly. Correspondingly, low-sensitivity areas use a basic lighting environment scheme.

[0096] (4) There is a coupling relationship between light environment and light comfort, thermal environment and thermal comfort and building energy consumption.

[0097] Based on the needs of the lighting environment and the building's energy consumption, a balanced intermediate value is obtained to meet the comfort requirements of the lighting environment while minimizing building energy consumption. The real-time lighting environment scheme is determined and continuously fed back and adjusted. Based on the optimal scheme analyzed and evaluated in (3), real-time control is carried out and corresponding lighting environment measures are taken. The lighting environment effect is fed back to the building BIM model in real time through sensors. The lighting environment scheme is determined based on the lighting environment prediction and deduction analysis. The obtained data is linked to the BIM model in real time. The influence of changes in micro-environmental solar intensity on building energy consumption is simulated using BIM. The energy consumption simulation data of different lighting environment combination schemes are compared. At the same time, the data are compared to obtain the energy consumption comparison results, determine the lighting environment scheme, and transmit it to the existing building for real-time control and take corresponding lighting environment measures. The lighting environment effect is fed back to the building BIM model in real time through sensors. Through positive and negative feedback, the lighting environment strategy and method are adjusted at any time.

[0098] Although the present invention has been described herein with reference to illustrative embodiments, the above embodiments are merely preferred embodiments of the present invention, and the implementation of the present invention is not limited to the above embodiments. It should be understood that those skilled in the art can devise many other modifications and implementations, which will fall within the scope and spirit of the principles disclosed in this application.

Claims

1. A BIM-based neural network control method for building lighting environment, characterized in that... Includes the following steps: Step 1: Building lighting environment information collection; Step 2: BIM Environment Creation; Step 3: Use Revit to create a BIM model, and use the urban thermal environment system software Envis-met to input the collected surrounding environmental information into the software to simulate the impact of changes in the surrounding environmental data on the building. Step 4: Based on the light environment analysis, determine the light environment solution and implement refined control. In step one, a 3m [meter] is used within a 1-kilometer range. A 3m grid was used to divide the light environment acquisition area, and light environment information was collected by light environment information collectors distributed on the rooftops of the building complex. Step two specifically includes: 1) Use Autodesk Revit to create a basic architectural model; 2) Establish a solar radiation model using daily light environment data collected by Autodesk Revit and a light environment information acquisition device; Step three specifically includes: The obtained data is linked to the BIM model in real time, and the Design Builder software is used to perform indoor thermal energy simulation calculations on the structure with surrounding landscape and building lighting environment to obtain lighting environment energy consumption simulation data; SPSS data analysis software is used to analyze the lighting environment simulation data to obtain lighting environment comparison results. Where LSG represents the light-to-heat ratio, T lum TIR represents visible light transmittance, while TIR represents near-infrared transmittance. Step four includes: Based on the optimal solution analyzed and evaluated in step three, real-time control is implemented, and corresponding lighting environment measures are taken. The lighting environment effect is fed back to the building BIM model in real time through sensors. Several lighting environment combination schemes are determined according to the lighting environment comfort requirements. The obtained data is linked to the BIM model in real time, and the impact of micro-environmental solar intensity changes on building energy consumption is simulated using BIM. The energy consumption simulation data of different lighting environment combination schemes are compared to obtain energy consumption comparison results. Under the condition of simultaneously meeting the lighting environment comfort requirements and minimizing energy consumption, the final lighting environment scheme is determined and transferred to the existing building for real-time control, and corresponding lighting environment measures are taken. Through positive and negative feedback, the lighting environment strategy and methods are adjusted at any time.

2. The BIM-based neural network control method for building lighting environment according to claim 1, characterized in that: The light environment information includes daylight factor and illuminance.

3. The BIM-based neural network control method for building lighting environment according to claim 1, characterized in that: Real-time control is achieved by constructing a fuzzy neural network control system, which includes a neural network prediction model and a fuzzy controller. The neural network prediction model takes the actual light intensity from the data acquisition device, the light intensity simulated by BIM, the visible light transmittance, and the heat transfer coefficient as input vectors, and the actual light intensity after taking corresponding shading measures as the output vector, thereby predicting the effectiveness of the shading measures. The fuzzy controller takes the deviation e between the actual light intensity after taking corresponding shading measures and the expected light intensity, as well as the rate of change e of the deviation, as input vectors. c As input, the shading area difference Δq and the indoor temperature difference Δt are used as outputs to obtain the shading area deviation value Δq. This deviation value is added to the shading area q to obtain the final shading area q+Δq, which is then used as the shading area value input to the next cycle of the neural network prediction model. The neural network prediction model uses a BP neural network with a three-layer structure. The first layer is the input layer with four nodes, which are the actual light intensity of the data acquisition device, the light intensity of the BIM simulation, the visible light transmittance, and the heat transfer coefficient. The second layer is the hidden layer with 4 nodes; the third layer is the output layer with 1 node, which represents the actual light intensity after taking the corresponding shading measures. The hidden layer transfer function of the neural network prediction model uses the sigmoid transfer function logsig; the batch gradient descent method TRAINGDM with momentum is triggered by the training function traingdm; the training algorithm uses the Quasi-Newton method; and the maximum number of training iterations is set to 10. 3 Training times; training target is 10. -2 The parameters for optimizing the quasi-Newton training direction are calculated initially, and then an appropriate learning rate is found. The BP algorithm uses the squared network error as the objective function and employs gradient descent to calculate the minimum value of the objective function. A database is established to store sensor data acquired by building roof sensors. The neural network prediction model selects representative data from this database and continuously learns autonomously.

Images