Light Environment Modeling And Intelligent Optimization Based On Indcor Personnel Distribution

With the implementation of my country’s carbon neutrality goals,sustainable development strategies and policies to address climate change,the establishment of an intelligent building technology system that adapts to China’s development has become an important development direction in my country’s construction field.The lighting system is the main component of the building system.How to achieve intelligence,low energy consumption and adapt to people’s growing lighting requirements has become a bottleneck for the development of intelligent building technology systems.In a word,the following problems exist.It is hard to model the indoor light environment.The temporal and spatial distribution of lighting control system and personnel is unbalanced.And it is difficult to balance the operating energy consumption of the lighting system and the comfort of the light environment.Therefore,this thesis focuses on the comfort,energy saving,high efficiency and other indicators of the lighting system.This thesis conducts in-depth research on indoor light environment modeling method based on mechanism and data and intelligent lighting control strategy based on personnel distribution.In this thesis,the effectiveness of indoor light environment modeling method and intelligent lighting control strategy is verified by building hardware and simulation platform.The integral contents and results of this study are as follows:(1)Modeling method of indoor light environment based on mechanism and data.In this thesis,a multi-neural network combination method based on BP neural network and RBF neural network is proposed,and a combined sunlight and light prediction model is constructed.The contradistinctive experimental results show that it is beneficial to boost the prediction precision and velocity of illuminance.In this thesis,an indoor light environment illumination calculation model is established by using the point illumination mechanism and the combined prediction model based on data training.The results verify that it can quickly and accurately obtain the mixed illuminance value of sunlight and light at the location of the person.In this thesis,an indoor light environment simulation model is built based on DIALux,and the experimental results show the calculation accuracy and universality of the indoor light environment illuminance calculation model.(2)Indoor personnel positioning and clustering area division.This thesis uses TB-RK3399 Pro D to realize the design of human target detection and positioning application technology based on embedded platform.This thesis firstly builds the corresponding hardware and software environment.Secondly,the distribution of personnel at a certain moment in the room captured by the real-time camera is directly collected through TB-RK3399 Pro D,and the actual position coordinates of the personnel at this time are obtained by using the YOLO V3 RKNN model and the Y-CMAC network framework model.Validation experiments show that it achieves high-computational and low-cost applications for object detection and localization.In this thesis,the mean shift algorithm is used to divide the actual position coordinates of indoor personnel into clustering areas,which lays the foundation for determining the priority search lighting objects.Comparative experiments show that the clustering area division effect of indoor personnel distribution based on mean shift algorithm is better.(3)Intelligent lighting control strategy based on personnel distribution.For non-adjustable brightness lamps,this thesis proposes a decision-making strategy and an intelligent lighting search algorithm for indoor illuminance group control driven by personnel distribution.The comparative experimental results show that the strategy can search for the optimal combination of lamps with the minimum number and optimal position on the premise of meeting the uniform illumination requirements of each person.This thesis proposes an indoor illumination optimization control strategy driven by personnel distribution and a best-guided multi-swarm particle swarm optimization algorithm based on competitive learning for adjustable brightness lamps.Comparative experiments show that the strategy achieves the optimal illuminance value of each lamp in the optimal lamp combination on the premise of meeting the requirements of illuminance uniformity,lighting comfort,and illuminance;the algorithm proposed in this thesis speeds up the convergence speed of understanding and elevates understanding accuracy.The intelligent lighting control strategy based on personnel distribution achieves lighting energy saving,elevates comfort and meets the lighting needs of personnel by reducing the number of lamps turned on,finding the optimal combination of lamps and optimizing the dimming level of lamps.