Research On Indoor Occupancy Detection Algorithm For Smart Buildings

In the development process of smart buildings,information technology has always played an important role as a driving force.The development of various new technologies and techniques has improved productivity and provided technical support and various solutions for the development of smart buildings.The definition of smart buildings is to provide healthy,applicable,and efficient use of space for people,while saving resources,protecting the environment,and reducing pollution throughout their entire life cycle.Therefore,the development of smart buildings needs to be centered around people,and research on how to use technologies such as artificial intelligence,big data,and the Internet of Things to enhance the human experience in building spaces,realize the value of buildings serving people,and ultimately improve people’s sense of acquisition,happiness,and security.To achieve a people-centered approach,it is necessary to intelligently adjust resources based on personnel distribution information within the building,thus providing a more comfortable and convenient environment for occupants,and minimizing energy consumption without affecting their comfort level.However,current typical application scenarios of smart buildings lack refined space occupancy management,and it is difficult to achieve accurate measurement of indoor occupancy information under the premise of protecting privacy.To address this key issue in the development of smart buildings,this paper studies space occupancy estimation algorithms within buildings,with the following specific contents:Firstly,in response to the problem of high-precision space occupancy counting,the current mainstream solution is to use optical images for human recognition,but this method raises privacy concerns.In order to achieve space occupancy estimation under the premise of privacy protection,this paper proposes a non-intrusive human counting solution based on an infrared array sensor.The solution uses a low-resolution infrared array sensor and a well-designed finite state machine model to accurately calculate indoor occupancy information by tracking and dynamically modeling the position of indoor targets.Secondly,in response to the problem of space occupancy behavior tracking,this paper constructs a multi-object tracking model for space occupancy behavior based on attention mechanism.The model uses far-infrared image data and can handle multiple targets and multiple cameras.By using the attention mechanism,the model can analyze behavior based on target position information,focus attention on the spatial features of individual targets,and more accurately analyze target behavior.Experimental results show that in the case of multiple targets and multiple cameras,the model can improve the accuracy and efficiency of target tracking.Finally,the above theoretical and technical research results have been applied in practice.The space occupancy counting algorithm has been implemented on specific hardware systems,and various sensors commonly found within buildings have been integrated into the system.Additionally,a neural network model has been deployed on embedded hardware to predict comfort,effectively reducing the number of sensors needed within the building.Through designing circuits from the ground up and integrating multiple Internet of Things devices,a composite multi-functional sensor has been developed.This device has the advantages of small size,low cost,and strong portability.Test results show that the accuracy rate of space occupancy can reach over 95%.