| According to statistical analysis,people spend 80% to 90% of the time indoors,and the results of indoor environment control directly affect the thermal comfort of the human body.Therefore,research on thermal comfort regulation is of great significance for improving the work efficiency and quality of life of indoor personnel.Research on thermal comfort control of buildings widely uses predictive mean voting(PMV)indicators for evaluation and analysis,including four environmental parameters and two human parameters.In practical applications,traditional detection methods cannot achieve real-time dynamic measurement of human body parameters,resulting in differences between realtime thermal comfort assessment and actual conditions,resulting in poor indoor PMV regulation and affecting the human thermal comfort level to a certain extent.As computer vision technology has achieved good results in the field of target detection and recognition,it has been widely used.Based on this,this paper uses machine vision technology to realize online estimation of human body parameters of indoor personnel;introduces the estimated human body parameters into the air-conditioning system control,and establishes and implements a dynamic thermal comfort control strategy based on real-time estimation of human body parameters.First of all,this paper establishes a deep learning-based motion state detection model to identify indoor people walking,reclining,resting,sitting and other motion states,and using look-up table mapping method to realize the real-time estimation of the personnel’s metabolic rate.The motion state detection model uses the C3 D basic network architecture to extract spatio-temporal features,and retains the effective information in the original image to the maximum extent through adaptive pooling operations,improves feature expression capabilities,and improves classification and recognition accuracy.The experimental results show that the method in this paper can better realize the recognition of people’s movement status and the real-time estimation of metabolic rate.Then,this paper establishes a dynamic estimation method of indoor clothing thermal resistance based on machine vision.This method is based on the Mask RCNN network and extracts image features through the Res Net-101 feature pyramid;uses the Mask branch to perform pixel-level segmentation operations to obtain the personnel’s clothing area;uses the area generation method to complete shorts,trousers,short sleeves,long sleeves and other categories detection;through look-up table mapping,online dynamic estimation of clothing thermal resistance is realized.The experimental results show that the method in this paper is applicable and effective in the detection of personnel’s clothing status and the estimation of clothing thermal resistance.Finally,this article realizes the indoor environment thermal comfort regulation based on the dynamic detection of human body parameters.Through visual analysis of indoor video image information,estimate human metabolism rate and clothing thermal resistance parameters;combine with environmental parameters such as temperature and humidity detected by measuring instruments to obtain dynamic description of indoor environment PMV;and then dynamically measure and set values based on PMV.The PID controller is used to adjust the air volume at the end of the air conditioner to realize the thermal comfort regulation of the indoor environment.The experimental results show that,compared with the traditional control method,the PMV thermal comfort control effect of the dynamic detection of human body parameters in this paper is better,and it saves energy under the premise of ensuring the thermal comfort of personnel. |
