Research On Indoor Occupants Detection And Occupancy Estimation Algorithm Based On Infrared Array

It is estimated that global building energy consumption will increase to 45% in the next 20 years,about half of which will be used in the HVACL(Heating,Ventilation,AirConditioning,Lighting)system.Obtaining basic personnel information is the primary prerequisite for building energy efficiency by using the HVACL-based regulation strategy.In addition,personnel information is also important for personnel evacuation and traffic statistics.The detection and estimation methods based on traditional environmental sensors generally have a slower sampling rate,and the real-time performance is lacking,and there is a large time deviation.Based on the Hidden Markov Model(HMM)and other methods of analysis and estimation of personnel,the time dependence of personnel information is neglected,and there is a large cumulative error.In order to meet people's needs,designing a high-precision,low-cost,robust personnel detection and estimation technology is a key issue to be solved urgently.This paper considers the cost of personnel detection and low intrusive demand,and proposes a human detection algorithm based on infrared array sensor.Firstly,according to the detection principle of infrared array sensor,the feasibility analysis of personnel detection was carried out,and a data acquisition system was designed.Then a data compression algorithm based on non-negative matrix factorization(NMF)is proposed to reduce the data dimension and reduce the interference caused by human movement.The data is then normalized to reduce the computational complexity.Finally,the matrix detection and connected region markers are used to realize the detection and analysis of personnel.This paper focuses on the estimation of the occupancy rate of high-rise buildings,and proposes an occupancy estimation scheme based on infrared array sensor and IHMMSoftmax.Feature extraction is first performed on the basis of the personnel detection algorithm.Then,the parameter training is combined with the video data to construct a nonhomogeneous hidden Markov model(IHMM).Specifically,in the parameter training phase,in view of the ambiguity of the relationship between the environmental parameters and the occupancy rate,Softmax regression is used instead of the Gaussian mixture model(GMM)to solve the performance probability matrix.In the estimation period of occupancy,online and offline estimation are realized by using forward algorithm and Viterbi algorithm respectively.Finally,the cross-validation method is used to obtain the estimation result,which can effectively improve the estimation accuracy.The experimental results show that the proposed human detection algorithm can effectively distinguish the target personnel and improve the recognition accuracy,which lays a foundation for the later occupancy estimation.The online estimation accuracy of the occupancy estimation scheme proposed in this paper is 0.8198,and the offline estimation accuracy is 0.8040.It can effectively reduce the accumulated error and improve the estimation accuracy,which is in line with the current development trend of indoor occupancy estimation.