2019 IEEE Region 10 Symposium (TENSYMP) | 2019
Environment Dependent Neural Network Model For Occupancy Detection
Abstract
Occupancy detection in buildings crucially involves with the improvement of energy efficiency, space utility, heating ventilation and air condition (HVAC) control and hence optimization of user comfort. The human occupancy information is extracted through different environmental sensors like CO2 sensor, light sensors, humidity sensors etc. This paper presents an environment dependent artificial neural network (ANN) approach to Figure out the problem of binary classification of human occupancy inside a building space. This ANN model has been modified to a hyper parameter tuned neural network, dependent of run-time environmental factors based on some empirical formulation. This study has been performed to enhance the classification accuracy with an ample quantity of training samples and hence extend an intelligent occupancy detection model. Employing the statistical concept of occupancy, this proposed ANN model is developed and evaluated with University of California, Irvine (UCI) repository occupancy data. Prospective results demonstrate a faithful performance of the model with guaranteed classification accuracy.