Using Deep Learning Methods to Monitor Non-Observable States in a Building

Abstract

This paper presents results from ongoing research with a goal to use a combination of time series from non-intrusive ambient sensors and deep recurrent neural networks to predict room usage at a university campus. Training data was created by collecting measurements from ambient sensors measuring room CO2, humidity, temperature, light, motion and sound, while the ground-truth counts was created manually by human observers. Results include analyses of relationships between different sensor data sequences and recommendations for a prototype predictive model using deep recurrent neural networks.