Abstract
Indoor air quality analysis is of interest to understand the abnormal atmospheric phenomena and external factors that affect air quality. By recording and analyzing quality measurements, we are able to observe patterns in the measurements and predict the air quality of near future. We designed a microchip made out of sensors that is capable of periodically recording measurements, and proposed a model that estimates atmospheric changes using deep learning. In addition, we developed an efficient algorithm to determine the optimal observation period for accurate air quality prediction. Experimental results with real-world data demonstrate the feasibility of our approach.
Highlights
With the proliferation of cheap but reasonably accurate sensors, indoor air quality can be determined by measuring various factors through the sensors installed in a given space
The gated recurrent unit (GRU) network is an long short-term memory (LSTM) variant with only two gates [2], implementing Equation (5)
We show the results of experiments for identifying the optimal time-step size of the gated recurrent units (GRU) model, and compare with those of the brute force method
Summary
With the proliferation of cheap but reasonably accurate sensors, indoor air quality can be determined by measuring various factors (e.g., fine dust density) through the sensors installed in a given space. Such measurements can be used to detect changes in the atmospheric state. Due to the above difficulties, until recently, many indoor air quality control systems have controlled the variables by establishing thresholds This method applies a given operation when current conditions exceed preset values, regardless of the number of variables or obstacles in the space.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have