Estimating the Fine-Grained PM2.5 for Airbox Sensor Fault Detection in Taiwan

Abstract

Recently, PM2.5 becomes one critical threat for the health of human. To monitor the PM2.5, the Airbox project that consists of more than 2000 PM2.5 sensors is executing in Taiwan. Thanks to the Airbox sensors, people can know the fine-grained air quality. However. Airbox sensors can fail and it is important to detect which sensor is failed to prevent the noise in the data. In this work, we focus on fault detection and value estimation for PM2.5 monitoring. To achieve our goal, we utilize the data from Environmental Protection Administration(EPA) for estimation. We firstly propose two estimation methods which consider the distance and similarity between the Airbox sensors and EPA monitoring stations for PM2.5 estimation. Then based on the estimation result, we detect which sensors is failed. We collect the data from Airhox Edimax web page and Taiwan's Environmental Protection Administration for our experiment. The experiment results reveal the good performance of our proposed methods.