Deficient air quality in industrial environments creates a number of problems that affect both the staff and the ecosystems of a particular area. To address this, periodic measurements must be taken to monitor the pollutant substances discharged into the atmosphere. However, the deployed system should also be adapted to the specific requirements of the industry. This paper presents a complete air quality monitoring infrastructure based on the IoT paradigm that is fully integrable into current industrial systems. It includes the development of two highly precise compact devices to facilitate real-time monitoring of particulate matter concentrations and polluting gases in the air. These devices are able to collect other information of interest, such as the temperature and humidity of the environment or the Global Positioning System (GPS) location of the device. Furthermore, machine learning techniques have been applied to the Big Data collected by this system. The results identify that the Gaussian Process Regression is the technique with the highest accuracy among the air quality data sets gathered by the devices. This provides our solution with, for instance, the intelligence to predict when safety levels might be surpassed.
Keywords: air quality monitoring; machine-learning; particulate matter; polluting gas.