Estudo e desenvolvimento de um sensor sem fio inteligente para monitoramento distribuído de poluentes atmosféricos no contexto de cidades inteligentes

Abstract

Air pollution is directly related to the cause and aggravation of various diseases, in addition to causing ecological damage to the environment. According to the World Health Organization (WHO), it is estimated that 4.2 million deaths are attributed to air pollution each year and 91% of the world’s population lives in places that exceed indicated air quality limits. In this work, an intelligent wireless sensor for atmospheric pollutants, specifically Particulate Matter (PM) and gases, such as Methane (CH4) and Carbon Monoxide (CO), was developed in order to alert the population and enable the control of pollutants emissions. The developed sensor will be part of a Wireless Sensor Network (WSN) distributed by the main avenues of the city of João Pessoa-PB, where it will be possible to identify routes, times and days of weeks with the highest pollution rates. The wireless communication technology adopted was LoRa, which was evaluated in urban and forest environments. In this work, the PPD42NS sensor was used, which uses the light scattering method to detect particles, and the MQ family sensors, which present cross sensitivity and form an Electronic Nose, capable of detecting various gases with the support of an pattern recognition algorithm. The PPD42NS sensor was subjected to a calibration process using the reference equipment, HiVol 3000, to correct its measured values. Furthermore, considering that one of the main factors in the development of WSN is the limited supply of energy to power the sensor nodes, this work proposes a reduction in the energy consumption of the sensor by replacing some of its electronic components with digital processing, calibrating the analog output from the digital output. The linear correlation values obtained in the two experiments were greater than 0.9, indicating a strong association between the variables. Experiments were performed with the gas sensors from the MQ family for calibration in clean air and then they were exposed to odorous substances. Each sensor reacts specifically to the detection of odors, making it possible to identify the gases through pattern recognition. The data obtained were trained, tested and validated via Artificial Neural Network (ANN), where it was possible to identify the different patterns, differentiating alcohol, water and air, with an accuracy above 98% at the output.