Desenvolvimento de um nariz eletrônico no contexto de internet das coisas para medições automatizadas de parâmetros de biogás em aterro sanitário

Abstract

The current model for environmental monitoring of biogas in landfills is characterized by periodic measurements in vertical gas drains and in the compacted soil cover layer, using classic instrumentation that is expensive and requires constant maintenance. An alternative to these measurements are Internet of Things (IoT) solutions that enable the development of an automated system with low-cost sensors and real-time data transmission. In this context, a device called Electronic Nose can be used. The objective of this research was to develop an Electronic Nose, in the context of IoT, with low cost, easy handling and low electricity consumption, for automated measurements of biogas parameters in landfills and data transmission in real time. This study was carried out at the Sanitary Landfill in Campina Grande-PB and the methodology was divided into three stages: biogas measurements in vertical drains, to map biogas emissions in the landfill; calibration of the IoT Electronic Nose, with planning of the calibration system, the collection of biogas at the landfill and the training of an Artificial Neural Network (ANN) to improve the selectivity of the sensors; and development of the IoT Electronic Nose, with the assembly of the final prototype, the installation of a data transmission tower with LoRaWAN technology and the planning of the field application. The process of measuring biogas in drains resulted in geospatial maps for the gases methane (CH4), carbon dioxide (CO2), oxygen (O2), carbon monoxide (CO) and hydrogen sulfide gas (H2S), which made it possible to identify the influence of operational and biodegradative parameters of waste on the gas emission process, demonstrating that the spatial analysis tool can support the monitoring of landfills. The ANN trained to calibrate the IoT Electronic Nose presented a good performance, allowing the reduction of the number of sensors used initially, without significant loss of network performance, and presenting Root Mean Square Error (RMSE) with values less than 1 for the most target gas outputs (CH4, CO2, O2 and CO), except for H2S, which was 1.47209 with the chosen activation function (ELU). The final prototype of the device developed presented a lower cost compared to classic instrumentation, and could positively impact the costs of environmental monitoring landfills.