AIR QUALITY MEASUREMENT AND PREDICTION SYSTEM USING ARTIFICIAL NEURAL NETWORKS

Abstract

THE PURPOSE OF THIS STUDY IS TO MEASURE THE LEVEL OF AIR POLLUTION IN THE DISTRICT OF VENTANILLA AND MI PERU, IN PERU. THE CONCENTRATIONS OF SUSPENDED PARTICLES RANGE FROM 2.5 G TO 10 G, ALSO KNOWN AS PM10, AND THE CONCENTRATIONS OF SUSPENDED PARTICLES LESS THAN 2.5 G, ALSO KNOWN AS PM2.5. THE TASK IS TO MEASURE CO2, PM2.5 AND PM10 POLLUTION TO PROTECT THE HEALTH OF PEOPLE LIVING IN THE REGION UNDER STUDY. A SYSTEM WAS IMPLEMENTED TO MEASURE THE CONCENTRATIONS OF PM10 AND PM2.5 POLLUTANTS IN CO2 POLLUTED AIR. THE MEASUREMENT SYSTEM INCLUDES A DUST AND CO2 SENSOR, AS WELL AS AN AMBIENT TEMPERATURE AND HUMIDITY SENSOR, A DHT11 SENSOR FOR THESE MEASUREMENTS, AND AN ESP8266 MODULE FOR WIRELESS RECORDING AND CLOUD RECORDING. AN ARDUINO UNO R3 AND ESP8266 BOARD USE WIFI TO PROCESS THE SENSOR VALUES. A GOOGLE SHEETS SPREADSHEET AND A PAAS CLOUD COMPUTING SERVICE PROVIDED BY GOOGLE. AN ANN WAS CHOSEN BECAUSE IT HAS PROVEN TO BE EFFECTIVE IN AIR QUALITY PREDICTIONS. COMPARED TO OTHER SIMILAR WORKS, ONLY ONE NETWORK WAS CREATED, BUT SEVERAL PROTOTYPES WERE DEVELOPED AND EVALUATED TO AVOID ARBITRARINESS IN DESIGN DECISIONS. DATA NORMALIZATION, ARCHITECTURE SELECTION, AND ACTIVATION FUNCTION SELECTION WERE THREE SPECIFIC COMPONENTS OF THE NR DESIGN THAT WERE EXAMINED. FINALLY, ARTIFICIAL NEURAL NETWORKS ARE USED TO PREDICT PM10 AND PM2.5 PARTICULATE MATTER CONCENTRATIONS. © 2024 LATIN AMERICAN AND CARIBBEAN CONSORTIUM OF ENGINEERING INSTITUTIONS. ALL RIGHTS RESERVED.