Fabricación de Sonómetros de Bajo Costo para Medición y Clasificación de Entornos Sonoros.
Loading...
Date
2022
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad de Concepción.
Abstract
El crecimiento en la población, industria y urbanización han provocado un aumento en la magnitud de ruido en los entornos sonoros que nos rodean. Este es un contaminante con el que interactuamos constantemente y provoca efectos adversos en la salud de las personas y en la fauna, por ello, es de urgencia reconocer y establecer medidas que gestionen el ruido ambiental. Para la medición y registro de los niveles de ruido, se utilizan instrumentos profesionales certificados llamados sonómetros, los cuales tienen un alto valor monetario, debido a esto, en este trabajo se desarrollaron dos sonómetros de bajo costo usando ordenadores de placa simple como alternativa a estos productos, los que poseen características que no tiene un sonómetro convencional como el funcionamiento continuo, respaldo en la nube, acceso remoto, comunicación con aplicaciones web y ejecución de modelos de aprendizaje profundo para la clasificación de entornos sonoros. Se realizó una calibración ambiental a los sonómetros desarrollados con un sonómetro profesional de clase 1 y se obtuvieron resultados óptimos en las mediciones, teniendo 0,3 decibeles de diferencia en valores de promedios minutales, 95 % de exactitud en la ejecución del modelo de aprendizaje y correcto funcionamiento en sus otras características, logrando un dispositivo que reduce diez veces el costo mínimo de un sonómetro de clase 1. A través de los resultados se evidenció la capacidad de obtener instrumentos de bajo costo y de gran calidad, siendo un instrumento más accesible para empresas, facultades, estudiantes y docentes en proyectos de investigaciones, para así suplir la necesidad enfática de regularización y registros de diferentes entornos sonoroso para mejorar la calidad de vida y salud de las personas.
The growth in population, industry and urbanization have caused an increase in the magnitude of noise at the sound environments that surround us. This is a pollutant with which we constantly interact and causes adverse effects on the health of people and wildlife, therefore it is urgent to recognize and establish measures to manage environmental noise. For the measurement and recording of noise levels, certified professional instruments called sound level meters are used, which have a high monetary value, due to this, in this work two low cost sound level meters were developed using single board computers as an alternative to these. products, which have features that a conventional sound level meter does not have, such as continuous operation, cloud backup, remote access, communication with web applications and execution of deep learning models for the classification of sound environments. The developed sound level meters were calibrated with a professional class 1 sound level meter and optimal results were obtained in the measurements, having 0,3 decibels of difference in values of minute averages, 95 % of accuracy in the execution of the learning model and correct operation in its other characteristics, achieving a device that reduces ten times the minimum cost of a class 1 sound level meter. Through the results, the ability to obtain low-cost and high-quality instruments was evidenced, being a more accessible instrument for companies, faculties , students and teachers in research projects, in order to meet the emphatic need for regularization and records of different sound environments to improve the environment and health of people.
The growth in population, industry and urbanization have caused an increase in the magnitude of noise at the sound environments that surround us. This is a pollutant with which we constantly interact and causes adverse effects on the health of people and wildlife, therefore it is urgent to recognize and establish measures to manage environmental noise. For the measurement and recording of noise levels, certified professional instruments called sound level meters are used, which have a high monetary value, due to this, in this work two low cost sound level meters were developed using single board computers as an alternative to these. products, which have features that a conventional sound level meter does not have, such as continuous operation, cloud backup, remote access, communication with web applications and execution of deep learning models for the classification of sound environments. The developed sound level meters were calibrated with a professional class 1 sound level meter and optimal results were obtained in the measurements, having 0,3 decibels of difference in values of minute averages, 95 % of accuracy in the execution of the learning model and correct operation in its other characteristics, achieving a device that reduces ten times the minimum cost of a class 1 sound level meter. Through the results, the ability to obtain low-cost and high-quality instruments was evidenced, being a more accessible instrument for companies, faculties , students and teachers in research projects, in order to meet the emphatic need for regularization and records of different sound environments to improve the environment and health of people.
Description
Tesis presentada para optar al grado académico de Ingeniero Físico.
Keywords
Contaminación acústica, Entorno sonoro, Aprendizaje profundo, Medidor de nivel de sonido