Loading...
Publication
Airmon: sistema IoT de monitorização e classificação da ocupação em salas de aula
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 36431 | 4.15 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
A gestão dos parâmetros ambientais em espaços fechados é de extrema importância para manter a salubridade e conforto dos recintos, principalmente em espaços públicos muito frequentados. Em particular, em ambientes escolares, a má gestão destes parâmetros, pode impactar negativamente o bemestar dos alunos e docentes, manifestando-se em dificuldades de concentração, fadiga e/ou dores de cabeça, afetando, consequentemente, o processo de ensino-aprendizagem. A utilização de inteligência artificial para classificar a ocupação desses espaços, oferece a oportunidade de otimizar a sua gestão e planeamento, tornando o processo educacional mais eficiente e adaptado às necessidades dos intervenientes. Este trabalho tem como objetivo contribuir para uma possível solução para a lacuna existente relativamente à gestão de parâmetros ambientais, através da conceção e implementação de um sistema de monitorização de baixo custo e alta escalabilidade, visando a recolha e agregação precisa de dados ambientais, oriundos de diversas salas de aula, distribuídas por diferentes estabelecimentos de ensino e com a capacidade de integrar todos os que se pretendam associar ao projeto. Foram desenvolvidas caixas equipadas com um conjunto diversificado de sensores de baixo custo e consumo energético. O sistema disponibiliza uma interface intuitiva para o acesso e monitorização em tempo real de variáveis como os níveis de CO2, humidade, temperatura e partículas, referentes a cada sala de aula das escolas monitorizadas. Adicionalmente, os dados ambientais foram complementados com a indicação da ocupação das salas, através da colaboração da comunidade escolar, fornecendo assim uma solução que respeita a privacidade das pessoas envolvidas, não requerendo a utilização de métodos de recolha de dados invasivos, como câmaras. A estes dados foram aplicadas técnicas de inteligência artificial com o intuito de classificar a ocupação das salas de aula, obtendo uma acurácia de, no mínimo, 83% na classificação da ocupação com um modelo geral para todas as salas de aula e de, pelo menos, 85% quando treinadas para uma sala específica.
The management of environmental parameters in enclosed spaces is extremely important in order to maintain the health and comfort of the premises, especially in highly frequented public spaces. In particular, in school environments, poor management of these parameters can have a negative impact on the well-being of students and teachers, manifesting itself in concentration difficulties, fatigue and/or headaches, consequently affecting the teaching-learning process. The use of artificial intelligence to classify the occupancy of these spaces offers the opportunity to optimise their management and planning, making the educational process more efficient and adapted to the needs of those involved. The aim of this work is to contribute to a possible solution to the existing gap in the management of environmental parameters by designing and implementing a low-cost, highly scalable monitoring system aimed at accurately collecting and aggregating environmental data from various classrooms distributed across different educational establishments and with the capacity to integrate all those who wish to be associated with the project. Boxes equipped with a diverse range of low-cost, energy-efficient sensors have been developed. The system provides an intuitive interface for accessing and monitoring in real time variables such as CO2 levels, humidity, temperature and particulates for each classroom in the monitored schools. In addition, the environmental data was complemented with an indication of classroom occupancy, through the collaboration of the school community, thus providing a solution that respects the privacy of the people involved and does not require the use of invasive data collection methods such as cameras. Artificial intelligence techniques were applied to this data in order to classify classroom occupancy, achieving an accuracy of at least 83% when classifying occupancy with a general model for all classrooms and at least 85% when trained for a specific classroom.
The management of environmental parameters in enclosed spaces is extremely important in order to maintain the health and comfort of the premises, especially in highly frequented public spaces. In particular, in school environments, poor management of these parameters can have a negative impact on the well-being of students and teachers, manifesting itself in concentration difficulties, fatigue and/or headaches, consequently affecting the teaching-learning process. The use of artificial intelligence to classify the occupancy of these spaces offers the opportunity to optimise their management and planning, making the educational process more efficient and adapted to the needs of those involved. The aim of this work is to contribute to a possible solution to the existing gap in the management of environmental parameters by designing and implementing a low-cost, highly scalable monitoring system aimed at accurately collecting and aggregating environmental data from various classrooms distributed across different educational establishments and with the capacity to integrate all those who wish to be associated with the project. Boxes equipped with a diverse range of low-cost, energy-efficient sensors have been developed. The system provides an intuitive interface for accessing and monitoring in real time variables such as CO2 levels, humidity, temperature and particulates for each classroom in the monitored schools. In addition, the environmental data was complemented with an indication of classroom occupancy, through the collaboration of the school community, thus providing a solution that respects the privacy of the people involved and does not require the use of invasive data collection methods such as cameras. Artificial intelligence techniques were applied to this data in order to classify classroom occupancy, achieving an accuracy of at least 83% when classifying occupancy with a general model for all classrooms and at least 85% when trained for a specific classroom.