2026-06-112026-06-112026-02-22COELHO, Gabriel da Silva Oliveira. Sistema inteligente residencial com IA e IoT para apoio à tomada de decisão baseada em alertas metereológicos contextuais. Orientadora: Evelin Helena Cardoso Gomes. 2026. 25 f. Trabalho de Curso (Bacharelado em Engenharia de Computação) – Faculdade de Engenharia da Computação, Campus Universitário de Castanhal, Universidade Federal do Pará, Castanhal, 2026. Disponível em: https://bdm.ufpa.br/handle/prefix/9596. Acesso em:.https://bdm.ufpa.br/handle/prefix/9596This study proposes an intelligent residential system based on the Internet of Things (IoT) and Artificial Intelligence, aimed at anticipating lo cal precipitation events. Unlike regional weather forecasts and models focu sed solely on global accuracy metrics, the proposed approach prioritizes the decision-making utility of alerts in residential environments. The system integra tes low-cost environmental sensors, temporal processing using recurrent neural networks, and event-oriented analysis, comparing a Baseline model (MR), a Very Short-Term Prediction model (MCP), and an Event-Based model (MBE). Results indicate that, although the MR exhibits superior overall statistical per formance, the MBE achieved greater anticipation capability and higher sensi tivity to atmospheric patterns preceding rainfall, correctly alerting 3 out of 5 observed precipitation events, while the MR and MCP generated only 1 and no useful alerts, respectively. These findings suggest that temporal and contex tual approaches are more suitable for residential alert applications than models focused exclusively on numerical precision.Acesso AbertoPrevisão meteorológicaSistemas de alertaInteligência artificialWeather forecastingAlert systemsArtificial IntelligenceCNPQ::CIENCIAS EXATAS E DA TERRATrabalho de Curso - Graduação - Artigo