Scholarlink

This study presents an Artificial Intelligence of Things (AIoT) system based on the Message Queuing Telemetry Transport (MQTT) protocol, incorporating Fully Connected Neural Network (FCNN) and Long Short-Term Memory (LSTM) models. The system comprises a micro weather station, a local central server, and a developed AI model. In the micro weather station, an Arduino board (ESP32) collects temperature, humidity, and water level signals. The local central server simulates four virtual environments using kernel-based virtual machine technology. It also receives sensing data from the ESP32 board and stores these data in a database server via the MQTT protocol. The developed AI model predicts temperature and humidity values, which are displayed on an organic light-emitting diode (OLED) panel and a web page. Experimental results show that the proposed AIoT system with the LSTM model achieved an accuracy rate of 99% on the validation set and 93% on the test set, demonstrating its capability to provide accurate sensing data for the next hour. Similarly, the AIoT system with the FCNN model achieved an accuracy rate of 99.8% on the validation set and 95% on the test set, making it suitable for providing sensing data for the next 10 min. Both AI algorithms effectively predict future weather conditions, delivering more accurate and comfortable weather data to enhance human quality of life.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Weather Prediction Based on Fully Connected Neural Network and Long Short-Term Memory Models for a Micro Weather System

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Guo-Ming Sung,
  • Hai-Ning Chu,
  • Yu-Cheng Chang,
  • Ching-Lin Lin

摘要

This study presents an Artificial Intelligence of Things (AIoT) system based on the Message Queuing Telemetry Transport (MQTT) protocol, incorporating Fully Connected Neural Network (FCNN) and Long Short-Term Memory (LSTM) models. The system comprises a micro weather station, a local central server, and a developed AI model. In the micro weather station, an Arduino board (ESP32) collects temperature, humidity, and water level signals. The local central server simulates four virtual environments using kernel-based virtual machine technology. It also receives sensing data from the ESP32 board and stores these data in a database server via the MQTT protocol. The developed AI model predicts temperature and humidity values, which are displayed on an organic light-emitting diode (OLED) panel and a web page. Experimental results show that the proposed AIoT system with the LSTM model achieved an accuracy rate of 99% on the validation set and 93% on the test set, demonstrating its capability to provide accurate sensing data for the next hour. Similarly, the AIoT system with the FCNN model achieved an accuracy rate of 99.8% on the validation set and 95% on the test set, making it suitable for providing sensing data for the next 10 min. Both AI algorithms effectively predict future weather conditions, delivering more accurate and comfortable weather data to enhance human quality of life.