Abstract: Wet weather is a weather phenomenon characterized by high humidity levels that can affect people's lives and health. Weather forecasts do not directly report on wet weather, which makes it difficult for people to prevent it. The scarcity of data on wet weather in the research field also poses a challenge to its prediction. Constructed, an artificial dataset for predicting wet weather and a wet weather prediction algorithm based on a hybrid expert system was proposed, which can predict wet weather for the next 24 hours. Based on the prediction results, a wet weather scene mode was designed to proactively send warning messages to users and recommend that they activate the autonomous intelligent humidity control mode to actively regulate indoor air. The test results show that this technology can effectively predict wet weather and provide intelligent humidity control strategies based on the prediction results, improving users' living comfort and health.

Key words: Air conditioning, Dehumidification, Deep learning, Mixture of experts