An Intelligent Long-Term Prediction Model for Indoor Air Quality Based on Informer

According to a relevant survey, the average indoor worker spends approximately 21 hours indoors per day. Therefore, maintaining good indoor air quality is particularly important for people's health. In this study, we employed the Informer model for indoor air quality prediction and investigated the impact of different parameter settings on prediction performance. By adjusting key parameters such as seq__len (input sequence length), label_len (decoder starting label length), and pred _len (prediction sequence length), we evaluated the model's performance on long-term sequence prediction tasks. The experimental results demonstrated the outstanding performance of the Informer model in long-term sequence prediction, with longer prediction lengths enhancing prediction accuracy while shorter prediction lengths may lead to a slight decrease in performance. The experimental results of this study confirmed the strong performance of the Informer model in long-term sequence prediction of indoor air quality. In particular, the model performed well in long sequence prediction tasks, achieving MSE and MAE values of 0.087 and 0.215, respectively, which significantly outperformed other time series prediction algorithms. Therefore, the Informer-based long-term sequence prediction model for indoor air quality can effectively forecast future indoor air quality, making it of great significance for protecting people's health.