FL-CNN-LSTM: Indoor Air Quality Prediction Using Fuzzy Logic and CNN-LSTM Model
Ruiming Bao, Yulong Zhou, Wenrong Jiang
Abstract
With generations spending about 90% of their time indoors, indoor environmental quality monitoring and prediction is a very important tool to protect the health of indoor people from indoor air pollution. This paper proposes a hybrid fuzzy logic and deep neural network method to predict indoor air pollution. And using PM2.5 pollutants as an example, we use the proposed model to train and predict the dataset collected by indoor sensors from November 2016 to March 2017 in Shanghai. In the procedure of prediction, we designed comparison experiments using LSTM (Long Short-Term Memory), CNN-LSTM, and our proposed FL-CNN-LSTM network based on the PyTorch framework. The results show that deep neural networks injected with fuzzy logic can provide both better predictive performance and interpretability for the intended application. This helps to provide control strategies for smart IoT devices to improve the quality of human life.