Prediction of temperature change with multi-dimensional environmental characteristic based on CNN-LSTM-ATTENTION model

The environmental temperature is an important physical quantity to measure the thermal comfort of the indoor environment, and accurate prediction of the environmental temperature change is essential to control the environmental thermal comfort and improve user comfort. This paper proposes a temperature prediction method with multi-dimensional environmental characteristic based on convolutional neural network (CNN) and long short-term memory (LSTM) network with an attention mechanism, wherein the time correlation between the environmental temperature data and other multi-dimensional environmental characteristic is considered. First, the CNN model is used to capture the data characteristic of environmental temperature data and other environmental parameters. Then the LSTM model is used to extract the multi-dimensional environmental time series data. To improve the prediction accuracy of the proposed method, the weight of the attention mechanism to the output of the LSTM model is added to adjust the prediction results. Experimental results show that the proposed method has lower complexity, higher training efficiency and prediction accuracy in predicting the temperature changes in the next hour and adjacent moments, and prove that the proposed method is applicable.