TCCE_LSTM_A: A Temperature Correction Model with Clustering and Expansion by Combining Long Short-term Memory with Attention Mechanism

The paper presents a model called temperature correction model with clustering and expansion combing LSTM (Long Short Term Memory) with attention mechanism (TCCE_-LSTM_-A) based on deep learning for multi-meteorological factors is proposed. In this paper, a novel comprehensive evaluation index (AMM) is constructed by combining mean square error and mean absolute error with temperature prediction accuracy to objectively evaluate the temperature correction effect and selected the optimal temperature correction model. The results show that the accuracy of the highest temperature prediction in the next 24 hours was improved to 89.5%, and the accuracy of the lowest temperature prediction was improved to 90.8%. The comprehensive performance of the proposed temperature correction model was better than that of the traditional Kalman filter, neural network, and machine-learning model.