A novel axle temperature forecasting method based on decomposition, reinforcement learning optimization and neural network

Hui Liu, Chengming Yu, Chengqing Yu, Chao Chen, Haiping Wu

Abstract: Highlights•The axle temperature is predicted by time series method.•Q-learning method is used to optimize the initial parameters of neural network.•EWT decomposition algorithm can preprocess the original axle temperature data.•A novel hybrid model is used to predict the trend of axle temperature.