With the intensification of energy crisis and environmental crisis, countries have accelerated the development of new energy sources. Lithium-ion energy systems occupy an important position in the energy storage market because of their excellent performance, but temperature-related issues still hinder their further development. In order to solve this problem, researchers are committed to more accurate prediction of the temperature of lithium-ion energy system. Long and short-term memory network (LSTM) has always been considered to be able to process time series well. The emerging temporal convolution network (TCN), as a special convolutional network, has also been proven to be able to handle sequential tasks well. In this paper, a new allied temporal convolution-recurrent diagnosis network (TCRDN) is constructed by combining LSTM and TCN using an adaptive boosting algorithm. The proposed model is experimentally demonstrated to be able to predict the change of surface temperature of lithium-ion energy system more accurately.
Keywords lithium-ion energy system, long and short-term memory network, temporal convolution network.