A Prediction Algorithm for Vehicle Torque Demand based on LSTM

In this paper, a model of vehicle torque demand prediction based on Long Short Term Memory networks (LSTM) is presented. The correlation analysis of the data used to train the network is carried out, and appropriate characteristic parameters including acceleration pedal, brake pedal, speed and torque are selected as the input of the network. The characteristic parameters of the past 2 steps are used to predict the torque demand of the future. The data collected from vehicle controller is divided into two sections which are training set to train the LSTM and testing set to verify the performance. The results compared with that of artificial neural network show that prediction model can effectively improve the prediction accuracy.