Study on prediction of electric vehicle energy consumption based on driving parameters and environmental factors

Accurately estimating the energy consumption of electric vehicles (EVs) can provide users with a clear understanding of the energy consumption of EVs in different usage scenarios. Therefore, constructing a high-precision energy consumption estimation model is crucial for improving the reliability of EV travel. In this study, a real-world data-driven EV energy consumption prediction model is proposed. The model utilizing principal component analysis method and by radial basis function neural network (RBFNN) investigates not only the effects of vehicle speed, acceleration, and driving parameters such as drive system power and heating, ventilation, and air conditioning system power on the energy consumption of EVs but also the relationship between driving parameters and electric energy consumption of EVs. Furthermore, the effect of ambient temperature on the energy consumption of EVs is investigated. The prediction results based on RBFNN show that the model prediction errors are as low as 3.4%, 14.1%, and 18.5% for the evaluation indexes of mean squared error, mean absolute error, and root mean squared error, respectively, compared with other neural networks and integrated learning methods (support vector machine, backpropagation neural network, and random forest). The correlation index R2 is as high as 99.5%.