Multistep electric vehicle charging station occupancy prediction using hybrid LSTM neural networks

Public charging station occupancy prediction plays key importance in developing a smart strategy to reduce electric vehicle (EV) operator and user inconvenience. However, existing studies are mainly based on conventional econometric or time series methodologies with limited accuracy. We propose new mixed long short-term memory neural network incorporating both historical state sequences time-related features for multistep discrete prediction. Unlike the LSTM networks, proposed model separates different types of handles them differently architecture. The is compared number state-of-the-art machine learning deep approaches EV data obtained from open portal city Dundee, UK. results show that method produces very accurate predictions (99.99% 81.87% 1 step (10 min) 6 steps (1 h) ahead, respectively, outperforms benchmark significantly (+22.4% one-step-ahead +6.2% ahead). A sensitivity analysis conducted evaluate impact parameters