An electric vehicle charging load prediction model for different functional areas based on multithreaded acceleration

This paper proposes an electric vehicle (EV) charging load prediction model for different functional areas based on multithreaded technology. This model comprehensively incorporates the preference characteristics of EV users in charging behavior and mode, as well as accounting for diverse travel purposes. It is worth emphasizing that a comprehensive analysis of the spatiotemporal travel characteristics exhibited by users during different seasons and on working days versus rest days is provided. In terms of temporal scale, refined probability models are employed to accurately fit the user's travel time variables. In terms of the spatial scale, the initial position and spatial transfer probability of users are analyzed. Subsequently, leveraging graph theory, a regional transportation network model is established. On this basis, a novel improved time optimal path decision-making algorithm is proposed by holistically considering factors such as road grade, saturation, traffic conditions, one-way/two-way streets, and delay at signalized intersections. Additionally, the impact of temperature on the upper limit capacity of batteries is analyzed, as well as driving/air conditioning energy consumption models are established. It is noteworthy that a two-stage charging power variation model is introduced, enhancing the precision of charging power calculation. Finally, the effectiveness of the method is validated through case analysis. The results demonstrate a significant disparity in load demand among various functional areas, and the spatiotemporal distribution of load demand is significantly influenced by seasons and working days versus rest days. Furthermore, when employing 4 threads, the proposed method achieves a speedup ratio exceeding 2.5 compared to conventional serial methods.