Remaining Driving Range Prediction of Electric Vehicles Based on Personalized Driving Behavior in Complex Traffic Scenarios

Abstract

With the development of electric vehicles (EVs), accurate prediction of the remaining driving range (RDR) is a crucial issue. However, predicting RDR for EVs still poses significant challenges, particularly under complex traffic scenarios and personalized driving behaviors, where existing methodologies struggle to meet adaptability requirements. This study proposes a novel RDR prediction method for EVs that incorporates personalized driving behavior and emphasizes the influence of future route information on prediction performance. Real-world driving data, encompassing driving patterns, traffic conditions, and road characteristics, serve as the foundation for constructing a precise driving behavior model with Hidden Markov Model (HMM) and Fuzzy Markov Model (FMM). Furthermore, an energy consumption model that integrates the physical model with machine learning techniques is constructed to accurately predict energy consumption rates under future driving cycles. By coupling this consumption rate with battery load in an equivalent circuit model (ECM), the RDR is accurately predicted. To validate the performance of the proposed method, it is compared with baseline models under various driving scenarios. Experimental results demonstrate the accuracy and adaptability of the model, with a relative error within 10% in practical driving validation. This methodology offers insights into energy efficiency optimization and intelligent decision-making for EVs in complex traffic environments.