Speed Prediction Method and Energy Management Strategy for a Hybrid Electric Vehicle Based on Driving Condition Classification

In order to effectively improve the performance of a dual-mode hybrid electric vehicle, a predictive-control-based energy management strategy is devised to conduct online power distribution through real-time optimization, and a vehicle upcoming speed prediction method is proposed. Driving conditions are classified into stationary condition and quickly-changing condition though K-means clustering algorithm. Then current vehicle driving condition is determined real-time, and for obtaining best prediction accuracy, vehicle speed is predicted based on Markov-chain for stationary condition, while vehicle speed is predicted based on radial basis neural network for quickly-changing condition,. The comparison of simulation results verifies the correctness of vehicle speed prediction method proposed and the effectiveness of energy management strategy.