基于马尔可夫链的混合动力汽车行驶工况预测研究

A Research on Driving Condition Prediction for HEVs Based on Markov Chain

针对目前混合动力汽车行驶环境适应性的问题,提出一种适应于随机多变的交通状况和基于实时行驶工况预测的能量管理策略。该策略采用一阶齐次马尔可夫链状态转移概率矩阵,对主干道和快速路建立了行驶工况特征参数预测模型,并通过对预测准确度的对比分析,确定了滚动时间窗和预测视距的最佳值。研究结果表明,与传统基于既定循环工况的整车能量管理策略相比,基于工况预测的能量管理策略具有更好的燃油经济性,为新能源汽车能量管理实时控制提供了一种新方法。

Focusing on the issue of HEV adaptability to driving environment,a novel energy management strategy adapting to random and varying traffic situation is proposed based on the real time driving condition prediction. The strategy uses the state transition probability matrix of first-order homogeneous Markov chain to set up a driving condition characteristic parameters prediction model for both trunk roads and expressways,and the optimal values of rolling time window and preview distance are determined by comparative analysis on prediction accuracy.The results of research indicate that the energy management strategy based on driving condition prediction model can achieve better fuel economy than that with conventional energy management strategy based on preset driving cycle,providing a new way of real-time energy control for new energy vehicles.