基于GPRS的电动汽车道路行驶工况自学习

GPRS Based Driving Cycle Self-Learning for Electric Vehicle

提出了一种基于GPRS的道路行驶工况数据的远程采集方法,并将其应用在电动汽车的实际运行中,获得电动汽车道路试验原始数据库.同时将自组织映射(SOM)神经网络引入到行驶工况的自学习中,通过SOM网络对原始数据进行运动学片段的聚类分析,构建出了电动汽车在实际运行中的3种典型工况,为电动汽车基于行驶工况的自适应优化控制策略提供了基础环节.所构建的行驶工况和其他行驶工况相比具有一般规律,表明应用SOM网络能够很好地实现道路行驶工况的自学习功能.

A methodology to collect the driving cycle data remotely based on GPRS was presented and applied to a running electric vehicle to build a driving cycle database for road test. The self-organizing map （SOM） network was introduced into self-learning of driving cycle, so the cluster analysis was performed to classify kinematic sequence of original data. Based on the classification of kinematic sequence ,three types of typical driving cycles of electric vehicle road test were constructed and provided foundation for self-adapt optimal control strategy for electric vehicle. Compared with other driving cycles,the constructed driving cycles have common regularity,which shows that self-learning of driving cycle is perfectly realized by the application of SOM network.