基于驾驶员类型识别的双离合自动变速器换挡规律研究  

Shift schedule of dual clutch automatic transmission based on driver type identification

作　　者：刘玺[1] 何仁[1] 程秀生[2] 

机构地区：[1]江苏大学汽车与交通工程学院,镇江212013 [2]吉林大学汽车工程学院,长春130025

出　　处：《农业工程学报》2015年第20期68-73,共6页Transactions of the Chinese Society of Agricultural Engineering

基　　金：江苏省高校自然科学研究重大项目(13KJA580001);江苏大学高级人才科研启动基金(14JDG155);江苏省自然科学基金(BK20150515)

摘　　要：为了使车辆驾驶性能满足驾驶员需求,提出了基于数据融合决策的驾驶员类型识别方法并建立了基于驾驶员类型的换挡规律。首先基于驾驶员的驾驶行为和驾驶意图,对驾驶员类型进行分析,制定了基于驾驶风格的驾驶员类型识别方案。选定能表征驾驶员驾驶风格的有效工况及相应的表征信号后,先采用BP神经网络分类器对驾驶风格进行辨识,再采用贝叶斯融合决策方法先后对同类操纵的驾驶风格辨识结果和所有操纵类型驾驶风格辨识结果进行数据融合决策,最终辨识出驾驶员类型。根据驾驶员类型,引入动力性系数,通过不同类型驾驶员对应的动力性系数值的改变,实现换挡规律中动力性因素和经济性因素所占比例的调整,最终形成基于驾驶员类型的DCT换挡规律。最后,以搭载6DCT的某试验车为对象,对不同驾驶员的换挡过程进行仿真实验,结果表明基于驾驶员类型的DCT换挡规律能够适应不同类型的驾驶员需求。该研究为驾驶员类型识别和智能型换挡规律的制定提供了参考。Shift schedule is one of the major factors for drivability. When using traditional method to establish shift schedule, it considers power performance and fuel economy, but neglects driver characteristics. Speed and throttle in traditional two-parameter shift schedule may reflect vehicle performance for driver to some extent, but driving characteristics of different drivers can＇t be considered. In this paper, a shift schedule method based on driver type was proposed for making vehicle maneuverability meet drivers＇ characteristics. In order to obtain the drive type, driving behavior and intention were analyzed according to drivers＇ operations in driving process, different driver characteristics were obtained, and then drivers could be classified into conservative and sport type. So identification scheme of driver type was proposed. Driver＇s operations, road condition and vehicle state were transformed into electrical signals by vehicle sensors. These electrical signals could be identified by electronic control unit and used to classify driving style, and then driver type could be obtained by fusion decision of driving style. Firstly, BP（back propagation） neural network classifier was employed for driving style identification from the obtained signals. The classifier designed had 3 layers, and any 2 layers were linked by nonlinear S-functions. The data of effective driving cycles and corresponding characteristic signals, which could remarkably characterize the driving style, were placed in the input layer, the different driving styles were obtained from the output layer, and the node number of the middle layer was optimized by the empirical formula. Moreover, the classifier was trained off-line on the basis of the driving data under various working conditions. Secondly, the driving styles were fused by Bayesian to obtain driver type. Because there were many different effective driving cycles while driving, the fusion decision process was performed in 2 stages. The fusion decision of driving style date of

关 键 词：车辆 变速器 识别 驾驶员类型 BP神经网络 贝叶斯融合 换挡规律 

分 类 号：U463[机械工程—车辆工程]