基于EEMD和熵理论的电动汽车制动意图识别方法  

Braking Intention Identification Method for Electric Vehicles Based on EEMD and Entropy Theory

作　　者：王波[1] 唐先智[1] 王连东[1] 杨树军[1] 马雷[1] Wang Bo;Tang Xianzhi;Wang Liandong;Yang Shujun;Ma Lei(College of Vehicles and Energy,Yanshan University,Qinhuangdao 066004)

机构地区：[1]燕山大学车辆与能源学院,秦皇岛066004

出　　处：《汽车工程》2018年第8期935-941,共7页Automotive Engineering

基　　金：国家自然科学基金青年基金(51505414);国家自然科学基金面上项目(51675462);河北省研究生创新项目(CXZZBS2017042);河北省高等学校科学技术研究重点项目(ZD2016012)资助

摘　　要：为抑制制动踏板信号中存在的间歇性成分或脉冲成分所造成的信号分解过程中的模式混叠现象,进一步提高制动意图识别的准确率和实时性,本文中提出了基于平均经验模式分解(EEMD)和熵理论的电动汽车驾驶员制动意图聚类识别法。首先,运用EEMD算法将制动踏板信号分解为IMF分量,以抑制模式混叠现象,更准确地提取制动踏板信号特征。接着,运用Shannon熵对IMF分量进行筛选,以减少特征提取的计算量。再用样本熵提取筛选后的制动踏板信号IMF分量的特征,得到不同制动意图的制动踏板信号特征向量。最后,运用聚类算法对制动意图进行识别。离线试验和实时试验的结果表明,基于EEMD和熵理论的制动意图聚类识别法比基于HHT的制动意图识别法具有更高的识别准确率和更好的实时性。For suppressing the mode mixing phenomena in the process of signal decomposition caused by intermittent or pulse compositions in brake pedal signals and further enhancing the correctness and real-time performance of braking intention identification,a cluster identification method for driver braking intention based on ensemble empirical mode decomposition(EEMD)and entropy theory is proposed for electric vehicles in this paper.Firstly brake pedal signals are decomposed into IMF components by EEMD to suppress mode mixing and more accurately extract brake pedal signal features.Then IMF components are selected by using Shannon entropy to reduce the computation efforts for feature extraction,and the features of IMF components of brake pedal signals selected are extracted by sample entropy to get the eigenvectors of brake pedal signals for different braking intentions.Finally clustering algorithm is adopted to perform braking intention identification.The results of off-line test and real-time test show that the clustering identification method for braking intention based on EEMD and entropy theory has higher accuracy rate and better real-time performance compared with that based on Hilbert-Huang transform.

关 键 词：电动汽车 制动意图 平均经验模式分解 熵理论 聚类识别 

分 类 号：U463.5[机械工程—车辆工程]