基于深度学习法的无人驾驶车中自动变速器换挡模型  

Automatic Transmission Gear Shifting Model in Unmanned Driving Vehicle Based on Deep Learning Method

作　　者：于文涛[1] 李海斌[1] 刘阳 Yu Wen-tao;Li Hai-bin;Liu Yang(Tianjin Vocational Institute,Tianjin 300410,China;Tianjin Yunlian Technology Co.,Ltd.,Tianjin 300300,China)

机构地区：[1]天津职业大学,天津300410 [2]天津云联科技有限公司,天津300300

出　　处：《天津职业大学学报》2018年第6期93-96,共4页Journal of Tianjin Vocational Institute

基　　金：天津职业大学科学研究基金项目"基于Simulink的双离合自动变速器换挡品质优化与控制策略分析"(20171107;主持人:于文涛);天津职业大学"十三五"教育教学改革立项研究项目"基于悉尼协议的汽车运用与维修专业创新创业人才培养研究"(JY20170213;主持人:于文涛);天津市教育科学规划课题"京津冀一体化下高职院校专业布局研究"(VE3141;主持人:成起强)的研究成果

摘　　要：本文基于深度学习法建立了无人驾驶车中自动变速器换挡模型。该模型由两部分组成,第一部分基于PSO-Simulink算法建立了汽车自动变速器换挡模型,模型的输入层为换挡过程中的档位、换挡时间和发动机转速,输出层为油门开度和汽车速度。第二部分采用深度学习中CNN算法建立了无人车行驶过程中的路况障碍分析模型。基于深度学习法的无人驾驶车中自动变速器换挡模型具有匀速耗时少,迭代步数低,判定准确率高的特点。该模型可根据路面障碍物信息和距离,为无人驾驶车中自动换挡技术提供了一种新技术。同时,模型表明自动换挡模型之间的差异和路况障碍物样本采集图的采集误差是模型准确率的主要原因。In this paper,an automatic transmission gear shifting model in unmanned driving vehicle is established based on the deep learning method.There are two parts in this model.In the first part,an automatic transmission gear shifting model is established based on PSO-Simulink algorithm.The input layer of the model is gear position,gear shifting time and engine speed in the process of gear shifting,and the output layer is throttle opening and vehicle speed.In the second part,the CNN algorithm is used to establish on-the-road condition obstacle model in the process of driving unmanned vehicle.The automatic transmission gear shifting model based on the deep learning method has the characteristics of less uniform speed,less time,less iterative steps and high accuracy.The model can provide a new technology for automatic gear shifting in driverless cars according to the information and distance of obsta-cles on the road.At the same time,the model shows that the main reason for the accuracy of the model is the difference between the automatic gear shifting model and the collection error of the road condition obstacle sample.

关 键 词：自动变速器 换挡 深度学习 无人驾驶 优化分析 

分 类 号：G712[文化科学—职业技术教育学]