基于牵引力-滑转率特性的高速电驱履带车辆行驶路面辨识方法研究  

Research on Road Surface Identification Method for High Speed Electric Drive Tracked Vehicles Based on Traction Slip Rate Characteristics

作　　者：侯云龙 盖江涛[1] 袁艺[1] 曾根[1] 李训明 马长军[1] HOU Yunlong;GAI Jiangtao;YUAN Yi;ZENG Gen;LI Xunming;MA Changjun(China North Vehicle Research Institute,Beijing 100072,China)

机构地区：[1]中国北方车辆研究所,北京100072

出　　处：《车辆与动力技术》2024年第1期15-21,共7页Vehicle & Power Technology

摘　　要：电驱动高速履带车辆质量大、行驶速度快、行驶条件复杂,识别地面特性对其动力学控制具有重大意义.基于电驱动高速电驱履带车辆与地面之间的牵引力-滑转率特性,本文提出了一种利用长短期记忆神经网络(LSTM)进行路面识别的方法.通过采集车辆行驶过程中的驱动电机转矩、转速信号及惯性测量单元得到的车辆纵向、横向加速度信号,记录并筛选出在一个滑转率变化过程中的牵引力-滑转率特征数据,通过LSTM神经网络,将特征数据归类,识别出当前行驶路面类型.该方法具有采集设备简单、信号易获得、算力要求较低等优点.仿真结果表明,该方法能够将车辆行驶路面准确归类于三种典型高、中、低附着路面.Electrically driven high-speed tracked vehicles have large mass,high speed and complex driving conditions.There is great significance to identify the ground characteristics for its dynamic control.Based on the traction-slip characteristics between electrically driven high-speed tracked vehicles and the ground,this paper proposes a ground recognition method using long short-term memory neural network(LSTM).The vehicle longitudinal and lateral acceleration,drive force,rotation speed are obtained by collecting the driving motor controller and inertial measurement unit during the driving process.The traction characteristics of the vehicle in a sliding change process are recorded and screened,which are highly correlated with the ground characteristics.Through the LSTM neural network,the feature data are classified to identify the current ground characteristics.This method has the advantages of simple acquisition equipment,easy access to signals,and low computility requirements.Simulation results show that the method can accurately classify the vehicle driving state into three typical adhesion surfaces:high,medium and low.

关 键 词：履带滑转 长短期记忆神经网络 路面辨识 

分 类 号：TJ811[兵器科学与技术—武器系统与运用工程]