基于CNN-LSTM的重型自卸车侧翻预警模型  

Rollover Warning Study of Heavy Dump Truck Based on CNN-LSTM

作　　者：汪佳铭 胡明茂 师国东 朱天民 WANG Jiaming;HU Mingmao;SHI Guodong;ZHU Tianmin(School of Mechanical Engineering,Hubei Institute of Automotive Technology,Shiyan 442000,China)

机构地区：[1]湖北汽车工业学院机械工程学院,湖北十堰442000

出　　处：《滨州学院学报》2024年第2期81-89,共9页Journal of Binzhou University

基　　金：湖北省重点研发计划项目(2020BAA005);工信部工业互联网创新发展工程项目(TC200A00W,TC200802C)。

摘　　要：为解决重型自卸车的侧翻预警问题,基于CNN-LSTM神经网络构造了重型自卸车的侧翻预警模型。利用Trucksim与MATLAB/Simulink搭建了重型自卸车仿真模型,以横向载荷转移率等于±0.85为侧翻阈值,提取了不同工况下的车辆运行参数,利用车辆运行参数,训练CNN-LSTM重型自卸车侧翻预警模型,并分别与基于CNN、LSTM搭建的预警模型对比。结果表明:CNN-LSTM重型自卸车侧翻预警模型预测准确率为98.31%;感受性曲线的曲线下面积为0.999,高于由单一神经网络所搭建的侧翻预警模型。In order to solve the rollover warning problem of heavy dump truck,based on the CNN-LSTM neural network,a rollover early warning model for heavy-duty dump trucks is constructed to achieve real-time determination under different working conditions.Using Trucksim and MATLAB/Simulink joint simulation,a heavy dump truck simulation model is built.With the lateral load transfer rate equal to±0.85 as the rollover threshold,the vehicle operating parameters under different working conditions are extracted and the vehicle operating parameters are used.The CNN-LSTM heavy-duty dump truck rollover early warning model is trained and compared with the early warning models based on CNN and LSTM respectively.The results show that the prediction accuracy of CNN-LSTM heavy dump truck rollover warning model is 98.31%,and the area under ROC curve is 0.999,which is higher than the rollover warning model built by a single neural network.It is clear that the CNN-LSTM heavy dump truck rollover warning model has some advance warning significance and is useful for reducing the occurrence of heavy dump truck rollover accidents.

关 键 词：重型自卸车 CNN LSTM神经网络 横向载荷转移率 侧翻预警模型 仿真 

分 类 号：U469.4[机械工程—车辆工程]