基于深度置信网络甘蔗转运车状态识别

Status Identification of Sugarcane Transporter based on Deep Belief Network

针对丘陵地区甘蔗转运车田间转运作业容易发生车身倾斜、过载等不稳定的危险状态,本文在课题组设计的一种适用于丘陵地区作业的双剪叉式甘蔗转运车模型基础上,针对转运车的工作状态不易判断、转运车的平衡调整难以掌控等问题,提出了一种基于深度置信网络的甘蔗转运车不稳定性状态的快速检测判断方法。通过在转运车车架位置安装一个无线压电式加速度传感器检测转运工作过程的振动信号,并在剪叉举升架关键节点粘贴直角花应变片采集节点应力,通过对振动信号进行数据预处理,通过训练深度置信网络,并对转运车状态进行辨别。试验分析表明,采用经预处理后的数据对深度置信网络进行训练,在测试集上对转运车状态预测的平均准确率可以达到90.90%,对倾斜、过载等不稳定危险状态的预测准确率可达100%,时间约为0.1 s,证明了该甘蔗转运车工作不稳定状态实时预测方法的有效性,可实现对甘蔗转运车状态的实时监测,为丘陵地区甘蔗转运车的实时监测控制奠定了基础。

Aiming at the unstable dangerous state such as body tilt and overload in the field transfer operation of sugar cane transfer vehicle in hilly area,this paper puts forward a rapid detection and judgment method based on the unstable state of sugarcane transfer vehicle the model of double shear fork sugarcane transfer vehicle designed by the group,which is not easy to judge the working status and the balance adjustment of the transfer vehicle.By installing a wireless ballast acceleration sensor in the carrier frame position to detect the vibration signal of the transshipment process,and pasting the right angle flower strain sheet to the key node of the shear lift frame to collect the node stress,pre-processing the vibration signal data,training the deep confidence network,the identifying the status of the transshipment vehicle are carried out.The experimental analysis results show that the average accuracy of the prediction of the state of the transporter can reach 90.90%in the test set,the prediction accuracy of the unstable dangerous state such as tilt and overload can reach 100%,and the time is about 0.1 s,which proves the effectiveness of the real-time prediction method of the unstable state of the sugarcane transfer vehicle,and can realize the real-time monitoring of the state of the sugarcane transfer vehicle in the hilly area.