摘要
为解决汽车超载运行和相应的运输业管理问题,本文中提出了一套基于BP神经网络的车载称重系统。通过检测车桥随载荷量变化而产生的微小变形,设计了2阶低通滤波和数字滤波算法,以提取有效载荷信息,利用BP神经网络建立载荷模型,并根据在某轻型厢式货车上进行的试装和加载试验得到的样本数据,在Matlab神经网络工具箱中,采用Levenberg-Marquardt学习算法完成了神经网络的学习、检验和预测。结果表明:预测载荷误差在5%以内,满足工程要求,方案可行。
To solve the problem ol truck＇s overload running and related transportation management, a vehi-e on-board weighing system based on BP neural network is proposed. By measuring the tiny deformation on truck xles caused by load, second-order lowpass filtering and digital filtering algorithms are designed to extract effective oad data. Then load model is built by using BP neural network,and according to the sample data obtained in load-ng test on a light van with weighing system installed,a process ol learning,testing and prediction ol neural network completed by adopting Levenberg-Marquardt learning algorithm in Matlab neural network tool box. The results how that the error ol predicted load is within 5%,meeting engineering requirements and indicating the feasibility ol roposed scheme.
出处
《汽车工程》
EI
CSCD
北大核心
2017年第5期599-605,共7页
Automotive Engineering
基金
国家自然科学基金(61371076)资助
