基于RBF神经网络识别路面谱的新方法  

A New Method of Road Surface Spectrum Identification Based on RBF Neural Network

作　　者：张丽霞[1] 赵又群[1] 吴杰[1] 尹浩[1] 

机构地区：[1]南京航空航天大学能源与动力学院,江苏南京210016

出　　处：《公路交通科技》2007年第6期135-138,共4页Journal of Highway and Transportation Research and Development

基　　金：高等学校博士学科点专项科研基金资助项目(20040287004);江苏省博士后科研基金资助项目(2004300)

摘　　要：路面不平度是车辆行驶中振动的重要激励。为了识别路面不平度的功率谱密度函数(路面谱),提出了一种基于径向基函数(RBF)神经网络识别路面谱的新方法。该方法以7自由度汽车振动模型为基础,以MATLAB软件仿真得到的汽车车身质心垂直加速度谱为神经网络理想输入样本,以GB7031-86建议的路面谱为神经网络理想输出样本,应用RBF神经网络建立汽车车身质心垂直加速度谱和路面谱之间的非线性映射模型。另取一组仿真得到的车身质心垂直加速度谱代入已训练好的网络进行路面谱识别。结果表明:该方法具有较强的抗噪声能力和较理想的识别精度,识别的路面谱与拟合的路面谱吻合一致。Road surface roughness is the important vibration excitation factor in the vehicle running. In order to identify power spectrum density function of road surface roughness （road surface spectrum）, based on Radial Basis Function （RBF） neural networks, a new method of road surface spectrum identification is put forward, Based on seven-degree-of-freedom vehicle vibration model, vehicle body centroid vertical acceleration spectrum which is got by MATLAB simulation is regarded as neural networks ideal input sample, road surface spectrum proposed by GB7031-86 is regarded as neural networks ideal output sample, the nonlinear mapping relation between vehicle body centroid vertical acceleration spectrum and roarl surface spectrum is found by RBF neural networks. Another vehicle body centroid vertical acceleration spectrum calculated by simulation is used to identify road surface spectrum by trained networks.The resuhs show that the method has better ability of anti-noise and ideal identification accuracy, road surface spectrum of identification fits the imitated road surface spectator.

关 键 词：汽车工程 路面谱 径向基函数神经网络 载荷识别 

分 类 号：U416.4[交通运输工程—道路与铁道工程]