摘要车轮的垂直动载识别路面可以直接应用于汽车室内道路模拟试验的响应模拟,对于汽车的研发和生产具有重要的工程意义。此外还可以判断路面状况,准确决定维修时机,节约大量的维修费用。
本课题对路面激励下汽车的垂直载荷进行了理论分析,并建立了1/4 二自由度汽车振动模型, 通过车轮力传感器采集多种路面的垂直载荷,根据路面动载数据特点,对动载进行小波特征提取,得到了典型路面特征的参数,然后推导出以汽车车身垂直动载特征参数作为输入信号、路面类别作为输出信号的数学模型,利MATLAB 工具箱函数建立了能根据路面激励下的载荷识别路面类型的径向基神经网络型,本文采用改进的动态加速常数协同惯性常数权重的粒子群算法对支持向量机的参数进行优化,最终求得分类结果。另外本课题还将PNN 网络与RBF 网络进行了比较分析, PNN 网络的识别正确率为92%,RBF 网络为32%,这说明PNN 网络对函数的逼近是较优的,可以获得较优解。51243
毕业论文关键词:路面不平度;模式识别;垂直载荷;RBF 神经网络;
Abstract
The vertical dynamic load to identify the road can be applied directly to the road simulated tests indoors, it has important project significance to the research and production of the automobile .In addition it can judge the road conditions, decided accurate repair time and save a lot of maintenance costs.
Vertical load subject to the road excitation motor is analyzed, and the establishment of 1\/4 two degree of freedom vehicle vibration model, the vertical load of wheel force transducer based on the collection of various pavement, pavement dynamic load characteristics of the dynamic load data, wavelet feature extraction, the parameters of typical pavement features, then deduce the car body vertical dynamic load characteristic parameters as the input signal, the road category as a mathematical model of the output signal, established according to the load identification of pavement type pavement under the stimulation of radial basis function neural network using MATLAB toolbox function, this paper adopts the improved dynamic acceleration constant collaboration inertia constant weight particle swarm algorithm to optimize the parameters support vector machine, to obtain the final classification results. Also in this issue also PNN network and RBF network are compared and analyzed, the correct recognition rate of PNN network is 92%, the RBF network is 32%, which indicates that the PNN network is better for function approximation, can obtain a better solution.
Keywords :The roughness ;Model identification ;Vertical load ; RBF neural network ;
目录
第一章 绪论1
1.1 引言1
1.3 基于载荷的路面不平度研究的意义.3
1.4 研究方法..3
1.5 国内外神经网络理论研究的发展现状..3
1.5.1 神经网络概述..4
1.5.2 神经网络的国内外研究现状..5
1.6 本文研究的主要内容.5
第二章 车辆振动力学模型的建立..7
2.1 汽车振动分析.7
2.2 车辆振动力学模型的建立7
第三章 实验数据采集及小波特征提取…11
3.1实验数据采集.11
3.1.1实验目的11
3.1.2 数据采集系统硬件构成11
3.1.3 车轮力传感器介绍.13
3.1.3.1自行研制的车轮力传感器介绍.13
3.1.3.2 车轮传感器原理介绍16
3.1.4 数据采集和分析系统软件介绍22
3.1.5 实验主要内容..23
3.2 小波特征提取.24
第四章 神经网络分类器….28
4.1 神经网络..28
4.1.1 神经网络的学习与训练..28 基于载荷的路面类型识别研究+MATLAB程序:http://www.751com.cn/zidonghua/lunwen_54808.html