摘要
针对智能轮胎的实时磨损监测需求,提出了一种新型轮胎磨损检测方法,使用三轴加速度传感器集成设备对轮胎进行加速度波形采集,使用凯撒最大化正态方差法对加速度波形特征进行主成分分析,基于分析结果进行波形特征值提取与筛选,并通过误差反向传播(BP)神经网络对筛选后的特征值数据进行训练,实现轮胎磨损值的实时检测。最后基于实车检测数据进行了测试与对比,结果表明该算法能在较低的算力需求下,将磨损检测的平均误差降低到0.1 mm。
In order to meet the real-time wear monitoring demand of smart tires,this paper presents a new tire wear detection method,which uses a three-axis acceleration sensor integrated device to collect the acceleration waveform of the tire,and uses the Caesars maximum normal variance method to perform principal component analysis on the acceleration waveform characteristics to make waveform feature value extraction and filtering based on the analysis results.The filtered feature value data is trained through the Error Back Propagation(BP)neural network,to achieve real-time detection of tire wear values.The test and comparison based on the real vehicle detection data show that the algorithm can reduce the average error of wear detection to 0.1 mm under low computational power demand.
作者
储昊昀
张峰瑞
张越
张峰
张士文
Chu Haoyun;Zhang Fengrui;Zhang Yue;Zhang Feng;Zhang Shiwen(Shanghai Jiao Tong University,Shanghai 200240)
出处
《汽车技术》
CSCD
北大核心
2023年第1期44-48,共5页
Automobile Technology
关键词
轮胎磨损监测
加速度特征提取
主成分分析
BP神经网络
Tire wear monitoring
Acceleration feature extraction
Principal component analysis
BP neural network
