摘要
研发了一种机动车尾气专用紫外差分检测仪。采用紫外差分法获得专用CMOS数字化数据,其S测量波长选择181.975 nm,N测量波长选择183.617 nm,采集数据后,采用卷积降维神经网络进行数据分析。使用精度持久性实验测定检测仪的校准周期,并与两种技术成熟的机动车尾气紫外差分测量仪进行比较,其中比较设备A采用多项式回归刚性数据分析算法,比较设备B采用模糊多列神经网络算法。该设备在不进行全面校准的前提下,使用1000次以内测试精度均高于比较设备A和比较设备B。该设备属于一种具有技术替代性的新型紫外差分机动车尾气测量仪器。
A special ultraviolet differential detector for motor vehicle exhaust is developed.The special CMOS digital data is obtained by UV difference method.The S measurement wavelength is 181.975 nm and the N measurement wavelength is 183.617 nm.After collecting the data,the convolution dimension reduction neural network is used for data analysis.The calibration cycle of the detector is measured by precision persistence experiment,and compared with two mature motor vehicle exhaust UV differential measuring instruments.The comparison equipment A adopts polynomial regression rigid data analysis algorithm,and the comparison Equipment B adopts fuzzy multi column neural network algorithm.Without comprehensive calibration,the test accuracy of the device is higher than that of comparison device A and comparison device B within 1000 times.The device belongs to a new type of ultraviolet differential motor vehicle exhaust measuring instrument with technical substitution.
作者
段继伟
付志勇
赵军
钟守君
Duan Jiwei;Fu Zhiyong;Zhao Jun;Zhong Shoujun(China Institute of Testing Technology,Sichuang Chengdu,610021,China)
出处
《机械设计与制造工程》
2021年第10期123-126,共4页
Machine Design and Manufacturing Engineering
关键词
机动车尾气
紫外差分算法
卷积降维神经网络
精度保持
实测比较
vehicle exhaust
ultraviolet difference algorithm
convolution reduced dimension neural network
accuracy maintenance
measurement comparison