摘要
根据煤矿井下粉尘衍射光角谱特征,研制出利用模式识别算法实时井下检测的光电粉尘传感器。为了提高测量精度,对误差进行分析,提出模式分类算法,利用模式的特征向量分布的差异对模式进行了分类,检测时可快速从全域找到最贴近的模式,在此基础上,本文又提出加入修补函数的模式识别反演算法,测量时可动态选择合适的修补模式对选定模式进行局域修补,使修补后模式的特征向量与待测尘样信号的方差和明显变小。仿真实验表明,改进的算法使传感器测量精度得到很大提高,快速性也得到有力保障。
The pattern recognition algorithm was presented according to dust diffraction angular.To improve the precision,the error was analyzed and patterns was sorted by eigenvectors difference.It took shorter time to recognize the optimum.Amendment function had been presented,so amendment pattern improved the optimum.The variance between the pattern eigenvectors mended and signals becomes less.The simulation indicates the precision and speed had been improved.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2006年第z1期359-360,共2页
Chinese Journal of Scientific Instrument
关键词
衍射光角谱
模式识别
模式分类
模式修补
angular distribution pattern recognition pattern sorting pattern amendment
