摘要
本文在IBM PC/XT微机上,采用模型参数法和模式识别技术,进行了纱条不匀的分析和生产中故障识别的初步研究。研究表明:使用18阶的AR模型所得到的熟条功率谱曲线,能有效地反映出熟条的不匀结构。采用修正的K-L近邻法使故障识别准确率达到85％。这种模型参数法和故障识别技术对改进纺纱工艺有一定的应用价值。
According to the principle of Fourier transform, Uster Evenness Tester takes form of wave-length spectrum to analyse sliver irregularity. But this paper analysed the irregular structure by AR model, and also carried on an initial research of troubles recognition on PC/XT computer. It concluded that sample length is short, the power spectrum of sliver got by means of 18 th order AR model can be known effectively the irregular structure of sliver; Accurate rate of troubles recognition approaches 85 per cent when using an amended K-near neighbor method whose K is 20; Model-parameter method and pattern recognition technique are worth improving spinning technology.
出处
《中国纺织大学学报》
CSCD
1991年第6期27-39,共13页
Journal of China Textile University
关键词
纺纱
纱条不匀
模式识别
功率谱
fourier transform, sample, power spectrum, pattern recognition, spinning, sliver irregularity.
