基于统计分析的烟支刺破与夹沫外观缺陷差异性检验方法研究  

作　　者：吕献周 祁林[1] 李庆松[1] 蒋铭 唐李洪 纳灵峰 刘继发[2] 

机构地区：[1]红云红河烟草(集团)有限责任公司会泽卷烟厂,云南会泽654299 [2]云南财经大学,昆明650021

出　　处：《科技创新与应用》2025年第5期142-146,共5页Technology Innovation and Application

摘　　要：在烟支生产过程中,保障产品外观质量对于维持消费者信任和市场竞争力至关重要。该研究针对烟支外观缺陷中的刺破和夹沫在微小尺度范围内难以区分的问题,通过面积、横向中心点、偏度和峰度形状特征进行数值化,然后采用非参数的检验方法来评估这些特征在刺破和夹沫缺陷之间的差异性。研究结果对于调整自动化视觉检测系统中的识别算法和分类策略具有重要意义,能够显著提高检测的准确性和效率。该研究不仅强化对烟支外观缺陷特性的理解,也为烟支质量控制和缺陷自动识别系统的设计提供实用的方法论支持,突显高级数据分析在质量保证中的应用潜力。In the production of cigarettes,ensuring appearance quality is crucial for maintaining consumer trust and competitiveness.Aiming at the problem that puncture and foam entrapment in cigarette appearance defects are difficult to distinguish within a micro-scale range,this study numeralizes the shape characteristics of area,transverse center point,skewness and kurtosis,and then uses a non-parametric inspection method to evaluate the differences between puncture and foam entrapment defects.The research results are of great significance for adjusting recognition algorithms and classification strategies in automated vision inspection systems,and can significantly improve the accuracy and efficiency of inspection.This research not only strengthens the understanding of the appearance defect characteristics of cigarettes,but also provides practical support for the design of cigarette quality control and automatic defect identification systems,highlighting the application potential of advanced data analysis in quality assurance.

关 键 词：外观缺陷 正态性检验 Mann-Whitney U检验 蒙特卡洛方法 质量控制 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]