皮棉轧工质量分级模型的建立  

Building of Lint Cotton Ginning Quality Grading Model

作　　者：肖春燕 侯加林[2] 

机构地区：[1]山东省棉花生产技术指导站,山东济南250013 [2]山东农业大学机械与电子工程学院,山东泰安271018

出　　处：《山东农业科学》2017年第10期134-138,共5页Shandong Agricultural Sciences

基　　金：山东省科技发展计划项目(2012GGB01084)

摘　　要：根据皮棉外观形态粗糙程度、所含疵点种类及数量的多少,我国将皮棉轧工质量分为好、中、差三个等级。目前我国皮棉轧工质量的分级是由人工完成的,检验时,由工作人员手持样本,与制作的国家等级实物标准对照,从而得出各样本所属的轧工质量档次。但该方法存在很多缺点,如:制作的等级实物标准样本会随着时间的推移发生改变,进而影响检测结果;由于检测人员的主观性差异,不同的检测人员在检测同一个样本时,可能会得出不同的结论。而轧工质量层次直接影响着棉花生产的经济效益。基于此,本研究首先利用皮棉感官质量分级仪对大量的白棉二级皮棉进行纹理提取,然后利用SPSS Modeler建模软件建立皮棉纹理特征与轧工质量的人工神经网络模型,经专家人工分级验证,该模型分级效果较好,准确性较高,利用此模型可以提高皮棉轧工质量分级的效率和准确性。The ginning quality is divided into 3 levels（ good,middle,bad） in accordance with roughness of appearance and types and defect number of lint cotton. At present,the ginning quality grading of lint cotton is finished by manual work. During the test,the workers hand the samples and contrast the tested cotton with the national-level cotton material standard samples so as to obtain the ginning quality grade. But the method has many disadvantages,for example,the national-level cotton material standard samples may change with time flying,or different workers may obtain different conclusions due to subjective differences and so on. However the ginning quality grade can directly influence the cotton economic benefits. In this paper,the texture extraction test of second-level white cotton was carried out by lint cotton sensory quality grading instrument. Then the neural network model about lint cotton texture features with ginning quality was built with the IBM SPSS Modeler. After identifyed by experts,the model had better grading effect and higher accuracy.So the model could improve the efficiency and accuracy of ginning quality grading of lint cotton.

关 键 词：棉花纹理 轧工质量分级 SPSS MODELER 神经网络模型 

分 类 号：S562.092[农业科学—作物学]