基于小波核极限学习机的烟叶烘烤过程的智能识别
Intelligent recognition of tobacco curing process based on wavelet kernel extreme learning machine
邢玉清 1樊彩霞 2豆根生 3宋朝鹏 4吴莉莉2
作者信息
- 1. 河南农业大学理学院,河南省农业物联网安全与创新工程技术研究中心,河南郑州 450002;信息工程大学网络空间安全学院,河南郑州 450001
- 2. 河南农业大学理学院,河南省农业物联网安全与创新工程技术研究中心,河南郑州 450002
- 3. 河南农业大学农学院,河南郑州 450002
- 4. 河南农业大学烟草学院,河南郑州 450002
- 折叠
摘要
烟叶烘烤设备操作复杂、技术含量高、熟练掌握烟叶烘烤技术人员不足等问题,影响了烟叶的烘烤质量.针对上述问题,本文提出了基于小波核极限学习机的烟叶烘烤过程的智能识别方法.实验中对三段式烘烤过程中的叶片变软、主脉变软、勾尖卷边、小打筒、大打筒和干筋6个烘烤阶段分别提取了颜色、纹理和温湿度特征,组建了 9维特征向量进入小波核极限学习机,通过增量型算法自适应地选择神经元个数,快速准确地识别了 6个阶段,得到了 98.33％的识别率.实验结果表明本文提出的基于小波核极限学习机的烟叶烘烤过程的智能识别方法具有一定的可行性,为研发烟叶烘烤智能调控系统奠定了理论基础.
Abstract
In order to solve the problems of large manpower consumption and unstable baking quality in the tobacco curing process,an intelligent recognition method of tobacco curing process based on wavelet kernel limit learning machine is proposed in this paper.In the experiment,the six baking stages including leaf softening,main vein softening,leaf curling,leaf small rolling,leaf large rolling and stem drying during the three-stage baking process were identified.The color,texture,temperature and humidity features were extracted from the six baking stages and a 9-dimensional feature vector was established to enter the wavelet kernel extreme learning machine.The number of neurons was adaptively selected through an incremental algorithm to identify the six stages quickly and accurately.The recognition rate was 98.33％.The experimental results show that the intelligent recognition method of tobacco curing process based on wavelet kernel extreme learning machine is feasible,which lays a theoretical foundation for the development of tobacco curing intelligent control system.
关键词
极限学习机/小波核函数/烟叶烘烤/特征提取/识别Key words
extreme learning machine/wavelet kernel function/tobacco curing/feature extraction/recognition引用本文复制引用
基金项目
中国烟草总公司科技重点研发项目(110202102007)
中国烟草总公司福建省公司资助项目(2021350000240019)
重庆中烟工业有限责任公司资助项目(YL202202)
河南农业大学自然科学类青年创新基金(KJCX2017A19)
出版年
2024
国家科技期刊平台
NSTL
维普
万方数据