河南农业大学学报2025，Vol.59Issue(3)：508-515,8.DOI:10.16445/j.cnki.1000-2340.20250407.001
基于GA-BP神经网络的烟叶打叶风分工艺参数优化
Optimization of processing parameters of threshing and pneumatic separation based on GA-BP neural network
摘要
Abstract
[Objective]This study aims to obtain the optimal technological parameters of flue-cured tobacco leaves in threshing and pneumatic separation and further optimize the leaf structure.[Method]Seven factors were selected,including the rotation speed of the first to fifth-stage threshing and the fre-quency of the 7th and 8th wind selection fans in the threshing and redrying process.Each factor was set at three levels to carry out the orthogonal test.The GA-BP neural network model was constructed with the combination of the process parameters of the orthogonal test result as the data sample set,and the process parameters were further optimized by the NSGA-Ⅱ method.[Result]The optimum process parameters for higher large and medium pieces rate were determined by orthogonal test as follows:the rotation speeds of the first to fifth stages were 493,471,620,798 and 794 r·min-1,respectively,and the fan frequencies of the seventh and eighth stages were 49 and 45 Hz,respectively.The optimum process parameters for lower fragmentation rate and stem content in leaves were as follows:the rotation speeds of the first to fifth stages were 503,489,621,792 and 792 r·min-1,respectively,and the fan frequencies of the seventh and eighth stages were 50 and 46 Hz,respectively.After the optimization of GA-BP neural network model,the rotation speeds of the first to the fifth stages of threshing were respectively set as follows:485,474,620,796 and 794 r·min-1,and the frequency of the seventh and eighth stage fan was 46,49 Hz.Under this condition,the rate of large and medium pieces increased by 1.52 percentage points,and the stem rate and fragmentation rate decreased by 0.09 and 0.08 percentage points,respectively.[Conclusion]On the basis of orthogonal test,the GA-BP neural network model was used to optimize the multi-process parameters,and the leaf structure was more rea-sonable,which provides a reference for improving the processing quality of tobacco leaves.关键词
叶片结构/BP神经网络/遗传算法/打叶风分/参数优化Key words
leaf structure/BP neural network/genetic algorithm/threshing and pneumatic separa-tion/parameter optimization分类
轻工业引用本文复制引用
田斌强,付龙,唐剑宁,刘辉,夏凡,黄沙,刘莉艳,郭筠..基于GA-BP神经网络的烟叶打叶风分工艺参数优化[J].河南农业大学学报,2025,59(3):508-515,8.基金项目
湖北中烟有限责任公司科技项目(2021JCYL4WH2C071) （2021JCYL4WH2C071）
