一种基于一维卷积神经网络的烟叶霉变预测方法  

A Prediction Method of Tobacco Mildew Based on 1D Convolution Neural Network

作　　者：翟乃琦 云利军[1,2] 叶志霞[1] 卢琳 ZHAI Nai-qi;YUN Li-jun;YE Zhi-xia;LU Lin(School of Information,Yunnan Normal University,Kunming 650500,China;Yunnan Province Key Laboratory of Optoelectronic Information Technology,Yunnan Normal University,Kunming 650500,China;Yunnan Tobacco Leaf Company,Kunming 650500,China)

机构地区：[1]云南师范大学信息学院,云南昆明650500 [2]云南师范大学云南省光电信息技术重点实验室,云南昆明650500 [3]云南省烟草烟叶公司,云南昆明650500

出　　处：《云南师范大学学报（自然科学版）》2021年第3期23-27,共5页Journal of Yunnan Normal University:Natural Sciences Edition

基　　金：云南省应用基础研究计划资助项目(2018FA033);云南师范大学研究生科研创新基金资助项目(ysdyjs2019152).

摘　　要：构建了基于一维卷积神经网络的仓储烟叶霉变预测模型,以烟叶样本霉变过程中产生的特征气体浓度为学习数据,对仓储烟叶霉变进行预测;实验结果表明:与传统的BP神经网络模型相比,所构建的模型预测效果更好.A mildew prediction model for tobacco in storage was established based on one-dimensional convolution neural network.The characteristic gas concentration produced during the process of tobacco sample mildew was adopted as the learning data to predict the storage tobacco mildew.The test results show that the prediction effect of the constructed model is better than the traditional BP neural network model.

关 键 词：卷积神经网络 烟叶霉变 电子鼻 传感器 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]