摘要
电子鼻在环境温度影响下会出现气体数据漂移现象,传感器在特征优化等流程中,可能会受到漂移因素的影响,因此提出一种部分补偿的去漂移补偿方式,在减少补偿模型特征复杂度的同时,保留被漂移因素影响较小的原传感器数据集共同参与分类。通过构建3种不同的补偿数学模型,对比常规的补偿流程和部分补偿流程的结果差异,证明部分补偿流程能够有效提高电子鼻在深度学习模型中的抗漂移能力,筛选出最佳的补偿模型。结果表明,最佳组合为随机森林的部分补偿组合,训练集和测试集的拟合系数R2分别达到0.94、0.89,均方根误差RMSE分别为0.14、0.20,茶叶分类精度分别提高至98%、96%。
In response to the phenomenon of gas data drift of electronic nose under the influence of ambient temperature,we believe that the sensors may receive the influence of drift factors in the process of feature optimization,etc.Therefore,this paper proposes a partially compensated de-drift compensation method,which reduces the feature complexity of the compensation model while retaining the original sensor dataset that is less affected by the drift factors to participate in the classification together.By constructing three different compensation mathematical models to compare the difference between the results of the conventional compensation process and the partial compensation process,it is demonstrated that the partial compensation process can effectively improve the anti-drift capability of the electronic nose in the deep learning model,and the optimal compensation model is filtered out to further illustrate the optimal compensation approach.The best combination is the partial compensation combination of random forest,and the R2 results of the training set and test set reach 0.94 and 0.89 respectively,while the RMSE is 0.14 and 0.20 respectively,and the resolution of the tea species is improved to 98%and 96%.
作者
蔡旻昊
徐赛
陆华忠
周星星
Cai Minhao;Xu Sai;Lu Huazhong;Zhou Xingxing(College of Engineering,South China Agricultural University,Guangzhou,510642,China;Institute of Facility Agriculture,Guangdong Academy of Agricultural Sciences,Guangzhou,510640,China;Guangdong Academy of Agricultural Sciences,Guangzhou,510640,China)
出处
《中国农机化学报》
北大核心
2026年第1期325-330,345,共7页
Journal of Chinese Agricultural Mechanization
基金
国家重点领域研发计划项目(2022YFD2002203)
广东省乡村振兴专项资金(2024TS—1—2)
广东省国际科技合作项目(2023A0505050129)
广东省农业科学院科技创新战略专项培训项目(农业科研主力军建设)(R2023PY—QN002)
广东省科协青年科技人才培育项目(SKXR2025519)。
关键词
电子鼻
温度补偿
茶叶分类
神经网络
随机森林
e-nose
temperature compensation
tea classification
neural network
random forest
