基于混合策略的电子鼻在线漂移抑制算法

Online Drift Suppression Algorithm for Electronic Nose Based on Hybrid Strategy

传感器自身的漂移问题会降低电子鼻系统检测的准确度以及系统寿命。为了解决该问题,提出一种基于混合策略的电子鼻在线漂移抑制算法。该算法结合主动学习与分类器集成方法构建算法框架,同时在主动学习样本选择的过程中结合样本的混合信息熵以及样本的相似性构建了样本评价的混合策略,混合策略进行样本筛选以样本的混合信息为基础。经对比,算法相比于其他方法,在在线漂移检测中,分类准确率分别能提高2.8、5.2、6.2、6.6个百分点。

The drift of the sensor can reduce the detection accuracy and system life of the electronic nose system.In order to solve this problem,this paper proposed an online drift suppression algorithm for electronic nose based on a hybrid strategy.The algorithm combined the active learning and classifier integration method to construct the algorithm framework.Meanwhile,in the process of active learning sample selection,the mixed information entropy of samples and the similarity of samples were combined to construct the mixed strategy of sample evaluation,the mixed strategy was based on the mixed information of samples.Compared with other methods,the algorithm can improve the classification accuracy in online drift detection by 2.8,5.2,6.2 and 6.6 percentage respectively.