基于稀疏自编码器的传感器在线漂移补偿算法

Online Sensor Drift Compensation with Sparse Autoencoder

传感器漂移是电子鼻系统中长期存在的问题。漂移现象会造成气体传感器的输出响应异常,使采集到样本的特征分布发生变化,进而导致分类精度明显下降。近年来学者们提出了多种传感器漂移补偿方法,但大多数针对离线场景,在实际应用中存在困难。针对这些问题,提出了一种基于稀疏自编码器的在线漂移补偿算法。该算法能够在仅使用未漂移样本的情况下,通过构建稀疏自编码器和分类器,对漂移样本进行特征增强和有效分类。将文中提出的方法用于公开数据集上进行漂移补偿实验,得到了与已有方法相似甚至更优的分类效果,因此,提出的算法能够有效地实现传感器的在线漂移补偿。

Sensor drift is a long-standing problem in electronic nose(E-nose)systems.The drift phenomenon can cause abnormal response of the gas sensor,which can make the feature distribution of the collected data change and lead a poor classification performance.Although various drift compensation methods have been proposed in recent years,but most of them are offline and not suitable for the practical applications.For those problems mentioned above,a novel online drift compensation based on sparse autoencoder method was proposed in this paper.The algorithm can make feature augmentation and realize the effective classification for drifted data by building sparse autoencoder and classifier only using the samples without drift.The method proposed in this paper is used to perform drift compensation experiments on a public dataset,and the classification results are similar to or better than the existing methods.Therefore,the proposed algorithm can effectively implement the online drift compensation of sensors.