A Novel Multisensor Detection System Design for Odor Classification

In order to cope with the demand for odor detection of herbal medicine, a multisensor detection system consisting mainly of metal-oxide sensors (MOS) is designed. For the constructed sensor array, the stepped chamber for sensors is designed based on the linear programming principle and simulated by the finite element analysis method, which greatly reduces the volume of the chamber. The simulations and experiments are provided to demonstrate the wide range of flow velocity universality with good sensor response. Besides, a Chebyshev polynomial-based weighted discriminant extreme learning machine (C-WDELM) is proposed to overcome the negative impact on the pattern recognition result due to the unbalanced number of samples of different classes in the dataset. Moreover, the odor data of six classes of herbal medicine are collected by our system. They are analyzed with various machine learning algorithms, and the algorithm we proposed obtains a mean accuracy of 0.9652 and a mean sensitivity of 0.9673, which are higher than other classical algorithms. It validates the effectiveness of the designed system and shows the potential of the system to be applied in the intelligent identification of herbal medicine.