Electronic nose sensor array optimization algorithm based on Mutal Information Feature Selection

Electronic nose sensor arrays are currently facing the problems of too many sensors and high data dimensions, which leads to bloated electronic nose systems and poor accuracy of pattern recognition algorithms. By selecting the optimal subset of sensors in the electronic nose sensing array through a feature selection algorithm based on mutual information, it is possible to reduce the volume of the electronic nose system and obtain better pattern recognition accuracy. At present, the traditional mutual information feature selection algorithm does not consider the characteristics of the gas and the sensor, and the recognition accuracy of the feature subset obtained by the electronic nose system is low. In this paper, a mutual information feature selection algorithm based on the performance weight of the electronic nose system is proposed. This algorithm can measure the features of distinguishability, redundancy and sensitivity to filter out a subset of sensors that are more suitable for the electronic nose array system. The method proposed is applied to the flow modulation sensor array data set and wound bacteria electronic nose data set for sensor screening, and the obtained feature subset achieves higher accuracy in the same pattern recognition algorithm.