Research on Recognition Model with Random Forest and Entropy Weight for Chemical Gas Sensor Array

Abstract

Electronic nose is one of the research hotspots in the field of engineering. In order to solve the chemical gas recognition problem of electronic nose (that is, sensor array), we designed and established a recognition model based on random forest, entropy weight and bootstrap aggregating. The model has been successfully applied to the UCI Gas Sensor Array Drift Dataset and achieved excellent effect and reliability, avoiding the adverse effects caused by drift problem and unbalanced data distribution to a certain extent. The design and implementation method of the recognition model has certain reference value to the research of related fields.