Recognition of infrared spectrum data of coal mine gas based on multiple hyperplanes classifier method

In light of the limitation of traditional linear searching way, which uses one hyperplane to classify in spectral recognition of coal mine gas, multiple hyperplanes of dendriform classifier method are introduced to classify in this paper, which has the good classifying effect on gas in complicated background environment. This article has introduced the principle of dendriform partition linear classifier, and used the dendriform classifier to do algorithm training and identificating classification. Applying to classify and identify complicated samples of remote sensing infrared spectrum datas of coal mine gas, the experimental results indicated that, in the same numbers of training sample, this method not only has less training iteration than linear classification, but also the weights calculated by it have better examination results than linear classification for the whole examination datas.