Abstract: According to high difficult development caused by high geostress and low permeability features of the coalbed methane in deep seam, the paper analyze d the rock core test data of the coalbed methane well in the south part of Qinshui Basin and the logging, well test, fracturing, production and other actual informations of the coalbed methane well and studied the relationship between the physical property parameters of main reservoir and depth. The study results showed that the variatio n law of different reservoir with seam depth would be different and would have jumped type variation features. The variation value of the inflection point would not be a certain buried depth value. The results of the changed inflection point of the physical property parameter simulated with BP neural network showed that due to the select ed key parameters different, the obtained physical property with the changed inflection point of the depth would be different. The inflection point value of the deep seam depth with the mechanics parameters as the key parameters was 1 043 m. The deep seam depth inflection point value of the particular physical property parameters wa s 659 ~ 950 m. The deep depth inflection point value with production factors as the key parameters was 927 ~ 1 171 m.