Study On Methods Of Parameters Analysis Of Coalbed Methane By SVM With Logging Data

As one of important unconventional gas resources, the coalbed methane draw many people’s attention. For low porosity and low permeability, low sensitivity to logging, and serious diameter expansion of CBM(Coalbed Methane)and so on, the evolution of CBM with logging data is very dificault. Moreover, we have few good well logging methods on the coalbed methane well exploration currently. As there are many differeces between CBM reservoirs and ordinary reservoirs, the traditional interpretation models cannot be used to the CBM effectively. We introduce the SVM (the Support Vector Machine) to deal with the problems with the logging data.In the analyzing of CBM content, we use the support vector classifier method to analyze the rank, and the support vector regression method to deal with the content, with the logging data. And the results show that we have got what we want, the accuracy of classification above 70%, and the regression is about 10%.In industrial components analyzing, we have used the SVM algorithm, the simple regression, multiple regression and BP neural network to create the math model to interpret these industrial components. The results show that the SVM algorithm has a good performance, and the accuracy is above 80%.During the evaluation of reservoir porosity, the SVM algorithm has been introduced to analyze the porosity, and we also have used the Acoustic porosity model, Density porosity model to make some comparisons, and have used the porosity-permeability model to evaluate the permeability, hoping to provide some advice. The application shows that SVM method also has a good performance in the interpretation of porosity either. And the estimation of permeabilitiy range from 0.3mD to 0.7mD.In the interpretation of gas production, we propose five parameters to descript the gas production of reservoir. They are as follows:the average gas production in low stable, the average gas production in high stable, the lowest gas production, the highest gas production and the average gas production. Where, we use the SVM to interpret these parameters. The results bring us some shines.In the interpretation of water production, we also propose six parameters ether. They are as follows:the average water production in low stable, the average water production in high stable, the lowest water production, the highest water production, the total water production and the average water production. And we alse use the SVM to do the evolutiones. And we make some difference in the practice, which could bring some ways to the interpretation of CBM production with logging data.