Research On Lithologic Identification Based On Bayesian Probability Model

Accurately describing rock types can provide important information for energy mineral exploration and deep structure research.A variety of physical parameters of geological body lithology can be obtained by actual sampling or inversion,including geophysical parameters,geochemical parameters,geological parameters,geographic information,etc.Using geophysical technology,lithology identification can be carried out according to the differences of physical parameters(such as density,magnetic susceptibility,resistivity,velocity,etc.)corresponding to different rocks,and the result of lithology identification using single physical property is not accurate enough,so it has become the development direction and goal of geoscience research to determine the lithology of geological body with multiple parameters.In this paper,Bayesian probability model is used to identify lithology.Bayesian method is a statistical classification method,which relies on probability to classify,and obtains probability through the correlation between geological body parameters.Based on this,we introduce Bayesian probability model based on adaptive kernel density estimation into lithology identification.Based on this,we introduce the Bayesian probability model based on adaptive kernel density estimation into lithology identification.In fact,the probability distribution of Bayesian probability model is based on the probability density of each physical property.To select independent physical property parameters for lithology identification,it is necessary to obtain the probability density of each physical property separately by using probability density algorithm,and obtain the probability of a single sample point relative to each type of lithology by synthesizing the probability density(the sum of probabilities is 1),and the lithology corresponding to the maximum probability is the prediction result.The theoretical model is used to study the classification effect of Bayesian classifier under the probability density of traditional Gaussian classification,kernel density estimation based on fixed bandwidth,and kernel density estimation based on adaptive bandwidth,mainly from the error rate and probability distribution of the model Considering the classification effect,the experiment proves that the lithology recognition result of this method is better.Compared with the traditional Gaussian algorithm and fixed bandwidth kernel density estimation,the classification result obtained by the adaptive bandwidth kernel density estimation is more stable and more accurate.In this paper,the lithology identification method based on Bayesian probability model comprehensively analyzes and studies various geological body parameters,which is of great significance for improving the utilization rate of data and the accuracy of geological body property identification.