Research On The Classification And Recognition Of Polyhalite In Central Sichuan Based On Pattern Recognition

As an important strategic resource,potassium is quite scarce in China.China's import of sylvine in 2012 ranked second in the world after India.As a result,we should draw more attention to discovery of our own potash resources and gradually strengthen the ability of self-sufficiency.Polyhalite is a valuable potassium minerals,there is a very high content of potassium.The reserves of polyhalite are very abundant in the Sichuan basin,especially in the central Sichuan Basin.However,because of interbedding with other lithology,the most polyhalite reservoirs are not pure.Therefore,the applicability of conventional logging identification methods in the study area is poor,which is not conducive to the later work of potash exploration and reserves evaluation.Therefore,it is of significance to carry out the research on the method of identifying the polyhalite.This paper is based on the polyhalite reservoir in the central Sichuan Basin with reference to the core analysis,well logging data and mud logging data.Firstly,a study of logging reaction of polyhalite was conducted.And identified the polyhalite by the method of overlapping and cross plot.Furthermore,a quality assessment and comparison of the identification and the mud logging data to found that the conventional logging identification method has a certain degree of discrimination,but the recognition accuracy is not high due to the impure factors.Secondly,this paper used three different pattern recognition methods,including support vector machine,BP neural network and extreme learning machine,which is used to structure the model for identifying the polyhalite and logging data as input.On the basis of identification models,identifing the lithology such as polyhalite,dolomite,gypsum,halite and mung bean rock.The results show that the recognition effect is good.Thirdly,based on the identification models which are built,combined with the influence to log response of the content of the polyhalite on the the gypsum and the mixed gypsum,constructing the further polyhalite classification model.The result is good,the recognition accuracy is above 80%.Through the comparison of four methods for the identification of polyhalite,we can find that the application of conventional logging identification methods in the identification of the bittern is poor.The pattern recognition method not only can accurately identify the polyhalite reservoir,but also can precisely further classify the polyhalite reservoir.It can be seen that the pattern recognition method has a good application prospect in potash exploration.