Geochemical Comprehensive Characteristic Anomaly Extraction Based On KPCA And Multifractal

In prospecting practice, exploration geochemistry data is an importantinformation source, one of the goals of the geochemical data processing is usingmathematical method to extract the useful information from a large number ofobservation data, and reveal the indicator variables (mainly chemical element)intrinsically linked with a variety of geological phenomena, look for mineralresources for exploration geochemistry and geological problems such as technicalservice method. This article focuses on introducing a new comprehensivegeochemical anomaly extraction method.Principal component analysis is an effective method of comprehensivegeochemical data feature extraction, which is constructed a new variables instead ofthe original variables. It can effectively reflect the compositive information of originalvariables, and indicate the exploration geochemistry of magmatite and geneticrelationship. However, the complexity of geological systems and multiple stagemineralization stage often lead to the nonlinear distribution of multivariategeochemical data. Therefore, compared with the traditional principal componentanalysis, the nonlinear principal component analysis is more suitable for extracting ofthe multivariate geochemical data.The Study of spatial distribution law of geochemical elements and is the core ofexploration geochemistry. It is the important ways to reveal the geochemical elementenrichment and change rule of space. Geochemical data is uncertainty in sampling andanalysis stage, and the distribution of elements in the earth’s crust area is randomnessand irregularity. So, traditional statistical methods often focuses on the frequencydistribution of elements content value, and the measurement of spatial variability ofgeochemical elements is powerless, fractal theory can effectively overcome theshortage of traditional statistical method, it can consider the frequency distribution ofthe geochemical data values, also it can reflect the spatial properties and nonlinearcharacteristic of the geochemical data.Main work of this article is extracting the single element and the synthesisanomaly based on the kernel function of principal component analysis andmultifractal exploration. The principal component analysis basing on kernel function,with the help of a "nuclear techniques" implicitly map the input space to a nonlinearcharacteristics of space. In this space, we carry out principal component analysis ofgeochemical data. The algorithm is in line with the exploration geochemistry datafeatures. We used S-A model of multifractal to get the geochemical anomaly. Thepractice shows that the algorithm meet the data characteristics of explorationgeochemistry, through to the Tibetan fierce lang-gilardino-liantang river workspace data stream sediment geochemical research, show that the effect of principalcomponent analysis based on kernel function and multifractal in extracting singleelement geochemical anomaly information and comprehensive anomaly informationis remarkable.