Research On Stratigraphic Lithologic Logging Recognition Based On Fuzzy Clustering And Development Of Supporting Software

Because of rising demand for energy during the development of modern economic, the requirements of petroleum geological exploration techniques increase. Using more ways and methods, the application of multidisciplinary becomes the trend to solve many technical problems in the field of oil-gas exploration.The emergence of artificial intelligence powerful promoted the development of the lithologic identification. Currently, the BP neural network, support vector machine (SVM), intelligent methods such as principal component analysis is widely used in the field of lithology identification, and has achieved remarkable results.As the expansion of traditional and artificial intelligence, lithologic identification of Fuzzy Clustering Method can make full use of natural gamma, neutron, sonic, density and resistivity logging data which contain abundant lithologic logging information, divide into raw data and determine the center of classification under the condition without rock core automatically, thus discriminate on all the sample data, improve the working efficiency and the accuracy of lithology classification. Around the topic of using Fuzzy Clustering Method to identify lithology. Thesis carried out the research as following four parts:(1)Have studied the principle of fuzzy clustering technology analysis and data preprocessing methods. The former includes three main aspects, fuzzy cluster analysis, fuzzy pattern recognition and fuzzy comprehensive evaluation. The latter selects the standard deviation and the range transformation as standardized ways in data processing.(2) Have worked on choosing lithology identification method. Compared the common lithology identification method with the fuzzy clustering recognition method. The paper selected the fuzzy clustering recognition method.(3) The selection method of the Characterization of lithologic logging parameters. Based on support vector machine (SVM) theory and logging response characteristics parameter analysis method to study to select sensitive lithology parameters.(4) Fuzzy clustering lithology recognition software compiling and application. Choosing formation lithology logging sensitive parameters as input variables, which is based on the fuzzy clustering theory. In addition, the software has been debugged and applicated. With all the above methods being made, and transformed the mathematical model into software application, research shows that, by using fuzzy clustering software and map for lithologic identification of lithology ratio pattern and lithologic distribution pattern, the macro lithologic distribution trend can be reflected. As a result, the coincidence rate is high. Choosing different degree calculation model can meet the needs of the different work area lithology identification, and was less affected by artificial factors. Through clustering analysis of significant assessment, which can ensure the reliability of the lithology classification results. After software debugging, application and comparing to the logging data of work area, it shows that using Fuzzy Clustering Method to identify lithology has high accuracy can satisfy the actual requirment of lithology recognition, which have a good application prospect and popularization value.