Study On Methods Of Fuzzy Pattern Recognition And The Applications In Oil And Gas Prediction

In the area of oil and gas prediction, the common used means is geochemical logging when classifying them by pyrolysis parameters. But the result of classification is not very satisfied and the precision is not very high. So we work over the methods of fuzzy pattern recognition and apply these methods to the oil and gas prediction of Tarim Basin. And the result shows that these methods are more stable and more effective than geochemical logging.We develop our research on fuzzy pattern recognition from three aspects: one is the technique of fuzzy pattern recognition based on modified templet method, another is the measure based on fuzzy clustering, and the third is fuzzy neural network. Firstly, we seriously analyzed the templet method and found the insufficiency. And the improved technique is integrated with PLS regression, which leads to a new measure called FPLS. And this method has great identify power on the chart of pyrolysis parameters of Tarim Basin. Secondly, in the research of fuzzy clustering methods, we gave out several useful formulae and an improved fuzzy transitive closure method according to the degrees of influence of various pyrolysis parameters. We also brought up a modified FCM algorithm according to the different functions of the samples. Finally, we produced a fuzzy neural network model based on the concept of fuzzy neural network. We also got a satisfied result in the application of Tarim Basin.Therefore, the production of this dissertation has great significance both in theory and in practice, especially in the research of oil and gas exploration.