Application Research Of Soft Computing In Pattern Recognition Of Sedimentary Microfacies

The analysis of reservoir sedimentary facies is significant for studying reservoir physical property structure,reasonable producing,analysing remaining resources and identifying atectonic subtle reservoir,and it's one of important methods often used in exploring and developing gas and oil resources.Based on the phase sequence grading rule and sedimentary environment characteristics,sedimentary identification is done by hands through observing the cores and logging data in conventional research of sedimentary microfacies.But this method is less efficient,the result accuracy is influenced by subjective factors.Therefore,looking for an effective identification method of sedimentary microfacies to replace the artificial recognition has become the important content of modern sedimentary facies study.For the characteristics that the information is imprecise,inconsistent and incomplete in pattern recognition,this dissertation constructs a reservoir sedimentary microfacies recognition model based on Rough Set and improved Support Vector Machine.First,the redundant or unimportant attributes are deleted by analysing the relationship among characteristic parameters,under the condition of keeping unchanged the classification ability and decision-making ability of characteristic parameters,provided optimized characteristic parameters for SVM identification model;Second,SVM is used to analyse the highly non-linear relationship between characteristic parameter and sedimentary microfacies based on the advantages of it is suitable to solve the problem of small sample,nonlinear,high dimension;Finally,in the process of sedimentary microfacies identification,combined the advantages of PSO and GA,the GA-PSO is used to optimize the parameters of SVM.The analysis of experimental results show that the identification capability of SVM which is optimized by the hybrid algorithm in terms of accuracy and running time is better than SVM's optimized by PSO or GA,the recognition accuracy of the reservoir sedimentary microfacies recognition model based on RS and SVM reachs 70% or more.The identification model based on RS and SVM established in this thesis can effectively solve the reservoir sedimentary microfacies recognition,it's an effective,reliable and potential sedimentary microfacies recognition method and can provide help for the oil field exploration anddevelopment.