There are some problems in conventional seismic facies analysis methods, such as easily plunge into local optimal solution, low sensitivity and without using prior knowledge. To solve the above-mentioned problems, we propose a pre-stack texture-based semi-supervised seismic facies analysis method with global optimization. Firstly, the pre-stack seismic texture attributes are introduced to highlighting the information of micro-spatial and amplitude variation with azimuth/offset in seismic reflection data. Then, the self-organizing map (SOM) neural network is used to compress a large amount of redundant information of the samples on the premise of maintaining the topology of the data. Finally, the artificial bee colony (ABC) algorithm is used to realize the global optimization of the clustering of neurons in the SOM output layer under the constraints of prior knowledge. Besides, according to the probability estimation results based on the probabilistic neural network (PNN), we define the confidence measures to quantitative analysis the classification results. The synthetic test and practical application results show that the proposed method can not only significantly improve the recognition ability of the seismic microfacies, but also improve the horizontal resolution and the accuracy of the seismic facies map. These satisfactory results illustrate the proposed method is an effective tool for seismic facies analysis.
分享
引用
