地质科技情报2011，Vol.30Issue(5)：127-131,138,6.
基于随机游走过程的多点地质统计学建模方法
A Random Walk-Based Multiple-Point Statistics Modeling Methods
摘要
Abstract
The reasonable reconstruction of the fluvial reservoir architecture has always been a difficulty in reservoir modeling. After analyzing the present situation and characteristics of domestic and international modeling methods,we propose a new method called RMPS. It synthesizes the advantage of random walk, better reproduction of mainstream of fluvial, and the superiority of MPS,representation of micro-shape of the channel in fluvial reservoir simulation. Firstly, the traditional random walk process algorithm is modified, and a new random walk is put forward, which contains a seven direction searching for channel nodes for calculating seven transition probability and a four direction channel source searching for solving the prediction of high curvature and network river. The modified method can simulate all types of fluvial mainstream line. On the basis of the predicted channel mainstream, the channel mainstream data are used to restrain the data events of multiple-point statistics so as to promote the rational choice of data events and to establish more accurate geological models. To examine the validation of the proposed algorithm,the satellite images of contemporary rivers are taken to make some simulations. The results show that RMPS can be better represented of the channel architecture than multiple-point statistic(MPS). Sensitivity analyses indicate that the method has certain stabilities. This method of stochastic reservoir modeling suggests that it will produce great significance for stochastic reservoir modeling and for establishing high-precision reservoir geological models.关键词
随机游走/多点地质统计学/GoogleEarth/河流相/主流线/迁移概率Key words
random walk/ multiple-point statistics/ GoogleEarth/ fluvial faces mainstream/ transition probability引用本文复制引用
石书缘,尹艳树,和景阳,冯文杰..基于随机游走过程的多点地质统计学建模方法[J].地质科技情报,2011,30(5):127-131,138,6.基金项目
国家自然科学基金项目(40902043) （40902043）
