摘要
由于在油藏描述领域中储层特性关系比较复杂,用传统方法进行油藏描述存在极大的局限性,为此提出运用学习矢量量化网络方法进行储层岩性识别。在学习矢量量化网络模型和学习算法分析基础上,运用该方法对某油井测井数据进行仿真试验。现场仿真试验结果与实际资料吻合较好,证明该方法在模式识别中具有较强的分类能力。与BP网络相比,学习矢量量化网络具有更明显的优越性。
Because of the complicated relation between reservoir attributes in reservoir description,traditional methods are limited to deal with these problems.Therefore the author proposes learning vector quantization neural network method to identify reservoir lithology.This paper describes the model and learning algorithm of learning vector quantization network,and a simulated test on a certain well to identify well lithology has been conducted.We analysis the simulation result and find that the result accords well with actual log data.It shows that this method has a higher classification capability in pattern recognition.Compared with BP algorithm,learning vector quantization network is more superior.
出处
《特种油气藏》
CAS
CSCD
2007年第5期32-34,共3页
Special Oil & Gas Reservoirs
基金
国家自然科学基金项目"油藏模拟混合软计算系统理论与实用方法研究"的部分研究成果(项目编号:40572082)
