论文详情
针对如何从勘探开发数据中获取更多有用信息问题,提出了一种数据驱动数学模型进化的方法。该方法将数据与数学模型本身关联起来,经过初始模型建立、数据整理、相关性分析、模型改进等过程生成与实测数据相吻合、具有明确地质意义的数学模型,达到加深地质规律认识的目的。将该方法应用于南海东部泥岩速度与砂质含量之间关系的分析,利用逐步优化建立的初始数学模型,获得与实际数据更吻合的进化模型,进一步将进化模型与地质知识关联,提出了泥岩差异压实机理模型。泥岩差异压实机理模型认为泥岩中的砂质颗粒会降低泥质压实程度,使泥岩速度变化的规律复杂化。研究区具体表现为同一深度下不同砂质含量的泥岩中泥质的速度差异超过1000m/s,泥岩速度在浅层随泥质含量增加而非线性减小,在深层又随泥质含量增加而非线性增大。研究结果表明,利用数据驱动数学模型进化的思路挖掘地质油藏数据中隐含规律的方法可行且有效,有一定的参考价值。
In this study,a data-driven method of mathematical model evolution was developed,with the aim to enhance the extraction of useful information from exploration and development data.The proposed method connects the data with the mathematical model itself and establishes a mathematical model with definite geological significance that is consistent with the measured data.This is achieved through initial model building,data sorting,correlation analysis,and model improvement,and results in a deeper understanding of the geological laws.The method was applied to the analysis of the velocity variation law in a sand-bearing mudstone based on logging data from multiple wells in a study area.A differential compaction petrophysical model for the mudstone was obtained by combining the evolution model with geological knowledge.This petrophysical model indicated that an increase in sand content can reduce the compaction of the argillaceous component,resulting in a velocity that increases with the sand content increasing in the shallow layer and decreases in the deeper layer.