支持向量机在回归预测有机碳含量中的应用研究——以川东南地区为例

Prediction of Organic Carbon Content Based on Rendezvous Graph and Support Vector Machine Regression—A Case Study of Maokou Formation in Southeast Sichuan Province

摘要:  四川盆地川东南地区为浅水–深水陆棚沉积环境，区内烃源岩发育，烃源岩富含大量有机质，最开始采用化学方法进行分析和判定，但是评价结果难以满足日益增长的生产需求，所以有机碳含量(TOC)计算模型作为一种有效的识别方式得到了广泛的应用。本文利用交会图法选出贡献率高的三条测井曲线：声波时差、自然伽马和深侧向电阻率，然后输入三条测井曲线值建立有机碳支持向量机回归预测模型。结果表明：有机碳含量在0.5以上时预测效果较好，当有机碳含量低于0.5时模型的精确度还需提高。

Abstract: The southeast area of Sichuan Basin is a shallow-deep-water shelf sedimentary environment, where source rocks are developed and rich in organic matter. At first, chemical methods are used to analyze and determine the source rocks, and the source rocks are rich in organic matter. However, the evaluation results are difficult to meet the increasing demand for production, so the (TOC) calculation model of organic carbon content has been widely used as an effective identification method. In this paper, three log curves with high contribution rate are selected by cross plot method: acoustic time difference, natural gamma and deep lateral resistivity. Then three log curves are inputted to establish organic carbon support vector machine regression prediction model. The results show that the prediction effect is better when the organic carbon content is above 0.5, and the accuracy of the model needs to be improved when the organic carbon content is lower than 0.5.