基于机器学习的脆性指数直接反演方法

Nonlinear direct inversion for brittleness index based on machine learning

常规脆性指数预测方法是基于叠前振幅随偏移距的变化(AVO)反演,间接计算过程中会产生累积误差。提出了一种脆性指数直接反演方法,该方法使用BI_Zoeppritz方程来表示地震数据与脆性指数之间的关系,通过机器学习LSSVM算法建立非线性反演模型,直接反演出脆性指数。实际资料测试结果表明:该方法具有更高的精度和更强的抗噪声能力,利用该方法进行脆性指数的直接反演是可行的。

The conventional prediction method of brittleness index is based on pre-stack AVO inversion,which will produce accumulated error in the indirect calculation process.This paper presents a direct inversion method of brittleness index.The BI_Zoeppritz equation is used to express the relationship between seismic data and brittleness index.The nonlinear inversion model is established by machine learning LSSVM algorithm,and the brittleness index is directly inverted.The test results of actual data show that this method has higher accuracy and stronger anti-noise ability.It is feasible to use this method to directly inverse the brittleness index.