核磁共振弛豫信号多指数反演新方法

A New Method for Multi-exponential Inversion of NMR Relaxation Signal

孔隙介质核磁共振(NMR)弛豫信号的多指数反演在NMR岩心分析与测井解释中起着关键作用。通过将核磁共振弛豫信号多指数反演问题转化为一个带非负约束的非线性优化问题,先用非负最小二乘方法(NNLS)确定弛豫时间组分数,然后由差分进化(DE)算法求解优化问题。新方法计算非常简单,不依赖于初值,而且容易实现非负约束。计算机模拟和实际数据反演结果表明,该方法对核磁共振弛豫信号分析非常有效,尤其是在信噪比较低的情况下,仍能保证弛豫谱的真实性。

Multi-exponential inversion of NMR relaxation signal plays a key role in core analysis and log interpretation for oil exploration. The inversion is reduced to a non-linear optimization with non-negative constraints. First, the number of groups of relaxation time is determined by non-negative least squares method （NNLS）, and then the non-linear optimization is solved by DE algorithm. The new method is very simple and does not depend on initial value, and is also easy to realize non-negative constraint. Computer simulations and real data inversion results show that the method of analysis of NMR relaxation signal is very effective, especially in low SNR of circumstances, and can guarantee the authenticity of relaxation spectrum.