RBF神经网络法在泥页岩有机非均质性测井评价中的应用

探讨了通过RBF神经网络方法利用常规测井曲线评价泥页岩TOC和S1的可行性.利用RBF神经网络法和ΔlgR法对L69井TOC做测井评价并进行对比,二者建模结果中实测值与计算值R2分别为0.73和0.7,前者略优于后者,当声波或电阻曲线与TOC为非线性关系时表现尤为突出;利用RBF神经网络法对L69井S1进行测井评价,其建模结果中实测值与计算值R2为0.73,精度较高,成功实现了对泥页岩中S1的预测.研究结果表明,RBF神经网络法对泥页岩有机非均质性(包括含油非均质性)的测井评价可行性和模型精度均较高,具有较大应用前景.

Application of RBF Neural Network to Logging Evaluation of Clay Shale Organic Heterogeneity

The feasibility of evaluating clay shale TOC and $1 with conventional logging curves was dis- cussed by using the RBF neural network method. The RBF neural network method and the AlgR method were adopted to evaluate the TOC of L69 well,and the R2 of measure value and calculated value of the for- mer method was 0.73 and that of the latter method was 0.7 ,which showed the RBF neural network meth- od was a little better than the AlgR method, especially when the correlation was non-linear between acous- tic or resistance curve and TOC. And the RBF neural network method was used to evaluate S1 of L69 well, the R2 of the measure value and calculated value was 0.73,and the accuracy was higher, successfully esti- mating S1 contained in clay shale. The study shows that the RBF neural network method to evaluate the or- ganic heterogeneity （including the oil content heterogeneity） is feasible, and therefore has possessed a broad application prospect.