论文部分内容阅读
本文基于前馈多层网络结构发展了一种稳健的油气识别技术。全局优化方法的引入和学习样本输入模式的改进克服了BP算法的一些缺陷,该技术适用于检测由于地下岩性和孔隙流体性质变化而引起的波形特征的细微变化和进行储集层横向预测。本文给出的合成数据和实际数据算例证实了该识别技术的稳健性和有效性。
This paper develops a robust oil and gas identification technology based on feedforward multi-layer network structure. The introduction of the global optimization method and the improvement of the learning sample input mode overcome some shortcomings of the BP algorithm, which is suitable for detecting subtle changes of waveform characteristics and lateral prediction of reservoirs due to changes of underground lithology and pore fluid properties. The synthetic data and real data examples presented in this paper demonstrate the robustness and effectiveness of this recognition technique.