论文部分内容阅读
该文研究了BP神经网络对固井信号的分类能力.首先应用理论框模型分析了不同层状介质结构的回波,并系统深入地研究了层状介质物理参数的变化对BP神经网络分类的影响.理论研究结果表明,当各层介质参数在一定范围内变化时,不影响网络的正确分类,即BP神经网络对固井信号分类具有很强的Robust性.该文也给出了对实际困井信号的分类结果,其正确识别率在叨％以上.该研究结果反映出BP神经网络具有很强的信号分类能力,对将其用于固并信号分类的工程实际有重要价值.“,”The classifying ability of BP neural network on the logging signal is studied in this paperr First, the echoes of various layered media are analyzed with the theoretical models. Then the effect of the layered medium parameters on the chassifying of BP neural network are analyzed . The theoretical study how that the media parameter not effect the classifying of neural network. The experimental results of clasifying the theoretical simulative signal and measured signal from real wells are also introduced. The correct classification rate achieved by this neural network in over of 90%. The study result showi the classification ability of the nerual networrk is great.