论文部分内容阅读
反向传播（ＢＰ）双隐层神经元网络具有较高的容错性和分辨能力，可以形成任意复杂的判别区，对复杂地层条件的油水层具有效高的识别能力。该方法利用常规测井资料，提取了电阻率、自然电位、密度等三个对判断油水层有利的指标，并对原始的测井指标进行了归一化处理，提高了网络的运行速度。经对大庆西部地区实际井资料的处理和验证，取得了较准确的结果。此方法同样适用于复杂地层岩性的划分和水淹层的判别。
Backpropagation (BP) double hidden layer neural network has high fault tolerance and resolving power, can form arbitrarily complex discriminant regions, and has high recognition ability for oil-water layers in complex formation conditions. The method uses conventional well logging data to extract three indicators that are favorable for judging the oil-water layer, such as resistivity, natural potential and density, and normalizes the original well logging index to improve the operation speed of the network. After processing and verifying actual well data in western Daqing, more accurate results have been obtained. This method is also applicable to the division of lithology of complex formations and discrimination of water flooded layer.