地层岩性随钻识别的神经网络方法研究

钻井过程中随钻识别钻头当前位置的岩性信息对于合理选择钻头类型、快速建立岩性剖面、及时发现油气层和卡准取心层位有着重要意义。钻井实践证明，地层岩性的变化在钻井参数上有很好的综合体现。以录井资料为基础，结合已钻井的测井资料，根据BP神经网络原理，建立了地层岩性随钻识别神经网络模型。应用该模型在新疆油田进行了地层岩性随钻识别试验，试验结果与测井资料解释结果相比，效果较好，符合率可达80％。

A METHOD FOR LITHOLOGY IDENTIFICATION WHILE DRILLING BASED ON NEURAL NETWORK

During the drilling process, it is very important to identify formation lithology near bit while drilling for selecting bit types, for quick establishment of the lithology section, for discovering the oil and gas layers in time, and for locating the coring layer exactly. Drilling practice indicates that there is a direct or indirect relationship between drilling parameters and lithology. Based on the mud logging data and combining with the logging data of the drilled wells, BP neural network is applied to establish a neural network model for lithology identification while drilling. This model was verified in Xinjiang oilfield. Compared with the geological explanation of logging data, the prediction result of the model is much better than ever before and the coincidence rate can reach as high as about 80%.