基于BP神经网络的深井钻井井侵诊断与预测

Deep Well Drilling Invasion Diagnosis and Prediction Based on BP Neural Network

为了实现深井钻井地层流体侵入的智能化实时诊断与预测,分析了井侵的发生规律和机理,明确了井侵的风险征兆和表征参数以及各参数的权重等级,确立了实时识别模型和诊断预测流程。优选神经网络方法开展深井钻井井侵智能诊断,建立了深井钻井井侵的三层神经网络诊断模型。通过某井开展模型的实例验证,结果证明建立的模型能够准确地识别该井的井侵发生风险,为后续开展钻井复杂的神经网络智能化诊断奠定了基础。

In order to realize the intelligent real-time diagnosis and prediction of fluid invasion in deep well drilling,the occurrence law and mechanism of well invasion are analyzed,the risk signs and characterization parameters of well invasion and the weight level of each parameter are defined,and the real-time identification model and diagnosis prediction process are established.The neural network method is optimized to carry out intelligent diagnosis of deep well drilling invasion.A three-layer neural network diagnosis model of deep well drilling invasion is established.Through the example verification of the model in one well,the results show that the established model can accurately identify the risk of well invasion in the well,which lays a foundation for the subsequent intelligent diagnosis of drilling complex neural network.