地层压力BP网络预测方法研究及其应用

为了进一步提高井底下部地层压力预测精度,探讨利用BP网络法进行地层压力预测研究。着重介绍了合理网络结构的确定方法、学习算法的选择和改进等,并建立了适合地层压力预测的BP算法。通过井底下部地层的声波时差预测,介绍了神经网络在地层压力预测中的具体应用过程,并对其在现场的应用效果进行了验证。研究结果表明,利用神经网络来进行地层压力预测的方法,具有预测精度高、自适应好、容错能力强等优点。

Study on Formation Pressure Prediction with BP Network and Its Application

To further improve the prediction accuracy of the downhole formation pressure, the method of B P network was investigated for formation pressure prediction. The focus was on the selection and modification of the determining method and algorithm of the reasonable network structure, etc., and a B P algorithm was built for suitable formation pressure prediction. The specific application procedures of B P network in formation pressure prediction was described, taking the time delay prediction of acoustic wave in the bottom formation as an example, and its field application results were also verified. The study results showed the advantages of this method, e.g., high prediction accuracy, good self adaptation, and good error tolerance, etc.