BP神经网络方法预测钻井井帮缩径  

作　　者：赵厚胜[1] 谭昊 孟益平[3] 

机构地区：[1]淮北矿业股份有限公司建设发展部,淮北235000 [2]天地科技建井研究院,北京100013 [3]合肥工业大学,合肥230009

出　　处：《安徽建筑工业学院学报（自然科学版）》2012年第4期22-23,43,共3页Journal of Anhui Institute of Architecture(Natural Science)

摘　　要：在深厚的含水冲击层中建造井筒,需要特殊的凿井方法,钻井法是一种机械化程度高,安全、可靠的特殊凿井方式。施工中经常需要对井帮的变形数据进行监控,而传统的测井方式比较困难。通过BP神经网络预测的方式,可以以少量的监控数据为基础进行程序的训练,并通过算法进行预测计算,可以大大减少测量次数,提高钻井质量和效率。文中以有限元计算结果为数据基础,通过建立BP人工神经网络进行训练和预测,并与有限元结果和部分实地测量的结果作比较,论证神经网络对于煤矿钻井井壁缩颈预测的可靠性。To sink shaft through the deep alluvial, shaft lining deformation data is needed for monito- ring, but conventional logging methods is difficult to get much data where we need. Through BP neural network prediction we can greatly reduce the number of measurements with a small amount of mo- nitoring data as the basis for the training. Then the data help to improve the drilling quality and efficiency. By means of training the BP neural network with the data of FEM calculation and prediction, then comparing with measured data, it proves neural network reliable for shaft sinking.

关 键 词：钻井 神经网络 有限元分析 井帮变形 

分 类 号：TP311.52[自动化与计算机技术—计算机软件与理论]