径向水平井校直器的BP设计法

A DESIGN METHOD OF STRAIGHTENING DEVICE IN WHIPSTOCK FOR RADIAL HORIZONTAL DRILLING BY BP NEURAL NETWORK

根据径向水平钻井井下斜向器校直机构的设计要求 ,提出了具体的设计原则。运用反向传播多层神经网络 (BP网络 )对校直机构的滚轮位置进行了计算 ,同时利用BP网络优选了校直器阻力最小时的设计方案。结果证明 ,利用BP神经网络进行设计可显著地减少校直器设计与调试中的盲目性。

According to the design demand for straightening device in whipstock of radial horizontal drilling, the principles for design of the device are suggested. The positions of the wheels in the straightening device are calculated by using BP neural network. The results of calculation consist with those of experimental well. BP neural network is used to optimize the resistance of the straightening device. It is proved that this method can raise the accuracy of design and experiment remarkably.