基于人工神经网络的可取式桥塞胶筒密封性能预测

Sealing Prediction of Retrievable Bridge Plug Rubber Based on Artificial Neural Networks

可取式桥塞是井下作业中使用的主要工具,其中胶筒结构参数的合理性直接影响油井产量和安全生产。为此提出一种基于BP网络和正交试验法的桥塞胶筒密封性能的预测方法,在Abaqus软件环境中建立可取桥塞胶筒模型进行仿真,得到10组数据构成网络训练样本,用BP网络建立映射关系,建立胶筒的密封性能预测模型。结果表明建立的BP网络模型具有较高预测精度,使用BP网络预测胶筒密封性能可行、高效。

The retrievable bridge plug rubber is the main tool used in down-hole operation,the parameters rationality of rubber structure would directly affect the yield of oilwell and the safety in production.For this reason,a kind of sealing prediction method based on BP network and orthogonal test was put forward.To establish the retrievable bridge plug rubber model in the environment of Abaqus software and to carrying out simulation so as to obtain the network training specimen constituted with 10 sets of parameters.Using the BP network to establish the mapping relation rubber and by means of the BP network to achieve sealing prediction.The result of prediction showed that the BP network is efficient and reliable for sealing prediction of rubber.