【摘要】 整体式止屈器是一种对海底管道屈曲传播破坏起到阻止作用的保护装置.应用Abaqus有限元软件建立管道和整体式止屈器模型,采用一种特殊的流体质量加载法来模拟管道屈曲传播以及止屈器的穿越过程,并计算得到了止屈效率.通过有限元模型研究了止屈器长度和厚度对止屈效率的影响,研究表明增加止屈器的长度或厚度都可以提高止屈效率.采用基于贝叶斯正则化的神经网络对止屈效率的计算结果进行拟合,并应用该神经网络对6组随机尺寸止屈器的效率进行预测,最终预测结果与实际计算结果相吻合,证明了此网络模型在止屈效率拟合及预测方面的适用性.更多还原

【Abstract】 Integral buckle arrestors are devices that are strong enough to stop a propagating buckle failure of offshore pipelines.The finite element analysis software of Abaqus is used to model the pipe and arrestor.A special fluid mass-controlled loading procedure is adopted to simulate the buckle propagation and the crossover of arrestors.An extensive parametric study of arrestor efficiencies is conducted using the finite element model,in which the length and thickness of arrestors are considered.The Bayesian regularization neural network(BNN) is used to fit the calculation results and predict the efficiencies of 6groups of new random-sized arrestors.The agreement between predicted results and calculated results proves that BNN is a viable way of fitting and predicting arrestor efficiencies.更多还原