【摘要】 针对深井钻探液压卡瓦在夹持钻杆过程中易产生应力集中而造成钻杆损伤的问题，提出了一种融合贝叶斯神经网络（Bayesian neural network, BNN）与青蒿素黏菌算法（artemisinin slime mold algorithm, ASMA）的混合优化方法。以液压卡瓦平顶牙型结构为研究对象，首先建立卡瓦牙-钻杆接触模型，通过有限元分析计算卡瓦牙和钻杆的应力分布，并提取初始数据集；在此基础上，通过正交试验筛选敏感参数并进一步扩充样本数据集，用于代理模型训练。随后，利用BNN构建卡瓦牙齿形参数与力学响应的代理模型（决定系数R~2>0.95），并结合ASMA进行多目标优化。结果表明，优化后卡瓦牙的最大等效应力从582.96 MPa降至303.53 MPa（降低了47.9%），钻杆的最大等效应力从360.03 MPa降至235.87 MPa（降低了34.5%），卡瓦牙性能显著提升。研究结果为液压卡瓦牙型结构优化提供了高效、可靠的新思路。更多还原

【Abstract】 To address the issue of stress concentration in hydraulic slips during the gripping of drill pipes in deep well drilling, which often leads to drill pipe damage, a hybrid optimization method integrating Bayesian neural network(BNN) and artemisinin slime mold algorithm(ASMA) is proposed. Taking the hydraulic slip with flat-top tooth structure as the research object, a slip tooth-drill pipe contact model was first established, and the stress distribution of the slip tooth and the drill pipe was calculated through finite element analysis to extract the initial dataset. On this basis, orthogonal experiments were conducted to screen sensitive parameters, and the sample dataset was further expanded for surrogate model training. Subsequently, a BNN-based surrogate model was developed to fit the slip tooth profile parameters and mechanical responses(with a determination coefficient of R~2>0.95), followed by multiobjective optimization utilizing the ASMA. The results demonstrated that the maximum equivalent stress of the slip tooth was reduced from 582.96 MPa to 303.53 MPa(a reduction of 47.9%), while the maximum equivalent stress of the drill pipe decreased from 360.03 MPa to 235.87 MPa(a decrease of 34.5%), significantly enhancing the performance of the slip tooth. The research results provide an efficient and reliable novel approach for the structural optimization of hydraulic slip tooth profiles.更多还原