To avoid the considerable challenges and losses caused by stuck drilling to normal drilling operations, this article analyses the mechanism of stuck drilling, then combines the artificial fish swarm algorithm (AFSA) and support vector machine (SVM), and finally proposes an early warning model for the stuck-in medical drilling process based on the AFSA and SVM. The model realizes real-time sticking risk warnings by using the four parameters of riser pressure, torque, speed and hook load collected in real time and promotes real-time drilling parameter monitoring for the real-time dynamic warning of sticking risk. By comparing the AFSA-SVM sticking prediction model with the particle swarm optimization model and the traditional cross-validation optimization model, it is found that the AFSA-SVM stuck prediction model accuracy can reach 97.561% and that the training and testing times are 5.874 s and 0.76 s, respectively. Its accuracy and computational efficiency are higher than those of the particle swarm optimization model and traditional cross-validation optimization model. In comparison with the existing technology, the four-parameter early sticking warning model based on AFSA-SVM presented in this paper shows powerful comprehensive performance and field application value.

为避免卡钻给正常钻井作业带来相当大的挑战和损失，本文分析卡钻机理，然后结合人工鱼群算法（AFSA）和支持向量机（SVM），最后提出预警模型用于基于 AFSA 和 SVM 的卡入式医疗钻孔过程。该模型利用实时采集的隔水管压力、扭矩、速度和大钩载荷四个参数实现了卡钻风险的实时预警，促进了钻井参数的实时监测，实现卡钻风险的实时动态预警。通过将AFSA-SVM粘连预测模型与粒子群优化模型和传统交叉验证优化模型进行对比，发现AFSA-SVM粘连预测模型准确率可达97。561%，训练和测试时间分别为 5.874 s 和 0.76 s。其精度和计算效率高于粒子群优化模型和传统的交叉验证优化模型。与现有技术相比，本文提出的基于AFSA-SVM的四参数早期粘滞预警模型显示出强大的综合性能和现场应用价值。