«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:

34

期数:

2024年12期

页码:

200-206

栏目:

新型计算应用系统

出版日期:

2024-12-10

文章信息/Info

Title:

Logging Instrument Jam Prediction Based on Multi-strategy Improved Grey Wolf Algorithm

文章编号:

1673-629X(2024)12-0200-07

作者:

高雅田; 李英楠

东北石油大学 计算机与信息技术学院,黑龙江 大庆 163318

Author(s):

GAO Ya-tian; LI Ying-nan

School of Computer & Information Technology,Northeast Petroleum University,Daqing 163318,China

关键词:

遇卡分析; 支持向量机; 灰狼优化算法; 多策略; 佳点集; 差分进化算法; 混沌干扰

Keywords:

stuck analysis;  support vector machine;  grey wolf optimizer;  multiple strategies;  good point set;  differential evolution; chaotic interference

分类号:

TP391

DOI:

10.20165/j.cnki.ISSN1673-629X.2024.0288

摘要:

针对测井作业中遇卡现象预测难度大、预测准确率低的问题,以及灰狼优化算法(Grey Wolf Optimizer,GWO)存在种群多样性不足、易陷入局部最优的缺陷,该文提出了一种基于多策略改进的灰狼优化算法( Improved Grey Wolf Optimizer,IGWO)结合支持向量机(Support Vector Machine,SVM)进行仪器遇卡分析。 利用佳点集理论初始化提高种群多样性,引入自适应调整机制与差分进化算法(Differential Evolution,DE)的交叉变异的处理机制以及混沌干扰避免局部最优问题。 同时,在种群迭代过程中加入贪婪策略指导个体的选择更新,从而加速收敛。 将 IGWO 算法与其他 5 种群体智能优化算法在 4 种测试函数上进行实验,并将其应用到测井遇卡预测问题中,实验结果表明,通过 IGWO 算法对模型参数进行调优,有效提升了算法的寻优能力和全局搜索能力。 优化后的模型在测试集上的平均交叉验证准确率为 86. 26% ,其中,几何遇卡的 MAE 为 0. 1,RMSE 为 0. 316 2;力学遇卡的 MAE 为 0. 05,RMSE 为 0. 223 6。 整体上,模型表现出较高的准确率和较小的误差,具有较强的预测能力,为解决测井作业中的遇卡问题提供了有效的解决方案。

Abstract:

Aiming at the problem of difficulty in predicting the stuck phenomenon and low prediction accuracy in well logging operations,as well as the defects of Grey Wolf Optimizer ( GWO) in insufficient population diversity and easy to fall into local optimality,we propose an Improved Grey Wolf Optimizer (IGWO) based on multiple strategies combined with Support Vector Machine (SVM) for in-strument stuck analysis. The population diversity was improved by initialization using the good point set theory, and the crossover mutation processing mechanism of the Differential Evolution (DE) algorithm and chaotic interference were introduced to avoid the local optimal problem. At the same time,the greedy strategy was added to guide the selection and update of individuals in the population iteration process, thereby accelerating convergence. The IGWO and five other swarm intelligence optimization algorithms were experimented on four test functions and applied to the problem of well logging stuck prediction. The experimental results show that the optimization of the model parameters by the IGWO effectively improves the algorithm’s optimization ability and global search ability.The average cross-validation accuracy of the optimized model on the test set is 86. 26% ,among which the MAE of geometric stucking is 0. 1 and the RMSE is 0. 316 2,the MAE of mechanical stucking is 0. 05 and the RMSE is 0. 223 6. Overall,the model shows high accuracy and small error,with strong prediction ability,and provides an effective solution to the stuck problem in well logging operations.

常用功能

更新日期/Last Update: 2024-12-10