Application of particle mutation in drilling accidents diagnosis of Bohai shallow layers

Abstract

Drilling accidents occur frequently in Bohai shallow layers. The existing drilling accidents analysis model lacks real-time performance and has poor operability. By introducing accidents mechanism, the data model based on SVM is established to improve the processing efficiency of drilling accidents, and the targeted measures for handling drilling accidents are recommended. In this study, to perform data optimization and parameter selection for SVM, we propose a quantum variation particle swarm algorithm (QVPSO), which combines K-mean classifier, quantum variation, and particle swarm algorithm. Using the problem of accidents mechanism identification, the proposed drilling accidents diagnosis, termed QVPSO-SVM, was compared with multiple competitive SVM models based on other optimization algorithms including the original algorithm, K-mean classifier, particle swarm optimization, grid method, and genetic algorithms. The experimental results demonstrate that, although the drilling accidents diagnosis needs more time than the original SVM model to solve the identification of accidents mechanism. Drilling accidents diagnosis can significantly improve the accuracy of model (96%). The targeted measures recommended by drilling accidents diagnosis can improve the processing efficiency of drilling accidents and reduce nonproductive time.