Drilling accidents, complex and diverse, occur dynamically uncertain, as the traditional prediction methods are generally of low prediction accuracy and poor adaptability. In order to improve the accuracy of drilling accidents prediction, an adaptive prediction model for drilling accidents based on support vector machine with particle swarm optimization (PSO-SVM) is proposed. The model optimizes SVM parameters by means of the strong global search ability of PSO algorithm to reduce the blindness of SVM parameters selection; it retrain, re-optimize and regenerate the new prediction model after the misclassification accidents have been added to the sample set in order to correctly identify the similar misclassified accidents. The innovation of this model is the adaptive mechanism introduced on the basis of the traditional PSO-SVM model which can be initiative to re-generate prediction model for complex drilling accidents to improve the accuracy of drilling accidents prediction and adapt with different drilling conditions. Finally, verification of the model is completed through predicting the actual accident instances and comparing with the traditional PSO-SVM model. The results show that this model has stronger adaptive ability and higher prediction accuracy, so it will be of great significance for accurately predicting drilling accidents and reducing the cost of drilling.
展开▼