A sticking predictor construction and evaluation method for drill tools sticking prediction.

Abstract

The construction and evaluation of the sticking predictor are the basis of drill tool sticking prediction. This paper proposes a method to construct and evaluate the sticking predictor for rod-deflection sticking accidents. First, one uses various feature extraction methods to extract the sticking features from the sticking signal. Second, we introduce the Mann-Kendall method to test the obtained feature parameters and select the feature parameters that can reflect and track the sticking evolutionary trend. Third, the sticking predictor is constructed by calculating the weight values of the screened features. Finally, to test the effectiveness of the sticking predictor, we propose the sticking predictor evaluation model. The experimental result shows that the constructed sticking predictor in this paper is superior to other input features and provides a reference for predicting sticking accidents in engineering practice.