Lost circulation events during drilling operations are known for their abruptness and are difficult to control. Traditional diagnostic methods rely on qualitative indicators, such as mud pit volume changes or anomalous logging curve patterns. However, these methods are subjective and rely heavily on empirical knowledge, resulting in delayed or inaccurate predictions. To address this problem, there is an urgent need to develop efficient methods for a timely and accurate lost circulation prediction. In this study, a novel approach is proposed by combining principal component analysis (PCA) and empirical analysis to reduce the dimensionality of the model data. This dimensionality reduction helps to streamline the analysis process and improve prediction accuracy. The predictive model also incorporates an improved fruit fly optimization algorithm (IFOA) in conjunction with support vector machine (SVM) techniques. The actual instances of lost circulation serve as the evaluation criteria for this integrated method. To overcome the challenges associated with irregular population distribution within randomly generated individuals, a tent map strategy is introduced to ensure a more balanced and representative sample. In addition, the model addresses issues such as premature convergence and slow optimization rates by employing a sine-cosine search strategy. This strategy helps to achieve optimal results and speeds up the prediction process. The improved prediction model demonstrates exceptional performance, achieving accuracy, precision, recall, and F1 scores of 96.8, 97, 96, and 96%, respectively. These results indicate that the IFOA-SVM approach achieves the highest accuracy with a reduced number of iterations, proving to be an efficient and fast method for predicting the lost circulation events. Implementation of this methodology in drilling operations can lead to improved efficiency, reliability, and overall performance.
© 2023 The Authors. Published by American Chemical Society.