Collecting Performance Prediction for the Rubber Collector in Horizontal Wellbore Based on AutoML

To objectively study the influencing factors of rubber collector performance in horizontal wellbores and identify parameter optimization directions, this paper introduces a modeling approach for rubber collectors in horizontal wellbores and develops a corresponding performance prediction method using AutoML(implemented with AutoGluon V1.2.0). Specifically, the study establishes a simulated model of a rubber collector in a horizontal wellbore, simulating its operational state under eccentric conditions due to gravity. It evaluates the impact of rubber elastic element properties (such as material and size) on collecting performance using the integral value of the stress distribution as a metric. Based on the simulated model, an AutoML-based regression model is constructed to predict the collecting performance of the rubber collector under various parameters of the rubber elastic component. The analysis reveals that, as the eccentricity of the rubber collector increases, the most significant influencing factors on collecting performance, in descending order of importance, are axial extrusion range, hardness, radius, bulk modulus, and concave degree. The experimental findings indicate that, during parameter optimization for rubber collectors with varying eccentricities, priority should be given to the aforementioned factors. Additionally, adjusting properties like bulk modulus can enhance the axial extrusion range, thereby effectively boosting the overall collecting performance of the rubber collector in a horizontal wellbore.