Integrated Diagnosis Model and Optimization Algorithm for Beam Pumping Unit Faults

To address the challenge of fault diagnosis in beam pumping units, we propose a methodology based on vibration analysis and an enhanced integrated learning model. The Stacking integrated learning framework employs random forest (RF), support vector machine (SVM), gradient boosting (GB), and extreme gradient boosting (XGboost) as base learners, with multiple linear regression as the meta-learner, aiming to improve the accuracy and generalization capacity of a single model. Additionally, we introduce an improved sand cat swarm optimization (ISCSO) algorithm to optimize the hyperparameters of the model, addressing the challenges associated with manual parameter tuning. Experimental comparisons of prediction results between the ISCSO-stacking model and other models demonstrate that the ISCSO-stacking model achieves a prediction accuracy of 97%. Furthermore, the optimized hyperparameters substantially enhance model performance and reduce the risk of overfitting.