Riveting Quality Inspection Based on Bayesian Optimization Gradient Boosting Decision Tree

In industrial manufacturing, the inspection of rivet quality is of great significance because it is directly related to structural integrity and safety. In this paper, a gradient boosting decision tree (BOT-GBDT) method based on Bayesian optimization tuning is introduced for the automatic detection of rivet quality. A comprehensive feature dataset is constructed by extracting rivet surface features using image processing and fine-tuning the hyperparameters using Bayesian. The experimental results show that the optimized BOT-GBDT model achieves an overall accuracy of 92.5% on the test set and outperforms the support vector machine (SVM) and the unoptimized GBDT model in defect detection. The proposed method shows strong potential for application in rivet quality inspection and provides new insights for machine learning-based quality control in manufacturing.