Faulty Detection Study of Single Joint of Industrial Robot on Feature Index

To detect faulty status of single joint of industrial robot in real production scenario, it’s constructed a feature index directly from Motor Current Signals Analyze (MCSA) with limited sampling frequency. The features are generated under various velocities and selected from Wilcoxon signed-ranks test for status classification via SVM algorithm. This detecting method is verified with three independent practical samples. The results reveal that performance of single feature is restraint and the proposed index could improve the robustness of faulty detection in real scenario.