System‐Level Anomaly Detection in BLDC ‐Actuated Robotic Arms Using D‐Bi‐ LSTM and BSLO

In robotic arms, Brushless DC (BLDC) motors are important in offering proper motion control and provide smooth and better operation. These motors are mounted on the frame of the robotic arm, providing smooth and stable joint movements. Anomaly detection in robotic arms is important to provide operational efficiency and reliability. This work introduces a deep learning (DL) Dual Bidirectional Long Short‐Term Memory (D‐Bi‐LSTM) based Blood Sucking Leech Optimizer (BSLO) for anomaly detection, formulated as a binary anomaly detection task that distinguishes normal and anomalous operating states in the BLDC based robotic arm using velocity data. The proposed model combines the temporal learning capabilities of D‐Bi‐LSTM with the strengths of the BSLO algorithm for automatic hyperparameter tuning and faster convergence, and improves anomaly detection accuracy. The suggested work detects these motor‐related faults effectively by analyzing sensor data and extracting key features. The outcomes show that the dual Bi‐LSTM with BSLO attained superior anomaly detection accuracy of 98.9% and created it a robust solution for early anomaly detection in robotic arms. The proposed approach detects system‐level anomalies reflected in joint motion behavior, rather than detecting internal electrical or mechanical BLDC motor faults.