Crankshaft Wear Fault Detection Method for Welding Robot Reducer based on Belief Rule Base Inference

Abstract

Aiming at the nonlinear relationship between the welding machine servo motor torque signals and the RV （Rotate Vector） reducer crankshaft wear states， a wear fault detection method based on belief rule base inference （BRB） is designed. Firstly， the inputs of BRB system are considered as the mean values of the motor torques and torque derivatives， the outputs are set as the crankshaft wear fault levels. As a result， a belief rule base describing the mapping relationship between the inputs and the outputs is established. After the input signals are online obtained， the evidential reasoning （ER） algorithm is used to fuse the belief rules activated by inputs to obtain a belief distribution about the fault levels， and the degree of the crankshaft wear is evaluated by the distribution. Finally， using the measured torque data to verify the proposed method， it shows that the designed BRB fault detection method can largely replace the maintenance engineer to realize the automatic detection of the faults.

Keywords

• Industrial robot;RV reducer;Crankshaft wear detection;Belief rule base;Industrial alarm system