Detection of arcing in DC motors

This paper outlines an online condition-monitoring method for detecting arcing around the brush-commutator interface in DC motors. These motors are the critical components in steel-rolling mills, and tend to have excessive arcing and motor flashover under certain circumstances. The proposed method utilises system-identification and detection techniques. First, various motor parameters are identified. Then a parameter related to the arc-discharge voltage is estimated. This parameter is monitored for shift or drifts in its mean. Its mean value is different when the condition of the motor changes from no-arc to arcing. When this condition is detected, a warning is flagged to the mill operators. The detection involves a simple test between a no-arc hypothesis and an arcing hypothesis. A prototype system has been implemented in one of BHP´s steel-rolling mills for system evaluation