Health monitoring of welded structures using statistical process control

This paper presents health monitoring of welded structures using acceleration time response data. Residual errors are extracted from the measured acceleration time response data using an auto-regressive model. Damage identification is done by monitoring the residual errors using Shewhart and exponentially weighted moving average control charts. The applicability of the proposed method is tested with the welded structure model. Five damage levels are investigated and the damage is introduced by cutting a slot in the weld using an electrical discharge machine. Acceleration time response data are collected using piezoelectric sensors for all damage levels. The results show that both Shewhart and exponentially weighted moving average control charts are capable of identifying the presence of damage in the welded structure model under consideration. Exponentially weighted moving average control charts are more sensitive in damage identification than Shewhart control charts.