ADVANCED ROBOTICS FOR AUTONOMOUS ARC WELDING IN A HIGH-PRODUCTIVITY YARD

Abstract

Intelligent industrial robots are widely used in various industries. In marine and offshore industry, robotic welding system is used for joining thick materials in order to enhance the productivity and safety. In this thesis, challenging issues in these applications are discussed. Firstly, a novel filling control algorithm for suppression of seam boundary error propagation is proposed for robotic multi-pass welding systems. A non-causal feedforward controller is designed for compensating the error from previous welding passes using the H-infinity loop shaping technique. Secondly, the proposed feedforward control algorithm is extended to a complex weld geometry. A welding profile control with data-driven fast input allocation (PC-FIA) algorithm is proposed. Experimental results show that the required industry standard is achieved. Finally, an adaptive weld quality monitoring framework is proposed for multiple-conditioned robotic welding tasks based on a swing high temperature sensor system. Both known and unknown welding conditions can be identified and the probabilistic weld quality estimation accurately reflects the weld defects in real-time using proposed method.