On-line measurement of weld fusion state using weld pool image and neurofuzzy model

Proper fusion is crucial in generating a sound weld. Successful control of the fusion state requires accurate measurements of both the top-side and back-side bead widths. A top-side sensor based system is preferred so the sensor can be attached to and move with the torch. Thus, the system must be capable of estimating the back-side bead width using top-side parameters. Because skilled human operators can estimate the fusion state through observation of the weld pool, in this work a neurofuzzy system is developed to infer the back-side bead width from the pool geometry. It is found that the back-side bead width can be estimated with high accuracy by the identified neurofuzzy model. Thus, accurate feedback of the fusion state can be provided for controlling the fusion state.