Research of neural network for weld penetration control
- Autores: Dukun Ding
- Localización: Applied Mathematics and Nonlinear Sciences, ISSN-e 2444-8656, Vol. 7, Nº. 2, 2022, págs. 849-860
- Idioma: inglés
- Enlaces
-
Resumen
A method has been proposed for welding penetration status prediction in the paper. First, an experimental system was setup and welding experiments were performed. Some groups of welding images could been obtained. A composite filteringsystem composed of a neutral light reduction filter and a narrow band filter was developed to filter the weld arc disturbance.Some operations were performed to the images, namely the median filter and gray transformation. Then a neural networkwas setup, containing three layers. The inner widths of poolxn, the outer widths of poolxw, the width difference valuesebetween the inner and outer of pool, ratios of inner pool widthsRnand ratios of outer pool widthsRwbetween two adjacentimages were determined to be the input parameters. The penetration parameterpwas chosen to be the output. Based on theimages, groups of pool parameter data have been obtained and used to train the network. In this way, the weld penetrationprediction model can be deduced. Finally, verification tests have been done. It showed that weld penetration situationpredicted by the model is fit to its real condition. The accuracy rate is up to 96%, which affords a new way for penetrationdetection
