A Neural Network Closed-loop Control of CO_2 Welding Spatter by means of Arc Sound(Physics, Processes, Instruments & Measurements)

Search this Article

Author(s)

Abstract

This paper deals with the problems of decreasing CO_2 welding spatter by utilizing neural networks to achieve intelligent control. The arc sound energy produced at the beginning and end of short circuit transfer was used to sense the spatter ratio. One multi-layered neural network was utilized to identify the relation of the arc sound energy to the frequency of short circuit transfer. Both the arc sound energy and the frequency of short circuit transfer were taken as the input of the neural network controller and the output parameters were welding voltage and current (wire feed rate). The result of simulation and experiment showed that this system could control CO_2 welding process and decrease the welding spatter remarkably through some times self-learning. The technique explored here could be applied to a wide variety of nonlinear control problems in welding.

Journal

  • Transactions of JWRI

    Transactions of JWRI 30(2), 1-4, 2001-12

    Osaka University

Codes

  • NII Article ID (NAID)
    110006486115
  • NII NACSIS-CAT ID (NCID)
    AA00867058
  • Text Lang
    ENG
  • ISSN
    03874508
  • Data Source
    NII-ELS 
Page Top