| Abstrakt: |
Modern level of automation of surfacing processes requires development of the appropriate computer systems for setting and maintaining the set process modes in real time, analysis, processing and, if required, automatic correction of these parameter values, allowing for their influence on penetration depth, base metal proportion in the deposited metal (BMP), spattering losses of electrode wire (SLM), as well as formation of the deposited beads, their size, and quality. Experiments were conducted and practical methods of development of neural network models (NNM) were demonstrated in the case of such parameters of flux-cored wire arc surfacing as BMP and SLM. These NNM were applied with success for prediction of the above technological parameters of arc surfacing. Widening practical application of NNM in arc surfacing requires expansion of the data base of surfacing technological parameters, material properties, quality of the obtained results, etc. The data base should be focused on application during NNM development and use. 12 Ref., 5 Tabl., 5 Fig. [ABSTRACT FROM AUTHOR] |
