Optimization and neural modelling of infiltration rate in ultrasonic machining

Ultrasonic machining is a processing method typically practiced for processing the highly brittle/hard materials. The proposed research work is attempted at exploring the influence of varying input conditions namely; cobalt %, power rating, thickness of work, different tools, tool geometry, and abrasive size on the infiltration rate in ultrasonic drilling of WC–Co composite through neural modelling. The design of experiments methodology has been practiced for scheming out the experiments. The significant process variables have been acknowledged using variance analysis test which has revealed the abrasive size, power rating, and tool profile as the most influential factors for the infiltration rate. An artificial neural network (ANN) model is suggested to analyze the infiltration rate in USM with striking parameters. Multiple layer feed frontward neural architecture is restrained through error-back propagation-based training algorithm. Predicted results show the effectiveness of the proposed neural structure with maximum error of 6%. The optimized parametric combination for infiltration rate has been revealed as; cobalt- 6%, work thickness- 3 mm, tool- hollow, tool material- nimonic-80A alloy, abrasive size- 200, and power rating- 80%. Microstructure analysis revealed that good edge quality with no appearance of cracks or burr/chipping on the edge of the drilled holes which further ensured the quality level of hole drilling through attempted work.