Machining characteristics of hastelloy c-2000 in end milling using artificial intelligence approach
This research work deals with the machining characteristics of Hastelloy C-2000 in the end milling operations. The mathematical model was developed through the response surface method (RSM) which basically focuses on machining characteristics such as surface roughness, tool life and cutting force us...
| Summary: | This research work deals with the machining characteristics of Hastelloy C-2000 in the end milling operations. The mathematical model was developed through the response surface method (RSM) which basically focuses on machining characteristics such as surface roughness, tool life and cutting force using coated and uncoated carbide cutting inserts in wet conditions. The accuracy of this aforementioned technique and model was verified by ANOVA. The minimum and maximum of machining performance was presented followed by the confirmation test to validate the design variables. It is found that the models are able to predict the longitudinal component of the surface roughness, cutting force, and tool life close to those readings recorded experimentally with a 95% confident level. Artificial Neural network (ANN) prediction model was developed with back propagation algorithm with the use of multilayer perceptron and activation function of hyperbolic tangent. Feed rate is the most influential factor, followed by axial depth and cutting speed for surface roughness, tool life and cutting force. The mean absolute relative error for surface roughness of RSM models (first, second order) and ANN is 4.386 %, 2.324 % and 0.1790% for coated carbide inserts and 9.878 %, 6.681 % and 0.136 % for uncoated carbide inserts respectively. In addition, for tool life model, 8.3130 %, 4.8760 %, 0.2% for coated carbide inserts, 9.7880%, 7.6270 %, and 0.1580% for uncoated carbide inserts. Furthermore, for cutting force model 4.386 %, 2.324 % and 0.4181 % for coated carbide and 9.878 %, 6.681 % and 0.5% for uncoated carbide. The PVD coated-carbide cutting tools perform better than the uncoated-carbide in terms of the surface roughness, cutting force, and tool life. Surfaces finish and wear surfaces were characterized using an optical video measurement system, scanning electron microscope (SEM) and electron dispersive X-ray (EDX). The tool failures found in this research was flank wear, notching, and chipping. Adhesion and plastic lowering at cutting edge were the main tool wear mechanisms seen in the present work, which is clearly demonstrated by the adhered workpiece material and the formation of a built-up edge (BUE) on the tool flank. There have been a few chips found in this research and broadly they can be divided into two types. Type 1: unstable and type 2: critical. Due to the research done on the earlier models, RSM established prediction and optimization models. However, ANN serves more efficiency and accuracy because its error is very less compared to RSM. ANN has characteristics of predicting machining and they work far better when compares to mathematical modelling. |
|---|