الفهرس | Only 14 pages are availabe for public view |
Abstract Abrasive water jet (AWJM) machining is one of the most recently developed manufacturing technologies. It is superior to many other cutting techniques in processing various materials, particularly in processing difficult- to-cut materials. This thesis presents a comprehensive study on the effect of cutting parameters such as standoff distance, nozzle traverse speed, abrasive flow rate and material type on cutting performance for two types of marble which called Carrara white and Indian green. The main cutting performances are surface roughness, surface waviness and Kerftaper ratio. from the results, it was found for the two types of marble that surface roughness increased by increasing standoff distance and traverse speed but increasing abrasive flow rate reduced surface roughness meanwhile for Carrara white the roughness increase again beyond certain limit. Also, it was found for both types of marble that surface waviness decreased by increasing abrasive flow rate and standoff distance. But increasing standoff distance beyond certain limit causes increasing in surface waviness. High traverse speed results in an increase of surface waviness. Finally, it was found that kerftaper ratio increased by increasing standoff distance and traverse speed. Meanwhile, it decreased with increase in abrasive flow rate. Based on the results that were obtained from the experiments, artificial neural network (ANN) technique was used to construct relationship between cutting parameters and cutting performance and predict the unknown values for different input parameters. One, two three hidden layers were used for each individual performance and also a general network which contains all performance was constructed. Then a comparison between individual and general network was achieved. After applying ANN among inputs and single output, to estimate surface roughness, surface waviness and Kerf taper ratio, it was found that the accuracies of ANN were 95.5%, 95%, and 96.5% .While, by conducting multiple outputs in one network, the accuracy decreased to about 89%.So, using single output ANN is more accurate than multiple outputs ANN. |