A Master of Science thesis in Mechatronics Engineering by Azza Rashed Al Hassani entitled, "Modeling of Tool Wear when Turning of TI-6AL-4V Titanium Alloy," submitted in February 2013. Thesis advisor is Dr. Ibrahim M. Deiab and thesis co-advisor is Dr Khaled Assaleh. Available are both soft and hard copies of the thesis.
Difficult-to-cut materials are widely used particularly in the aerospace and automotive industries. However, the high cost of processing these materials limits the use of their improved mechanical properties. Tool life is one of the most important factors in machining operations of such materials and it is mainly affected by cutting conditions including the cutting speed, feed, depth of cut and cooling environment along with the generated temperature and cutting forces. In addition, the modern industry is moving towards automating the manufacturing processes. Therefore, tool life monitoring is important to achieve an efficient manufacturing process. In this study, a tool wear prediction model during the turning of Titanium alloys is studied. It is based on the monitoring of tool performance in controlled machining tests with measurements of cutting forces and vibration under different combinations of cutting parameters (cutting speed, feed rate, depth of cut and coolant). The influence of cutting parameters on the tool life was studied experimentally by performing more than 300 cutting tests. A prediction model was then developed to predict tool wear. The basic steps used in generating the model adopted in the development of the prediction model are: collection of data; analysis, pre-processing and feature extraction of the data, design of the prediction model, training of the model and finally testing the model to validate the results and its ability to predict tool wear. In this work, tool wear prediction was developed using three different modeling methods: Feed-forward Back-Propagation Neural Network, Regression Analysis and Gaussian Mixture Regression (GMR). Comparing the predicted tool wear values with the measured ones showed reasonable agreement. Neural Network modeling yielded the least prediction error with prediction accuracy of 90.876% which is 2.702% and 1.23% higher than the prediction accuracy of the GMR and regression models respectively. Search Terms: Titanium alloys, Turning process, Tool wear, Neural Network, Regression Analysis, Gaussians Mixture Regression.