Process acceptability in turning of titanium alloys based on cutting force sensor monitoring

Turning tests were carried out on a difficult-to-machine Ti alloy (TiAl6V4) under diverse cutting conditions. The principal objective and scope of this work was the reliable and robust identification of Acceptable / Not Acceptable process conditions for such materials. Decision making on process conditions acceptability is performed using supervised neural network approaches applied to single cutting force component signal specimens and sensor fusion of cutting force signal data. The quality parameters utilised for Acceptable / Not Acceptable process conditions classification were the vibration level during machining and the type of chip generated ranked between 1 (good) and 5 (bad) on the basis of expert turning operator knowledge. The rule-of-the- thumb to classify process conditions as Acceptable or Not Acceptable is the following: whenever a 4 or 5 classification is verified either for Vibration Level or Chip Type, process condition are deemed Not Acceptable. The obtained results showed that the NN SR (Success Rate) for Acceptable process conditions identification are always significantly higher than for Not Acceptable process conditions. By resorting to sensor fusion of the 3 cutting force components (Fx + Fy + Fz) instead of using single cutting force components (Fx, Fy, Fz), the NN SR tend to increase also for Not Acceptable process conditions identification. This emphasizes the NN capability to realize the concept of sensor fusion.