| Monitoring of tool wear condition for the drilling process, which is one of the major cutting processes, is a very important economical consideration in order to prevent tool failures, increase machine utilization and decrease production costs in an automated manufacturing environment. New techniques are proposed in this study for on-line identification of tool wear status based on cutting force and power signal measurements during drilling.; Hidden Markov Models (HMM), which are very popular in speech recognition, are adopted for tool condition monitoring using two approaches in this study. The first is the bargraph monitoring of the HMM probabilities that shows the progress of tool wear during operation and the second is the classification of the sensor signals that correspond the various types of wear status, e.g. sharp, workable and dull, using the multiple modeling method.; Several other innovative techniques, namely the phase plane method, the transient time for the torque signals, parameter estimation of a mechanical model of torque and statistical analysis of all the sensor signals are also discussed.; Finally, a new algorithm, Decision Fusion Center Algorithm (DFCA), is proposed which combines the output of the methods developed in this work to produce a global decision variable that shows the tool wear status during the drilling operation. Experiment results demonstrate the effectiveness of the proposed methods and the DFCA algorithm.; Although this work focuses on on-line tool wear condition monitoring for drilling operations, the concepts introduced are general. Process and machine monitoring based on HMM and the decision algorithm (DFCA) developed in this thesis can be applied to many other cutting and machining operations. |
