A signal processing of deep drilling process

Drilling process is a material removal process to produce a hole. Any hole 10 times to its diameter is considered a deep hole. There are a lot of applications in industry that demand on the depth of hole to be drilled such as die, engines and aerospace industries. The depth of hole can minimize the operation and save money. Since the drilling process are move to the automated manufacturing environment nowadays. One of the primary issues in deep drilling technique is tool wear and failure which can affect the sustainability of the process. Therefore, based on collective data, classifying the tool wear mechanism and failure of deep drilling, the tool life stage can be identified and tool major fracture can be avoided. In this experiment, signal processing method was chosen to monitor the tool condition. By using the two sensors which is dynamometer and accelerometer, the signal data obtained was then being analyzed using three different signal processing techniques which are Fast Fourier Transform (FFT), Short Time Fourier Transform (STFT) and Hilbert-Huang Transform (HHT). The SKD61 material and the High Speed Steel (HSS) drill bit was used to carry out the experiment. There are 25 sets of experiments with different parameter used for each set. The parameter used was determined using Design of Experiment (DOE) method. Every sets of experiment were repeated three times to increase the accuracy of the signal data obtained. Based on classification data, the feedrate and cutting speed above 298.8 mm/min and 1592 rpm will lead to tool failure; blunt or fracture. Time domain graph shows the force produced at z axis is the highest. Using FFT, there is no dominant frequency for the good tool condition. However there are some dominant frequencies for blunt and fracture tool. To differentiate between blunt and fracture, the amplitude of FFT gives the higher value when the tool is fracture. This is due to the tool bending and chip clogging. STFT was used to illustrate when is the high frequency region was occur. Then, the signal data was analyzed using HHT which decompose the time series into a set of components called intrinsic mode functions (IMF). IMF was used to detect tool failure by means of the energies of the characteristics IMF associated with characteristics frequencies of the drilling process. When the tool failure occurs, the energies of associated characteristics IMF change in opposite directions. Based on signal data and the tool condition, the type of tool failure was classified whether the tool is good, blunt or fracture. The optimization usage of machining parameter also influences the tool condition during the drilling process was perform. Since the time domain just can capture the time and force produced during the process, the FFT in needed to measure the frequency along the experiment. The FFT amplitude may control the tool life and failure. However, the STFT is used to capture the right time when the high frequency region was occur. Other than that, the time when tool failure occurs can be traced through the associated IMF characteristics generated using HHT methods. Consequently, the signal data processing is not only used to detect the tool failure, but it can be used to develop a system for online tool failure detection which can detect the failure and control the machine parameter to be optimized with the tool conditions