Feature: AI
Figure 9: Results screen for tool wear and surface roughness Te graph on the frequency plane for the same signal is shown
in Figure 11. Tis approach enables examination of signals on the time and frequency planes based on the Fourier series functions. Equation 2 determines the Fourier transform, F(w), of a signal obtained on a complex exponential display; Equation 3 determines e jwt
:
(2) (3)
Figure 8: Artificial neural network and the prediction of tool wear and surface roughness
The program During ANN training, it was possible to predict vibration wear and surface roughness based on our previously-trained network using 13 different parameters (cooling water high pressure, breaking speed, breaking depth, progress speed, force RMS, average force, peak-value force, vibration RMS, vibration average, vibration peak value, and others). Te program’s interface is shown in Figure 8. Tool wear and surface roughness results are provided to the user
in a separate window; see Figure 9. In addition, with the aid of the program, we determined the time
intervals and frequencies at which vibrations occurred, effectively predicting future ones; see Figures 10 and 11. As can be seen from these figures, the actual and predicted signals are very similar to one another, indicating a high degree of success of the training used for this program.
ANN in projects In this study we developed an interface in Matlab by using an ANN on data obtained from the cutting workpiece made of the Inconel 718 super-alloy, at different pressures of cooling liquid, which was shown to be capable of making surface roughness and tool wear predictions according to 13 different input parameters. Te reliability of the data trained with an ANN was tested
using the regression curve method. Tool wear and surface roughness estimations were found 96% and 97% reliable, respectively. Te regression curves of the predictions are shown in Figure 12. Using this interface, surface roughness and tool wear can be
determined without using any physical measurements methods, saving considerable time. Furthermore, using another interface developed with the ANN method, it was possible to use cutting forces as inputs to predict the frequencies at which vibrations occurred during tests. Tis, again, allowed us to determine the
Figure 10: The determined signal on the vibration time plane based on the force signal, with ANN
www.electronicsworld.co.uk September/October 2020 35
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68
