WEEK 9

PREFACE TO WORK

As the next part of the project the contact material should be identified using the tactile sensor probe. A time domain signals were analysed for all five materials to get the material property .

Tactile sensor get the readings by giving small touch/hit to the object. On the other hand it is given energy to the object. By energy conservation the given energy should transfer to the antenna itself and the object. So according to the object's material , the amount of energy absorption is differ. Hence the energy gain to the antenna probe also differ from material to material of the object (Law of Energy conservation ). Therefore the damping time of the antenna probe may indicate the material properties of the contact objects.

PROGRESS

With the above idea the damping time for the five different materials were calculated. The norm signal of the X and Z axis was taken for the calculations. But since it was a very high frequency it was difficult to fit an exponential function directly to this signal to find the damping time. Therefore, a suitable filter was applied and curves were fitted through the peaks as shown in Figure 22.

As in the Figure 22 it was unable to fit a suitable curve to the signal, but anyhow by using the curve fitting command in the Matlab a general exponential curve was fitted for each and every point. The general exponential model was given in the equation 1.

General model Exp1:

F(x) = a*exp(b*x)------------------------------------------------ (1)

Time constant = τ = 1/b

Gain = a

The exponential curve fitting was shown in Figure 23.

Since vibration details for 25 contacts were taken for each and every contact distances, 25 time constants were calculated for one contact distance and average them. Followed the same for gain as well. After that, they were plotted for the five materials. The resultant graphs were shown in the Figure 24 and 25.

Since 1000Hz was chosen as the sampling frequency the data consist the frequencies up to 500Hz. There may have some significant frequency range that able to separate every materials. With that idea frequency bands were chosen randomly but without taking 50Hz due to the electrical line interference. The low frequency band was selected up to 50 Hz and high frequency band was selected above 50Hz.

Chebyshev 2^nd order filter was used to filter the norm signal and follow the procedure as before. Some resultant graphs for the gain and time constant were shown in the following figures. When analysis the figures it was clearly identified the separation of the materials was varied with the frequency band.

RESULTS

(a) (b)

Figure 26. Variation of the gain (a) and time constant (b) over the contact distance for five materials

2. high frequency band (over 50Hz)

(a) (b)

Figure 27. Variation of the gain (a) and time constant (b) over the contact distance for five materials

After applying several frequency ranges, one range was selected since it has more separation than the others. The resultant graphs of the Gain and Time constant for that particular range were shown in the Figure 28 and 29 respectively. The filter response of the used filter was shown in the Figure 30.

With the signal processing algorithms we are able to find time constant as the unique feature to separate the materials. But it has a limitation since that feature works in a unique frequency band. The details of the used filter as follows.

Fs = 1000 ; Sampling Frequency

Fn = Fs/2; Nyquist Frequency

Wp = [60 160]/Fn; Normalised Passband

Ws = [55 165]/Fn; Normalised Stopband

Rp = 1; Passband Ripple (dB)

Rs = 30; Stopband Ripple (dB)