Thus, modified values of the experimental data were used; depth of cut and vibration data are multiplied by 5 and 10, respectively and the spindle speed and feed rate data are divided by 500 and 100, respectively.
At the specified spindle speeds and depth of cut, the vibration amplitudes are recorded by the following instruments: an accelerometer (PG 109 M0, frequency range 1 to 10,000 Hz), a 4-channel digital oscilloscope (model-DPO 43034) with charge amplifier (Model: CA 201 A0).
Two key parameters are linked to these vibrations: the depth of cut achievable without vibrations and the chatter frequency.
Spindle speed, feed rate, depth of cut and vibrations were selected as independent variables in this study.
Four independent variables were used as inputs, namely spindle speed, feed rate, depth of cut and vibrations and the output of the networks was surface roughness.
All neural networks used have a four neurons input layer, because there are four types of input data, i.e. spindle speed, feed rate, depth of cut and vibrations, one or two hidden layers with variable number of neurons and a single neuron output layer for the surface roughness which is the system output.
At a low speeds and depth of cuts the amplitude of vibration levles rises.
Thereafter, as the feed and depth of cut increases the vibration levels also raises significantly.
For each combination of spindle speed, feed rate and depth of cut, the corresponding vibration data (in μV) were recorded.
The main input parameters of this model are the cutting parameters such as cutting speed, feed rate, depth of cut and tool vibration in radial and in main cutting force directions.
