LIBRARY ı6ıul 0) - Cranfield University
LIBRARY ı6ıul 0) - Cranfield University
LIBRARY ı6ıul 0) - Cranfield University
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
" digital hardware, which is basically composed of a digital computer and<br />
analog-to-digital (A/D) converters.<br />
" signal processing component, provided by dedicated software algorithms.<br />
2.6.3 Signal processing and interpretation<br />
Signals in general can be classified into two groups, namely: deterministic and<br />
stochastic signals. Stochastic signals are those whose behaviour is highly<br />
unpredictable, that is, they occur randomly [ref. 155]. On the other hand,<br />
deterministic signals are those that have known characteristics and that can be<br />
explicitly described by mathematical and physical models [refs. 155,158].<br />
Stochastic signals are affected by random noise but if the noise content is<br />
negligible, the process may be regarded as deterministic. It should be noted that the<br />
welding current and voltage signals in their unsmoothed states are stochastic. [ref. 51]<br />
A deterministic signal can be formed from a stochastic signal provided the<br />
amplitude or time classes of the signal are formed over a sufficiently long period [ref.<br />
158]. Theoretically, for a precise information to be extracted from a stochastic signal<br />
an infinite record length is necessary and the information based on finite length<br />
records must always be qualified by statistical statements referring to the probability<br />
of the information being correct within a certain percentage [ref. 159]. The period for<br />
which data is collected (i. e. sampling time) should be sufficiently long such that the<br />
mean value of a definite portion of the signal is equal to the overall average of the<br />
total signal [refs. 156,158] and/or the Fourier transform of data collected over a<br />
longer period should not differ significantly from the Fourier transform of the data<br />
collected over the initially chosen sampling time [ref. 159]. This enables statistical<br />
analysis to be performed on the signal.<br />
For monitoring gas metal arc welding, sampling times ranging from 100<br />
milliseconds to 1 second and sampling frequencies21<br />
ranging from 200 Hz to 10 kHz<br />
have been reported in the literature [refs. 38,161,162].<br />
The minimum sampling frequency necessary for a sampled data to represent<br />
the continuous time signal without aliasingu is set theoretically by the Shannon<br />
sampling theorem as twice the maximum frequency-component of the signal [ref.<br />
163]. Sampling frequencies equal to or greater than 8 times the maximum signal<br />
frequency are generally used [ref. 163].<br />
Large amounts of data are usually collected during monitoring, most of this<br />
information is not useful for process control. The data need to be reduced to make<br />
analysis simpler, faster and to save on storage capacity [ref. 51].<br />
The most common data processing approach is to break the sampled transient<br />
data into its basic statistical features such as mean, minimum, maximum, standard<br />
deviation, etc. This is called feature extraction and significantly reduces the data<br />
without losing important information, filtering out irrelevant information [ref. 160].<br />
21 Sampling frequency is the frequency at which the Analog-to-Digital converter acquire and<br />
converts the analog signal to a discrete-time signal (series of consecutively sampled data). The data<br />
is normally acquired during the sampling time at a fixed sampling frequency.<br />
22 Aliasing is a term used in control theory to define the distorsion that occurs in a signal when it is<br />
reconstructed from a digitized signal which was sampled with a frequency not high enough to fully<br />
represent the original analogue signal.<br />
42