Signal classification is done by providing the classifier with a complex sample of the input signal across the chosen sampling bandwidth, for a chosen sampling time and a chosen sampling rate. This sample is examined for the properties of the signals contained within it. The results of the classification are output as a list of classified signal parameters.
Two classification modes are provided: A manual mode and a continuous mode. In manual mode the classifier will make one attempt at classification. In continuous mode, the classifier cyclically classifies signals with a user selectable interval.
Depending on the complexity of the signals and the selected classification bandwidth, after a few seconds the results of the classification attempt are listed in the results window. Each signal in the list is labeled and numbered as Signal n, and in manual mode this corresponds to the Sn signal marker in the FFT window – the marker is a box surrounding the signal envelope. The classifier processing state is visible in the status bar. After each restart, the results window is cleared.
The actual range of classification depends on the frequency settings. Thus it is important that the mode corresponds to the actual input signal frequency range.
The success of classification attempts is highly dependent on the selection of the proper sampling time. To recognize low speed signals, the higher sampling rate, i.e., 3.2 s should be selected.
It is worth keeping in mind that the classifier is based on statistically based estimates of the properties of the input signals, and many factors determine the reliability of these results – among these are signal quality and application settings. However, using the classifier in conjunction with common sense and good monitoring provides a powerful aid in analyzing unknown signals.