CONSTRUCTED FEATURES FOR AUTOMATIC CLASSIFICATION OF STATIONARY TIMING SIGNALS

The challenge for automation to reduce the dimension of the data has been solved for the construction of diagnostic and recognizing patterns, based on steady-state one-dimensional signals, distributed in time. The set of indicators characterizing the properties of the signals, which allows to reduce the description of the classified objects is firstly proposed. The experiments on studying of the proposed indicator set at practical problem solving s are conducted.