COMPARISON OF SIGNAL RECOGNITION METHODS BY COMBINED USE OF APPROPRIATE EVALUATION CRITERIA WITHIN THE ADDITIVE CONVOLUTION

Existing signal recognition methods have both their advantages and disadvantages, which are found when recognizing signals from classes defined by different characteristic standards. Therefore, for signals from different classes, the indicators of recognition quality by one method or another can differ significantly. There is a need to create a more balanced method capable of providing the necessary stability relative to the accuracy and reliability of the final results in the process of recognizing signals from various classes. As such signal recognition method, the article proposes to use an approach based on the combine using of weighted signal proximity criteria within the additive convolution. Euclidean distances between reference points are used as evaluation criteria, which are used in the context of applying the four most well-known recognition methods: the amplitude method (the trivial Euclid method), the DDTW method using the values of the first derivatives, and methods based on the Wavelet transform and the Fourier transform.