Estudio y comparación de métodos modernos y métodos inteligentes para la determinación del comportamiento en tiempo-frecuencia de una señal : desarrollo de métodos Inteligentes tiempo-frecuencia para reducir la carga computacional.

Abstract

This study shows the results obtained in the development of ""Intelligent"" methods to reduce the computational load. The process with which the different frequencies of interest can be obtained, and the interpretation of the information obtained with respect to the processing time of each of the stages. A signal composed of several independent components is used as a basis, with increasingly closer frequencies, which allow evaluating its Frequency Resolution, and increasingly smaller amplitudes, which allow evaluating the detection capacity of very small amplitudes, and with determined durations, which allow evaluating the Time Resolution, for any Time-Frequency method. For this study we will use the time-frequency diagram, spectrum plots using the 1-Hann window, which generates a narrower main lobe width than the rectangular window, and Blackman Harris, which generates very small secondary lobes of the order of 92 dB below the peak of the main lobe. In this study, 6 tests will be performed with different artificial signals to show and compare the capability of the methods with the mentioned characteristics. In this study, 2 tests and 6 scenarios with different artificial signals will be performed to show and compare the capability of the methods against the mentioned characteristics. The first scenario has standard parameters, the amplitudes are equal to one and the frequency components are considerably separated. The second scenario evaluates the frequency resolution characteristics. The third scenario evaluates the very small amplitude detection characteristics. The fourth scenario evaluates the Frequency Resolution and Small Amplitude Detection characteristics. The fifth scenario evaluates the temporal resolution. Finally, the sixth scenario evaluates the system behavior with the addition of white Gaussian noise.