Augstas izskirsanas spejas spektrala analize, pielietojot bazes funkciju adaptacijas pieeju

Spectral analysis is a common part of the signal theory dealing with possibilities of representation of signals in the frequency domain. The subject of the investigations is very wide, therefore in the present work attention is fixed on band-limited signals registered during the finite time interval. Selection of this particular signal class is not accidental because a lot of practical tasks in the field of signal processing faced with or can be reduced onto band-limited signal analysis, for instance, speech, radio and communications signals. Moreover main part of spectral analysis methods are developed without taking into account information could be extracted from frequency domain and based on frequency range unlimited signal models. In the current work a band-limited signal features are studied and hypothesis in order to estimate spectrum of different signal components are proved. A method for continuous time signal spectral analysis is developed. The proposed method is distinguished by the possibility of its application both for uniformly and nonuniformly sampled signals. Development of new high frequency resolution algorithms for spectral analysis satisfying the present-day requirements and its implementation by using advanced signal processing tools represents a topical problem of signal processing. The original basis functions adaptation approach is proposed to distinguish this goal and practically realizable algorithms for discrete signal processing are developed. The computer simulations results confirn effectively of new algorithms. Comparison with well-known classical, parametrical and modern spectral analysis algorithms for uniformly and nonuniformly sampled signals with corresponding classification of signal models and analysis methods is done. Nowadays nonuniformly digitized signal analysis is one of increasing interests directions for theoretical investigations and the research work has been done in this field allows more deeply understand problems connected with the subject of these thesisSeparate summaries in English, RussianAvailable from Latvian Academic Library / LAL - Latvian Academic LibrarySIGLELVLatvi