DEVELOPMENT OF A METHOD FOR COMPUTER ANALYSIS OF SPECTROMETRIC SIGNALS PARAMETERS USING DISCRETE WAVELET TRANSFORM

Abstract

Purpose of the work. Due to the rapid development of computer technology, computer analysis methods are actively used to process digitized spectrometric signals and construct spectra. The aim of the work is to increase the accuracy of recognition and measurement of spectrometric signal parameters in computer systems of spectral analysis by developing a new data processing method. Methodology. In the process of computer analysis, digital signal processing methods, discrete wavelet transform algorithms, methods and algorithms for intelligent analysis of large data sets are used. To generate digital images of spectrometric signals, mathematical and computer modeling methods are used. As part of the research, a cross-platform program was developed that allows for simulation of a digitized signal, data processing using existing or new approaches, and visualization of the results. This software tool was developed in the C++ programming language using the QT framework, which provides the ability to create cross-platform software. Scientific novelty. For the first time, a method of computer analysis of digitized spectrometric signals was developed, the feature of which is signal filtering with automatic determination of the level of electrical noise using discrete wavelet transform algorithms and the adaptive BayesShrink algorithm, as well as additional processing of pulse superposition, which allowed to increase the accuracy of spectrometric data processing. Conclusions. As part of the development of a new method for computer analysis of spectrometric signals, an approach to filtering the signal from noise with automatic determination of the noise threshold and an algorithm for recognizing pulse parameters with correction of their amplitudes during superposition were presented. To comprehensively verify the recognition accuracy and speed of the proposed method in comparison with several existing approaches, input data for analysis with fully known parameters with an artificial, idealized distribution of pulse amplitudes were simulated. After that, analysis of such data was performed using various methods. The presented results of computer processing indicate that the proposed analysis method allowed to increase the accuracy of recognition of spectrometric signal parameters in the studied scenarios, in comparison with alternative approaches. One of the directions for further research is to verify the operation of the developed method on data sets obtained during real experiments using spectrometric equipment and a digitizer.