Computer modeling of spectrometric signals with increased detailing

Abstract

Relevance. Computer methods for analyzing digitized spectrometric signals have been widely developed. However, there is a problem of objectively assessing the effectiveness of a particular digital processing method due to the randomness of processes at the input of the radiation detector, which can be solved by computer modeling of signal images with fully known parameters. To better match real digitized pulse signals, simulation algorithms should implement the possibility of a random shift in the time of occurrence and amplitude of pulses relative to the digitizer sampling points.

The goal of the research is to increase the accuracy of computer modeling of spectrometric signals by developing modeling algorithms with increased detailing of the digital signal image. An additional task is to estimate the maximum margin of error in determining pulse parameters using simple analysis methods, which arises when determining the pulse amplitude and which is caused by the process of discretization at different frequencies.

Research methods are based on mathematical and computer modeling, include numerical methods, and use proprietary algorithms. As part of the research, an application is being developed that is capable of generating digital images of spectrometric signals with specified parameters and contains software-implemented methods for their analysis.

The results. An approach to modeling pulse signals using an adjustable coefficient of their amplitude and time parameters detailing was developed. Using this approach, a detailed image of a signal with a constant amplitude of pulses and their random distribution in time was generated, and the modeling of its discretization process was carried out at different clock frequencies. The obtained data were analyzed using the Maximum method and the maximum margins of errors in recognizing amplitudes, which are caused by the discretization process itself, were determined.

Conclusions. The developed modeling algorithm allows to programmatically generate a digital image of a signal with a discreteness that brings it closer to a real analog signal. At the same time, the accuracy of its simulation exceeds the resolution of modern digitizers by at least an order of magnitude. The results of research using the developed algorithms indicate that to minimize margin of errors caused by discretization, it is necessary to use digitizers with a clock frequency of more than 1 GHz or to use mathematical methods to restore information partially lost as a result of the discretization process.