STATISTICAL MODELING OF COMPUTER EQUIPMENT RELIABILITY USING MATLAB

Abstract

The article discusses a method for creating a random number generator with a diffusion-nonmonotonic distribution in the MATLAB program. The method is developed on the basis of using a standard random number generator with a uniform distribution. The use of a diffusion-nonmonotonic distribution is the most promising in determining and modeling the reliability of electronic equipment and computer equipment using the probabilistic-physical method. Statistical modeling using a random number generator allows you to obtain statistical estimates of all quantitative reliability indicators during a statistical experiment by modeling failures of real electrical components. The transformation of random numbers with a uniform distribution generated by built-in Matlab generators into a diffusion-nonmonotonic distribution according to the Fokker-Planck formula with a left reflecting and right absorbing screen is performed by inversion transformation. As a result of the research, a transformation algorithm was developed and a program was obtained that converts random numbers to MATLAB. Verification of the obtained results using the Pearson criterion with a significance level of 99% confirms that the obtained random variables correspond to a diffusion-nonmonotonic distribution. Comparison of the results of modeling according to the Pearson criterion with the results of similar programs of other developers shows the best reproducibility of the developed algorithm by 6.92%. The developed generation method allows obtaining random numbers according to the Fokker-Planck formula with a left reflecting and right absorbing screen corresponding to the diffusion-nonmonotonic distribution by converting uniformly distributed numbers of the MATLAB generator. The presented random number generator with diffusion-nonmonotonic distribution significantly expands the capabilities of the MATLAB program in statistical modeling of random processes, and the developed algorithms can be used for other similar software products.