Drabich O. Statistical Analysis of Hidden Periodicities in the Systems for Determination of Technical Object Characteristics

The thesis is dedicated to the development of discovering and analysis methods of hidden periodicities based on their models in the form of periodically correlated random processes (PCRP) for the purpose of their application in data measuring systems for technical object characteristic determination. Using a small parameter method, a theoretical investigation of PCRP correlation period estimate properties was conducted. The estimates are derived per continuous and discrete Fourier transforms of signal realization sections, and also correlation and covariation products. For PCRP typical models, an analysis of estimate statistical characteristic dependencies was carried out on realization section length, discretization step and signal parameters describing its probabilistic structure. By means of computer simulation model of typical PCRP, an experimental investigation of developed methodology was performed. On its basis the structure of vibration signal stochastic recurrence was analyzed. The coherent estimate properties of noised pulse oscillation signal parameters were explored. The results of parameter noise-immunity analysis were used at development of analog and digital transducers at author participation.