Evolutionary search for processing the results expert compsrisons several homogeneous reliability

Purpose. There are regulations requirements for reliability in the projecting and exploitation of new technical elements. The determining task in the parameters of reliability is appeared. It is a requirement in the latest specifications. There are reliability indices among the standardized requirements for the development of new technical elements and systems. There are contradictions for the latest systems. It is necessary to give an estimate of reliability, when there were no long-term experiences of their long exploitation. The traditional method for receiving of reliability indices of equipment or technical systems is the statistical analysis of the results of their long-term exploitation. Statistical method is not a solution of this problem, but we think that using the method of expert estimations can be a better way to solve this problem. In particular, this situation became a real problem for tube gas heaters (TGH). These heaters are widespread for heat supply to industrial plants. To other side, the term of exploitation TGH is counted for only a few years. To get real statistical estimation is not possible on short-term exploitation. In this case for getting estimation of reliability parameters we can use the method of expert estimations. In particular, Thomas Saaty method of pairwise comparison of the factors. Following, T. Saaty method we can create a matrix of pairwise comparisons using approximate scale of importance. Traditionally the T. Saaty matrix is processing by finding the vector of their own values. The matrix must be completely filled. The result of the vector of matrix value depends on possible fluctuations in the values of this matrix. For processing of this matrix we offer to use an evolutionary random search of more preferable (Rs-optimal) solutions. According to formulation the purpose of choosing solutions in the form of rules of preference, the evolutionary search expands capabilities of expert estimations, including incorrect possible of looking for solutions though using of additional criterion (stabilizer). Methodology. General algorithm of evolutionary random search of more preferable (Rs-optimal) solutions is used. The rule of preference (choice of solutions) is create, which let to get the desired solution in evolutionary random search. Findings. The estimation of failure rates efficiency for tube gas heaters according to the Thomas Saaty matrix and evolutionary random search of more preferable solutions has been completed. The application of evolutionary random search of more preferable solutions allows to improve the processing of expert estimates using T. Saaty matrix. Such a method provides the ability to process the results of of expert estimations incomplete T. Saaty matrix. And to consider the additional matrix, except this one information on the compared objects at construction function selection decisions. Originality. The approach to the processing of the results of expert estimation of the matrix T. Saaty using evolutionary random search of more preferable solutions is proposed. Practical value. Using evolutionary search for the processing of expert estimations, which is based on the Thomas Saaty method. It gives advantages in the flexibilities to formulation requirements for the procedure of processing of expert estimations.