ALGORITHM OF SYNTHESIS OF DISCRETE TESTS BASED ON EVOLUTIONARY THEORIES

Abstract

The complexity of modern digital components and devices as objects of diagnosis, in most cases, makes it impossible for experts to synthesize tests manually in the time allowed to solve the problem of organizing diagnostic tests. This determines the relevance of the development and implementation of intelligent methods and tools for automated synthesis of test sequences. One of the promising options for the synthesis and optimization of discrete tests is the solution of this problem based on the theory of evolution or genetic algorithms.

The paper describes an algorithm for the synthesis of discrete tests using genetic algorithms. The algorithm of realization of diagnostic experiment contains stages of formation of initial population, crossing, processing of the received results and check of a condition of a stop of work of genetic algorithm. The main characteristics of the algorithm: the type of evolutionary algorithms used - hybrid; method of generating binary words of chromosome codes by crossing operator from values of diagnostic vectors of elementary tests; the method of selection of applicants for crossing - proportional, based on the probability function of fitness and the control operator of the compatibility assessment; type of crossover operator - uniform; the method of forming a new generation - the descendants of crosses of individuals based on the results of proportional selection and elitistly selected individuals that do not allow crossings under the control operator of compatibility assessment; the main way to stop the evolutionary algorithm - the loss of the ability to implement crossings of individuals in the resulting generation; additional (possible) ways to stop the algorithm - limiting the execution time or the number of iterations (generated generations).