Shukri N. The hierarchic evolutionary methods of test generation for digital systems

Object of research: the digital logic systems, set to communications of logic elements, inputs and exits in specialised language. Objective of the work: effectiveness increase of the automated diagnosing of digital systems with the devepopment the evolutionary methods of test generation and application of more powerful multiple strategy of observing of output signals. Various strategy of observing of output signals before testing of circuits with memory are considered. Multiple strategy of observing of output signals for circuits with memory at which various states pairs of good and fault circuits can differ during the various moments of time that allows to encrease tests coverage, but requires the big computing resources. The method of multiple-valued logic simulation which allows to process simultaneously distinguished state pares of good and fault circuits is modified. The genetic algorithm of generation of input sequences distinguishing the state pares of good and fault sequential cirquits is developed. The two-level algorithm of test sequence generation is proposed where the genetic algorithm of the first bottom level builds characteristic sequences which allow to set some elements of memory or to distinguish them. Use of these sequences does evolutionary search more "directed" on top level at test generation and encreases its efficiency. Generalisation of a method of distinguishing function for combinational schemes on circuits with memory on the basis of multiple strategy of observing of output signals that allows to raise coverage of checking tests is executed. The dissertation results are used at Institute of applied mathematics and mechanics NASU in the form of the program module of system of simulation and diagnostics ASMID.