Research On Self-feedback Test Vector Generation Method Based On Genetic Algorithm

With the increase of the complexity and integration of VLSI,the quality of integrated circuit is more difficult to reach the qualified standard.Even small faults can cause serious consequences in integrated circuits,so test of integrated circuits is more important.The generation of test patterns is very important in the whole test process.A test pattern generation method is proposed in this paper,which is optimized on the basis of the self-feedback test patterns generation method based on genetic algorithm.A template matching method is proposed,which can generate test patterns with high fault coverage by using the technology of design for testability and Tetra MAX simulation.This pattern set will be regarded as a reference test pattern set,and the pattern set produced by internal nodes of the tested circuit selected by genetic algorithm through the self-feedback loop will be compared with it,and the matching degree will be used as the fitness value.According to the statistics,it is confirmed that the probability of excellent offspring generated by using the double-points crossover as the crossover operator is higher than that by using the single point crossover.According to the results of experiment on the benchmark circuits ISCAS' 85,the fault coverage of the test pattern set generated by self-feedback test pattern generation method with the matching degree as the fitness value of the genetic algorithm proposed in this paper is increased by 5.5% on average compared with the self-feedback test pattern generation method before optimization;and the LUT resource is saved by 13.42% on average compared with the method of using technology of design for testability to generate test pattern set and store it for backup.Therefore,considering the hardware resources and fault coverage,this method is a one that can meet both requirements.