Abstract: In order to overcome the disadvantages of the traditional hierarchical analysis method which can’t denote fuzziness and time-varying system, a niche genetic algorithm was used to examine the consistency of the dynamic triangular fuzzy number reciprocal judgment matrix, correct its matrix elements and order the weights. The method takes the consistency examination，matrix correction and weights ordering as an integral process. The genetic algorithm was adopted to optimize the nonlinear function of consistency index coefficient. The algorithm can make the corrected judgment matrixes have smaller consistency index coefficients on the basis of correcting the initial judgment matrixes to the smallest extend. The complexity and stability of the algorithm were analyzed, and its effectiveness was also proved by an example comparison. The theoretical analysis and example comparison show that the algorithm can take advantage of the initial judgment matrix information and meet the requirements of off-line assessment. And it has the stability in weights ordering. It is meaningful in overcoming the fuzziness and time-varying features of evaluation systems.

Key words: system evaluation and feasibility analysis, dynamic triangular fuzzy number, reciprocal judgment matrix, consistency index coefficient, genetic algorithm