Abstract: Studying on the problem of maintenance strategy decision making of equipment can reduce the maintenance cost and improve the availability, reliability, safety of a machine, and product quality. Although it has had a certain extent of development, the problem of maintenance strategy decision making with mixed data has never been resolved. The local concordance index and discordance index is defined based on the stochastic dominance to compare one maintenance strategy with others under stochastic evaluation; the fuzzy evaluations are transformed to the corresponding fuzzy numbers, then, the local concordance index and discordance index are defined based on the distance index between the centroid point of a fuzzy number and original point to resolve the problem of maintenance strategy decision making under fuzzy evaluation. Based on the above processing, the fuzzy and stochastic evaluations are treated the same the the deterministic in ELECTRE III in order to solve the problem of maintenance strategy decision making with mixed data. The proposed approach is applied to selecting the most appropriate maintenance strategy for the steam turbine of steam-power plant. The result shows that the approach is sound and effective.

Key words: ELECTRE III, Steam turbine, Maintenance strategy, Mixed data, Multiple attribute decision making, Ranking fuzzy numbers, Stochastic dominance