To reduce the influence of greenhouse effect, pollution gas emission should be considered in economic dispatch problems. The reason is that a mass of pollution gas emission is produced by thermal generating sets. Naturally, renewable energy source such as wind should be considered to reduce pollution gas emission. Totally, original economic dispatch problems are transferred into wind/emission/emission dispatch problems. However, large execution time of economic dispatch is occupied by the addition of wind. Further, time-consuming computation and time-sensitive dispatching are inconsistent. Thus, to save the time-consuming problems, a surrogate-based multi-objective optimization method is proposed. In detail, original time-consuming objective functions are replaced by data-driven fuzzy neural network models. Large execution time is saved by the applying of surrogate models. Then, an enhanced mould algorithm is proposed for improving diversity and convergence of original algorithm. Improved algorithm obtains excellent performances in some benchmark test functions and wind/emission/emission dispatching problems. The excellent performances in this paper are illustrated by IEEE 40-units test systems with two wind turbines.