7.3 Comparison of three multi-objective optimization algorithms

Three criteria of Pareto optimal are proposed by Zitzler et al.(2008) toassess the performance of the algorithms： GD (Generation Distance) betweensolution set drawn by the algorithm and the ideal Pareto set; non-dominantPareto front spacing rate SP (Spacing); the ratio of number of solution setsdrawn by algorithm that do not belong to Pareto optimized solution by ER(Error).However，due to the complexity of hydrological models，the idealPareto solutions of the objectives cannot be drawn，and therefore when comparing the three optimization methods under the same iteration times，Pareto solution dominant rate，Pareto front spacing rate and iteration time can be usedas evaluation criteria.

7.3.1 The spacing rate of non-dominant solutions

Spacing rate is an important criterion to evaluate ......     （共5835字）    [阅读本文]>>