7.2 Parallelization test to MOPSO in calibration

The performance of parallelized MOPSO (parMOPSO) algorithm is tested by using a test function of low complexity and a semi-distributed conceptual hydrological model of higher complexity.The computational time of the algorithm is also evaluated and regarded as the main criteria the algorithm performance.The test function and the model were implemented on PC with 4 IntelCore (TM) 2，2.4GHz processors (Quad).

7.2.1 Test on function

The first testing case is a nonlinear bimodal function to seek the global Pareto front by using parMOPSO algorithm.The function was proposed by Debet al (2002) as below：

Minimize：

f1(x)=x1

Minimize：

where

The population size of the parMOPSO algorithm was set to be 200，and thenumber of generation was set to be 200，the mutation rate was set to be 0.5 whilethe ......     （共4070字）    [阅读本文]>>