Network coding can improve utilization of the limited number of wavele ngth converters in wavelength division multiplexing (WDM) optical networks.But network coding operations may increase the computational overhead,complexity an d buffer demand.In order to achieve optical multicast capacity with less number of network coding operations as possible,an improved genetic simulated annealing algorithm (GSAA),which combines the genetic algorithm and an individua l acce pting strategy based on simulated annealing algorithm,is proposed in this paper.By designing an algorithm fitness function, which can both reach the maximal multicast throughput and meet the constraint of wavelength continuity in the WD M network,the proposed algorithm can reduce the number of network coding operations.To enhance the feasibility of part c hromos omes in population space,a part of feasible artificial chromosomes are added into the initial population.Moreover,a local search proc ess and an adaptive mutation operation are designed to improve the local search ability and the convergence speed in each round of t he proposed algorithm’s iteration.Compared with other genetic algorithms,the simulation results show that the improved genetic simulated annealing algorithm can get better convergence speed and fewer number of network coding operations in WDM networks.