自适应退火遗传算法的起重机箱形梁优化

Crane box girder optimization based on self-adaptive annealing genetic algorithm

为克服标准遗传算法的固有缺陷——停滞和早熟现象,将具有较强局部收索能力的模拟退火算法融入其中,对适应函数进行退火拉伸,对接受算子进行退火处理,同时加入自适应机制来改进标准遗传算法的杂交率和变异率,尤其对变异率的调整,使其既能根据个体适应值的大小进行自适应修正,也能随进化状态的改变而改变,从而增强了算法摆脱局部最优解的能力。以最终形成了自适应退火遗传算法进行起重机主梁优化。经实例验证:与原标准遗传算法相比,在保证收敛结果不变的情况下,收敛速度和全局收敛性都得到了较大提高。

To overcome the inherent defects of stagnation and premature convergence in standard genetic algorithm, the simulated annealing algorithm with strong local search capability is integrated into the algorithm,annealing stretching is applied to fitness function,annealing treatment is adopted for acceptation operator,and meantime,self-adaption mechanism is added to improve crossover rate and mutation rate of standard genetic algorithm.Especially adjustment of mutation rate can change the mutation rate automatically according to size of individual fitness and different evolutionary status so as to strengthen algorithm＇s ability to break away from local optimum solution.In the end,self-adaption annealing genetic algorithm is formed for crane main girder optimization.Verification by example and comparison with the standard genetic algorithm shows that both convergence rate and global convergence of the new algorithm improve a lot under the premise of better convergence results.