基于模糊遗传算法的埋入式电阻热布局优化

Thermal placement optimization of embedded resistances by fuzzy genetic algorithm

针对标准遗传算法优化埋入式电阻热布局存在的过早收敛等问题,通过设计适应度函数、采用模糊逻辑控制器自适应调整交叉概率和变异概率,以及对长时间未进化的种群实施局部灾变等措施维持种群多样性,使算法最终收敛于全局最优解。仿真结果表明,该算法能够更好地抑制早熟收敛,算法优化布局结果的温度分布更平均,并通过热成像仪对实验样件进行温度分布测试验证了算法的有效性。

Aiming at the premature convergence problem of the standard genetic algorithm （SGA） for thermal placement opti- mization of embedded resistances, a series of measures were taken to maintain the population diversity and search the global opti- mization solution, such as the fitness function design, the crossover and the mutation probabilities adaptive adjustment by the fuzzy logic controller （FLC）, and the local catastrophe operation intervention when the population hadn＇t evolved for a long time. The MATLAB results show that this method can suppress the premature convergence better, and the finite element analysis（FEA） simula- tion results indicate that its thermal placement optimization result has more equal temperature distribution. The effectiveness of the algorithm was verified by testing the temperature distribution of experimental samples using the thermal imager.