基于模拟退火-单纯形法的模型参数估计研究

Estimation of Model Parameters Based on Simulated Annealing Simplex Algorithm

针对模型参数估计问题难以准确求解的不足,提出了一种模拟退火和单纯形算法结合的混合优化算法.该算法利用模拟退火的随机全局搜索能力和单纯形算法的确定性多面体搜索策略,把这2种算法进行结构上的组合,通过采用新的反射操作,构成了模拟退火单纯形算法用来求解带有约束的优化问题.对7种测试函数的实验结果表明:该混合优化算法比传统模拟退火算法和单纯形算法有着更好的搜索精度.最后将该算法运用在了模型参数估计问题上,能够准确地辨识出模型参数,证明了该算法在模型参数估计问题中的有效性.

There are some difficulties in underestimation of the problem to solve accurately for model parameters,and we propose a simulated annealing and the simplex algorithm combining hybrid optimization algorithm.The algorithm uses the simulated annealing random global search ability and the certainty of the simplex algorithm polyhedron search strategy,a combination of these two algorithms structure,reflected by the introduction of new operations,constitutes a simulated annealing algorithm is used for solving simplex there are constrained optimization problem.The experimental results of seven test functions show that the hybrid optimization algorithm than the traditional simulated annealing algorithm and simplex algorithm has a better search accuracy.Finally,the algorithm used in the model parameter estimation on the issue,can accurately identify the model parameters to prove the effectiveness of the algorithm in the model parameter estimation problems.