随机优化的改进交叉熵方法

随机优化的交叉熵方法具有高效性和自适应性的特点，在高维和非线性等复杂优化问题中具有巨大的开发潜力。针对传统交叉熵优化方法精度不足的缺点，提出使用“当前精英样本”和“全局精英样本”构建新的参数更新策略，以充分提取迭代历史中的有用信息。采用自适应的平滑策略和变异操作进一步提升计算性能。通过3个计算实例证明，改进后的方法比传统交叉熵方法具有更高的计算精度和更强的全局搜索能力。

Stochastic optimization method based on improved cross entropy

Cross entropy method is an efficient and adaptive stochastic optimization method and has im-mense potential in complex optimization problems with high dimension and nonlinear constraints.However, the traditional cross entropy method is lack of accuracy.In this study, both the concepts of current elite sam-ples and global elite samples are introduced to extract more useful information from the whole iterative history. Then, a new parameter updating strategy is established based on these two concepts.New adaptive smoothing strategy and mutation operation are also applied to improve its computing performance.The proposed algorithm is illustrated by three numerical examples.The computational results indicate that the improved cross entropy method has higher calculation accuracy and better global search capability.