摘要
基于不同需求提出了一种比较通用的指标评价函数,构建了地面站资源配置的统计模型和基于神经网络方法的仿真元模型。在元模型求解过程中,提出了回代法寻找最优配置的思想,即将每次迭代求解得到的优化配置点代入下次迭代训练中,以提高网络的逼近性能和泛化能力。算例表明,该方法能够求得地面资源的较优配置,求解效率高。
A universal indicator evaluation function meeting different requirements is proposed and a statistic model and a simulation meta-model of ground resources allocation based on neural network method are established.The paper also provides a method of back substitution to get best allocation in the solving process of the meta-model,i.e.,getting the optimal allocations of each iteration substitute to the next iteration of study to improve approximation performance and fitting capability.The example of calculation demonstrates that the method gets optimal ground resources allocation with high solving efficiency.
出处
《飞行器测控学报》
2010年第4期1-6,共6页
Journal of Spacecraft TT&C Technology
关键词
测控站资源
优化配置
元建模
神经网络
TT&C Station Resources
Optimal Allocation
Meta-Modeling
Neural Network