用模糊BP神经网估算小卫星成本

Cost Estimating for Small Satellite Using Fuzzy BP Neural Networks

提出了用BP神经网对卫星成本进行二维估算 ,即先求出样本点集扩散到控制点集中的模糊数 ,然后将对应同一控制输出变量的控制输入变量的模糊数作为学习样本 ,让BP神经网进行学习。学习后的BP神经网就能在给定输入变量的情况下 ,估算出输出变量。最后 ,给出了小卫星成本的质量和定点精度模糊BP神经网估算的实例 ,取得了满意的结果。

A new method for estimating satellite's cost based on fuzzy BP neural networks is presented in this paper. Sample set is spread through control set. Control input fuzzy set corresponding to the same control output is viewed as learning sample which is transmitted for BP neural networks to study. After learning, if given some input, the output can be obtained. In order to estimate the satellite's cost, we take satellite system parameters as input and satellite's cost as output. Finally, an example about estimating the cost of optical remote sensing small satellite is presented which is successfully solved with the method. As an example, the method is used to estimate the cost of small satellite according to the two parameters: mass and pointing accuracy. The results show that the method is powerful and feasible.\;