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。 Proposed to use BP neural network two-dimensional estimation of satellite costs, that is, to find the sample point set to diffuse to the control point set of fuzzy numbers, and then the same control output variables of the control input variable fuzzy number as a learning sample, let BP Neural network to learn. The learned BP neural network can estimate the output variables given the input variables. Finally, an example of fuzzy BP neural network estimation of the quality of small satellite cost and fixed-point accuracy is given, and satisfactory results are obtained.