基于RBFNN的卫星地面站系统建模方法

System Modeling of Satellite Ground Station Based on RBF Neural Networks

如何建立卫星地面站系统的快速模型,是卫星地面站资源配置优化和长远规划中必须解决的问题。提出了基于径向基神经网络的卫星地面站系统建模方法。分析评价指标,确定输入输出参数和取值范围,以仿真数据为基础,通过选择合适的径向基函数和利用遗传算法改进ROLS算法的方法,能够在有限的训练样本下获得具有较好测试结果的神经网络模型。最后针对案例进行了验证。实验证明,方法能够建立可信的卫星地面站模型,算法复杂度低、运行速度快,适用于卫星地面站的资源配置优化和长远规划。

How to build smart models of ground stations is a key point of the resource optimization and the planning of satellite ground stations. In this paper, a RBFNN - based modeling method of satellite ground station system is proposed. In this algorithm, the evaluation specifications of satellite ground station are analyzed and the inputs and outputs of RBFNN are chosen. The ROLS method is enhanced by using genetic algorithm and selecting compatible radial basis function. The presented algorithm is verified by examination in the end. The result shows that it can help to build a credible model of satellite ground station, and has a low complexity, good efficiency, and is suitable for the resource optimization and planning of satellite ground station.