基于神经网络的多卫星姿态协同控制预测模型

摘要： 针对多卫星近距离对空间目标监测中的姿态协同控制问题，研究基于神经网络的智能控制模型.首先设计了面向任意姿态控制方法的神经网络自适应控制模型，该模型不改变卫星姿态控制方法本身，而是接收由理想控制模型生成的理想控制量，理想控制量经过神经网络后直接生成适用于卫星的控制量.根据神经网络对控制系统的系统级状态预测，基于一致性控制协议，设计了多卫星姿态协同控制模型.采用等效姿态角，通过数值仿真，分别验证了预测控制方法、输出状态预测、协同控制的方法正确性和有效性.

A Multi-Satellite Attitude Cooperative Control Model Based on System-Level State Prediction by Artificial Neural Network

Abstract： An intelligent control model is proposed to solve the problem of attitude cooperative control for multi satellite close range monitoring of space targets. Firstly, a neural network adaptive control model for arbitrary attitude control method is designed. The model does not change the attitude control method itself, but receives the ideal amount of control generated by the ideal control model. The ideal control amount directly generates the control amount suitable for the satellite after passing through the neural network. The multi satellite attitude cooperative control model is derived, utilizing system level state prediction of control system by neural networks, and consensus control protocol. The correctness and effectiveness of prediction control, control result prediction, and cooperative control are demonstrated by numerical simulation using equivalent attitude angle.