大柔性变结构系统传输特性的反向传播网络辨识

Back propagation neural network method of identification for the flexible and variable structure system

对大射电望远镜光机电一体化设计中悬索和馈源舱这一大柔性变结构系统的传输特性进行了研究 .由于该系统的结构是动态变化的 ,采用常规方法无法确定其传输特性 ,因此用改进的反向传播算法对其进行了辨识 ,得到了系统的反向传播网络模型 .从测试数据的网络输出结果与理论计算结果的对比可以看出 ,该网络模型能够较好地反映这一系统的输入 /输出特性 .这一结果对大射电望远镜系统的整体建模和馈源舱位置的间接测量都具有意义 ,也为柔性结构系统传输特性的研究提供了一种可资借鉴的方法 .

The transfer characteristics of the flexible and variable structure cables and cabin system used in LT mechanical, electronic and optic integrated design is studied. Since the structure of this system is dynamic, we can not get its transfer characteristics by routine methods. This flexible and variable structure sysgtem is identified by the ameliorative back propagation neural network algorithm and the back propagation neural network model of this system has been obtained. Comparing the theoretical results of testing data with the neural network output results of testing data, the authors draw a conclusion that this neural network model can well express input/output characteristics of this flexible and variable structure system. This kind of neural network modeling method is helpful to modeling LT system and indirectly measuring the position of the cabin. It is also a reference method for the research on flexible and variable structure systems transfer characteristics.