【作者】 安二； 王保国；

【Author】 Er AN;Baoguo WANG;School of Aerospace Engineering,Beijing Institute of Technology;

【机构】 北京理工大学宇航学院；

【摘要】 基于小波神经网络(Wavelet Neural Network,WNN),本文建立了一种预测飞行品质的飞行员-飞机闭环系统数学模型。借助于Mc Ruer模型获得了小波神经网络的训练样本数据库,该数据库包含3125个飞行员的驾驶行为特性。用自己编制的程序和建立的数据库,通过反复迭代的方法来训练初始的小波神经网络模型。当小波神经网络的结构参数及层间权重矩阵收敛后,便可认为训练结束,此时小波神经网络的数学模型就建好了。本文还探讨了输入两种不同信号时,该数学模型中人机闭环系统的性能与飞行品质的预测。5个样本的预测结果表明该模型具有一定的可行性和良好的工程应用价值。更多还原

【Abstract】 A mathematical model of closed-loop pilot-aircraft system to predict aircraft flying qualities was proposed based on wavelet neural network(WNN).With Mc Ruer model a training sample datebase,which includes 3125 pilot behavior characteristics,was created to train the WNN.The WNN model was trained by iteration using the program compiled by ourselves and the created datebase.When the structure parameters of WNN and the weight matrix between each layer were convergenced,the training was thought to be over and the mathematical model of WNN was built.Besides,the system performance and aircraft flying qualities were discussed when the determined model worked for two different input signals separately.The predicting results of five samples indicate that this model is feasible and has good engineering application value.更多还原