论文部分内容阅读
为实现对承载后柔性机翼挠度的准确预测,在全面分析柔性机翼挠度的影响因素基础上,应用正交试验法确定的影响柔性机翼挠度的主要因子作为输入变量,挠度作为输出变量,以大量试验数据为训练样本,通过多次试取隐含层和各隐含单元,并选取trainlm作为最优训练函数,最终建立了预测柔性机翼挠度的BP(Back Propagation)人工神经网络模型.在此基础上,随机选取试验结果中的12组试验样本,连续进行10次挠度预测,预测结果和试验实测值最大相对误差和标准方差分别为4.481%,1.033 7.解析结果表明:柔性机翼挠度预测结果与实验值吻合的较好,建立的人工神经网络预测模型具有较高的预测精度.
In order to realize the accurate prediction of flexible wing deflection after bearing, based on a comprehensive analysis of the factors affecting the deflection of flexible wing, the main factors affecting the deflection of flexible wing determined by orthogonal test are taken as input variables and deflections as output variables, Taking a large amount of experimental data as the training sample, BP (Back Propagation) artificial neural network model for predicting flexible wing deflection is finally established by testing hidden layers and hidden units multiple times and selecting trainlm as the optimal training function. On this basis, randomly selected 12 test samples in the test results, the prediction of deflection for 10 times in a row, the maximum relative error and standard deviation are 4.481%, 1.033 7. The analytical results show that: flexible wing The deflection prediction results are in good agreement with the experimental ones. The established artificial neural network prediction model has higher prediction accuracy.