一种航空发动机燃油流量基线的建模方法

为实现航空发动机的在巡航过程中的实时监控及时发现发动机状态参数的异常变化,提高飞行安全水平,提出基于航空发动机燃油流量(FF)基线求解偏差值的一种算法。依据设定的飞行数据筛选原则和预处理方法建立模型样本,设计以高斯函数为隐含层激励函数和以线性函数为输出层激励函数的多输入单输出的RBF神经网络,通过Pearson相关性分析确定网络的输入节点。使用该网络得到预测燃油流量基线,再与实际燃油流量做比较可得燃油流量偏差值。最后对预测偏差值和观测偏差值实施两配对非参数检验以验证网络精度,结果表明该方法是计算航空发动机巡航状态下燃油流量偏差值的一种有效算法。

A Modeling Method for Fule Flow Baseline of the Aero-engine

For achieving real-time monitoring for the aero-engine during cruise phase, promptlyScatching abnormal shift of aero-engine status parameters and improving flight safety level,proposed a calculation method of fuel flow shift value based on fuel flow baseline.According to given data screening rules and pre-processing methods,built the model sample.Designed the multi-input and single-output RBF neural network with the Gaussian function selected as hidden layer transfer function and the Linear function selected ed as ouput layer transfer function,and input nodes were confirmed by Pearson correlation analysis.The predicted baseline was gotten by this model,and then fuel flow shift value was gotten by comparing the predicted baseline and actual fuel flow.Finally,Did two-sample matched-pairs nonparametric tests for observed values and predicted values to verify network accuracy, The results indicate that this method is an effective approach for calculating the fuel flow shift value of the aero-engine during cruise.