基于遗传算法优化神经网络的结冰环境中MVD和LWC预测

Prediction of MVD and LWC in Icing Environment Based on Genetic‑Algorithm‑Optimized Neural Network

水滴平均体积直径(Mean volumetric diameter,MVD)和液态水含量(Liquid water content,LWC)是两个影响飞机结冰的重要气象参数,但在实际中难以准确测得,如果能够实时、准确地获取这两个参数可以为积冰预测和飞机适航认证标准的建立提供一些指导。文中提出了一种基于遗传算法优化神经网络的结冰气象参数预测模型。以不同测点组合的冰厚和结冰速率、环境温度、飞行速度和机翼迎角为输入参数,结冰气象参数MVD和LWC为输出参数,构建遗传算法优化的结冰气象参数预测模型,并通过预测模型对数值计算测试组数据和结冰风洞实验数据的结冰气象参数进行预测。结果表明,基于遗传算法优化Elman神经网络的预测模型对结冰气象参数的测试组预测相对误差在10%以内,实验数据相对误差在20%以内,该方法具有一定的可行性。

The mean volumetric diameter(MVD)and liquid water content(LWC)are two important parameters that affect aircraft icing,but they are difficult to be measured accurately in practice.If these two parameters can be accurately obtained in real time,they can provide some guidance for icing prediction and the establishment of aircraft airworthiness certification standards.In this paper,a prediction model of icing meteorological parameters based on the genetic⁃algorithm⁃optimized neural network is proposed.We use the ice thickness and icing rate of different combinations of measuring points,ambient temperature,flight speed and wing angle of attack as input parameters,and the icing meteorological parameters MVD and LWC as output parameters,and develop a prediction model of icing meteorological parameters optimized by the genetic algorithm,This model predicts the icing meteorological parameters of the numerical calculation test group data and the icing wind tunnel experiment data.The results show that the relative error of the prediction model for numerical calculation of icing meteorological parameters of the test group is within 10%,and the relative error of the experiment data is within 20%.This method is feasible.