摘要
直升机旋翼系统的工作方式及其所承受的载荷形式使飞行实测载荷数据的有效性不高。研究如何利用有限的实测载荷及飞行参数数据建立直升机旋翼系统飞行参数识别模型,对于推进飞行载荷测试任务有重要意义。基于Matlab编程建立遗传算法优化的BP神经网络直升机旋翼系统飞行参数识别模型,实现通过现有载荷数据及飞参数据对旋翼系统飞行载荷预测仿真。预测的最大相对误差为10%、平均相对误差为3.7%,满足工程要求,并且较未使用遗传算法优化的BP神经网络预测结果好,表明所建立的飞行参数识别模型具有很好的学习能力和泛化能力。
The working ness of measured flying 1 manner and loading form of helicopter rotor system cause the low effective- oading data. It is of important significance to study how to use limited meas- ured loading and flying parameters data to establish helicopter rotor system recognition model of fly- ing parameters for carrying the task of measuring flying loading a step forward. Helicopter rotor sys- tem recognition model of flying parameters in this paper was established using BP networks optimized by genetic algorithm based on Matlab. This model realized the prediction and simulation of rotor＇ s flying loading with available loading and flying parameters data. The maximum relative error between the predicted results and the measured data was 10%, the average 3.7%, which met the engineer- ing demand. Furthermore, the predicted results were better than that of BP networks without genetic algorithm. It demonstrated that the recognition model of flying parameters in this paper had good learning ability and generalization ability.
出处
《直升机技术》
2016年第4期25-28,共4页
Helicopter Technique
关键词
直升机旋翼系统
飞行实测载荷数据
飞行参数识别模型
遗传算法
BP神经网络
helicopter rotor system
measured flying loading data
recognition model of flying pa- rameters
genetic algorithm
BP networks
