基于最小二乘法的飞行员模型参数辨识  

Pilot Model Parameters Identification Based Least Square Method

作　　者：高健[1] 郑淑涛[1] 韩俊伟[1] 

机构地区：[1]哈尔滨工业大学机电工程学院,黑龙江哈尔滨150001

出　　处：《江南大学学报（自然科学版）》2011年第6期648-652,共5页Joural of Jiangnan University (Natural Science Edition)　

基　　金：教育部新世纪优秀人才支持计划项目(NCET-04-0325)

摘　　要：针对飞行员与飞机相匹配的特点,采用最小二乘法对飞行员模型进行参数辨识。根据Hosman感知模型的特点,确立待辨识参数,为提高辨识精度,重点分析了辨识数据野值的剔除和补正。在俯仰工况下,利用最小二乘递推算法辨识参数,并在研究用Boeing 737-800飞行模拟机上对该方法进行了试验验证。结果表明,飞行员模型的仿真输出曲线与飞行测试数据之间的升降舵角度误差小于0.3°,俯仰角误差小于0.5°,说明通过该方法获得的模型参数能够反映飞行员实际的操作行为。Taking the performance of pilot adaptation to the aircraft into account, the pilot model parameters are identified by using the least squares method. It is validated which parameters will be identified according to characteristics of Hosman perception model, as well as outliers of data are deleted and corrected for improving identification precision. The recursive least squares is used to identify the model parameters in pitch task, and verified on a research flight simulator. The results show that the error for elevator angle between the output curves of pilot model and the test datum for nominal aircraft is less than 0. 3 deg, and that for pitch angle less than 0.5 deg. These smaller errors indicate the pilot model parameters can correctly describe pilot＇s manipulation behavior.

关 键 词：飞行员模型 参数辨识 最小二乘法 数据野值 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术] V249.11[自动化与计算机技术—计算机科学与技术]