摘要
文中提出了一种基于深度学习对飞机操纵杆的操纵舒适性预测方法。采用JACK人机工程仿真软件的Animation模块进行推拉杆操纵任务仿真,利用ForceSolver采集人体上肢关节力学数据,进行飞机操纵舒适性客观定量评价。该评价是通过SSP、力矩分析工具的组合方法;通过Bland-Altman法对仿真数据与数学模型计算结果进行一致性对比,通过非参数检验分析了用关节力学参数进行操纵舒适性预测的有效性,进一步分析了关节不同运动方向在负载感知中的敏感度;通过构建PSO算法改进的multi-BiLSTMs模型来对人体上肢关节操纵舒适性进行预测;该方法预测准确度为0.972,较之前模型预测精度提高了3.6%。结果表明:文中所述方法对操纵舒适性预测是可行的。
In this article,efforts are made to present a method of predicting the aircraft joystick’s handling comfort based on deep learning.The Animation module of the JACK human-engineering simulation software is used to simulate the push-and-pullrod operation task.ForceSolver is used to collect the mechanical data on the upper-limb joints,and objective quantitative evaluation of aircraft control comfort is carried out.The evaluation is conducted through the combination of SSP and the torque analytical tools.The Bland-Altman method is used to compare the consistency between the simulation data and the calculation results of the mathematical model;the non-parametric test is carried out,in order to verify that it is effective to employ the joint-based mechanic parameters for prediction of handling comfort;the sensitivity of the joints at different motion directions in load perception is further analyzed.The multi-BiLSTMs model improved by the PSO algorithm is constructed to predict handling comfort of upper-limb joints.The predicting accuracy of this method is 0.972,which is 3.6%higher than that of the previous model.It is shown that this method is feasible for prediction of handling comfort.
作者
张栋
李艳军
曹愈远
ZHANG Dong;LI Yan-jun;CAO Yu-yuan(College of Civil Aviation,Nanjing University of Aeronautics and Astronautics,Nanjing 210000)
出处
《机械设计》
CSCD
北大核心
2023年第3期55-64,共10页
Journal of Machine Design
基金
国家自然科学基金(50705097)
中国民航总局科技基金(MHRD07238)
南京航空航天大学研究生开放基金(kfjj20200725)。
关键词
上肢关节
JACK
深度学习
操纵舒适性
职业病防治
upper-limb joint
JACK
deep learning
handling comfort
prevention and control of occupational disease
