该文基于美国国家浮标资料中心(National Data Buoy Center,NDBC)浮标观测数据对哨兵一号搭载的合成孔径雷达(synthetic aperture radar,SAR)反演风速数据进行精度分析,并利用BP神经网络(back propagation neural network)对SAR反演风...该文基于美国国家浮标资料中心(National Data Buoy Center,NDBC)浮标观测数据对哨兵一号搭载的合成孔径雷达(synthetic aperture radar,SAR)反演风速数据进行精度分析,并利用BP神经网络(back propagation neural network)对SAR反演风速的偏差进行校正;同时针对环境要素、BP神经网络训练输入的样本量以及神经网络结构参数设计了敏感性试验;最后将SAR标量风场数据转换为用u、v矢量表示的风场数据,并对u向风和v向风分别进行了精度分析和校正。实验结果表明:SAR反演风速相较于浮标观测数据出现了低估现象;经过BP神经网络校正后,SAR反演风速数据的精度得到了改善,风速的平均偏差绝对值从0.78 m s下降到0.04 m s,均方根误差从1.98 m s下降到了1.77 m s;敏感性试验表明输入质量较差的环境要素数据时BP神经网络的校正效果有所下降,而增加训练集样本量能改善校正效果;将标量风场数据转换为u、v矢量风场数据后的校正结果也显示BP神经网络具有较好的校正效果。展开更多
In this paper, a curved path following control algorithm for miniature unmanned aerial vehicles（UAVs） in winds with constant speed and altitude is developed. Different to the widely considered line or orbit followin...In this paper, a curved path following control algorithm for miniature unmanned aerial vehicles（UAVs） in winds with constant speed and altitude is developed. Different to the widely considered line or orbit following, the curved path to be followed is defined in terms of the arc-length parameter, which can be straight lines, orbits, B-splines or any other curves provided that they are smooth. The proposed path following control algorithm, named by VF-SMC, is combining the vector field（VF） strategy with the sliding mode control（SMC） method. It is proven that the designed algorithm guarantees the tracking errors to be a bounded ball in the presence of winds, with the aid of the Lyapunov method and the BIBO stability. The algorithm is validated both in Matlab-based simulations and high-fidelity semi-physical simulations. In Matlab-based simulations, the proposed algorithm is verified for straight lines, orbits and B-splines to show its wide usage in different curves.The high-fidelity semi-physical simulation system is composed of actual autopilot controller, ground station and X-Plane flight simulator in-loop. In semi-physical simulations, the proposed algorithm is verified for B-spline path following under various gain parameters and wind conditions thoroughly.All experiments show the accuracy in curved path following and the excellent robustness to wind disturbances of the proposed algorithm.展开更多
采用径向基函数神经网络(Radical Basis Function Neutral Networks,简称RBF神经网络)来模拟大跨度结构的非高斯风压场.根据某大跨度结构的形式特点,将结构风场看成是屋面位置和时间的函数,将风压场分解为一系列径向基函数.再利用单调...采用径向基函数神经网络(Radical Basis Function Neutral Networks,简称RBF神经网络)来模拟大跨度结构的非高斯风压场.根据某大跨度结构的形式特点,将结构风场看成是屋面位置和时间的函数,将风压场分解为一系列径向基函数.再利用单调非线性无记忆转换映射和RBF中获得的风场函数定义向量过程,从而将非高斯场的模拟转换为互相关高斯过程的模拟.将RBF神经网络应用于一大跨度屋盖的非高斯场模拟,得到结构上非高斯风压场的分布.结果对比表明,RBF神经网络模拟非高斯风压场具有较高的准确性.该方法可直接利用RBF神经网络的输出结果,避免推导高斯过程和非高斯过程的关系式,因此具有较高的效率.RBF神经网络模拟非高斯风压场在准确性和效率上均具有显著优势.展开更多
文摘该文基于美国国家浮标资料中心(National Data Buoy Center,NDBC)浮标观测数据对哨兵一号搭载的合成孔径雷达(synthetic aperture radar,SAR)反演风速数据进行精度分析,并利用BP神经网络(back propagation neural network)对SAR反演风速的偏差进行校正;同时针对环境要素、BP神经网络训练输入的样本量以及神经网络结构参数设计了敏感性试验;最后将SAR标量风场数据转换为用u、v矢量表示的风场数据,并对u向风和v向风分别进行了精度分析和校正。实验结果表明:SAR反演风速相较于浮标观测数据出现了低估现象;经过BP神经网络校正后,SAR反演风速数据的精度得到了改善,风速的平均偏差绝对值从0.78 m s下降到0.04 m s,均方根误差从1.98 m s下降到了1.77 m s;敏感性试验表明输入质量较差的环境要素数据时BP神经网络的校正效果有所下降,而增加训练集样本量能改善校正效果;将标量风场数据转换为u、v矢量风场数据后的校正结果也显示BP神经网络具有较好的校正效果。
基金supported by the National Natural Science Foundation of China under Grant No.61403406
文摘In this paper, a curved path following control algorithm for miniature unmanned aerial vehicles（UAVs） in winds with constant speed and altitude is developed. Different to the widely considered line or orbit following, the curved path to be followed is defined in terms of the arc-length parameter, which can be straight lines, orbits, B-splines or any other curves provided that they are smooth. The proposed path following control algorithm, named by VF-SMC, is combining the vector field（VF） strategy with the sliding mode control（SMC） method. It is proven that the designed algorithm guarantees the tracking errors to be a bounded ball in the presence of winds, with the aid of the Lyapunov method and the BIBO stability. The algorithm is validated both in Matlab-based simulations and high-fidelity semi-physical simulations. In Matlab-based simulations, the proposed algorithm is verified for straight lines, orbits and B-splines to show its wide usage in different curves.The high-fidelity semi-physical simulation system is composed of actual autopilot controller, ground station and X-Plane flight simulator in-loop. In semi-physical simulations, the proposed algorithm is verified for B-spline path following under various gain parameters and wind conditions thoroughly.All experiments show the accuracy in curved path following and the excellent robustness to wind disturbances of the proposed algorithm.
文摘采用径向基函数神经网络(Radical Basis Function Neutral Networks,简称RBF神经网络)来模拟大跨度结构的非高斯风压场.根据某大跨度结构的形式特点,将结构风场看成是屋面位置和时间的函数,将风压场分解为一系列径向基函数.再利用单调非线性无记忆转换映射和RBF中获得的风场函数定义向量过程,从而将非高斯场的模拟转换为互相关高斯过程的模拟.将RBF神经网络应用于一大跨度屋盖的非高斯场模拟,得到结构上非高斯风压场的分布.结果对比表明,RBF神经网络模拟非高斯风压场具有较高的准确性.该方法可直接利用RBF神经网络的输出结果,避免推导高斯过程和非高斯过程的关系式,因此具有较高的效率.RBF神经网络模拟非高斯风压场在准确性和效率上均具有显著优势.
