A novel adaptive neural control strategy is exploited for the longitudinal dynamics of a generic flexible air-breathing hypersonic vehicle（FAHV）.By utilizing functional decomposition method, the dynamics of FAHV is ...A novel adaptive neural control strategy is exploited for the longitudinal dynamics of a generic flexible air-breathing hypersonic vehicle（FAHV）.By utilizing functional decomposition method, the dynamics of FAHV is decomposed into the velocity subsystem and the altitude subsystem.For each subsystem, only one neural network is employed for the unknown function approximation.To further reduce the computational burden, minimal-learning parameter（MLP）technology is used to estimate the norm of ideal weight vectors rather than their elements.By introducing sliding mode differentiator（SMD） to estimate the newly defined variables, there is no need for the strict-feedback form and virtual controller.Hence the developed control law is considerably simpler than the ones derived from back-stepping scheme.Finally, simulation studies are made to illustrate the effectiveness of the proposed control approach in spite of the flexible effects, system uncertainties and varying disturbances.展开更多
结合非结构网格对复杂外形精确描述的能力和IP(Information Preservation)方法降低统计耗散的特点,该文在非结构网格的基础上构建了改进型通量分裂式DSMC(Direct Simulation of Monte Carlo)-IP数值方法,对三维高超声速稀薄气体流场进...结合非结构网格对复杂外形精确描述的能力和IP(Information Preservation)方法降低统计耗散的特点,该文在非结构网格的基础上构建了改进型通量分裂式DSMC(Direct Simulation of Monte Carlo)-IP数值方法,对三维高超声速稀薄气体流场进行了数值模拟。采用渐变尺度的非结构网格划分流场区域,在满足DSMC方法对网格尺度需求的同时,对流场的结构特点进行精确描述;另外根据Van Leer格式的思想改进IP方法中的通量计算格式,解决了激波处的数值间断问题;并采用单元信息保存法对IP方法进行补充,提高了计算方法的稳定性。通过对三维圆球绕流和航天飞机头部绕流流场的数值模拟,验证了该文方法的有效性和可行性,并对模型中的气动特性和参数分布规律进行了分析。展开更多
基金supported by the Aeronautical Science Foundation of China (No.20130196004)
文摘A novel adaptive neural control strategy is exploited for the longitudinal dynamics of a generic flexible air-breathing hypersonic vehicle（FAHV）.By utilizing functional decomposition method, the dynamics of FAHV is decomposed into the velocity subsystem and the altitude subsystem.For each subsystem, only one neural network is employed for the unknown function approximation.To further reduce the computational burden, minimal-learning parameter（MLP）technology is used to estimate the norm of ideal weight vectors rather than their elements.By introducing sliding mode differentiator（SMD） to estimate the newly defined variables, there is no need for the strict-feedback form and virtual controller.Hence the developed control law is considerably simpler than the ones derived from back-stepping scheme.Finally, simulation studies are made to illustrate the effectiveness of the proposed control approach in spite of the flexible effects, system uncertainties and varying disturbances.
文摘结合非结构网格对复杂外形精确描述的能力和IP(Information Preservation)方法降低统计耗散的特点,该文在非结构网格的基础上构建了改进型通量分裂式DSMC(Direct Simulation of Monte Carlo)-IP数值方法,对三维高超声速稀薄气体流场进行了数值模拟。采用渐变尺度的非结构网格划分流场区域,在满足DSMC方法对网格尺度需求的同时,对流场的结构特点进行精确描述;另外根据Van Leer格式的思想改进IP方法中的通量计算格式,解决了激波处的数值间断问题;并采用单元信息保存法对IP方法进行补充,提高了计算方法的稳定性。通过对三维圆球绕流和航天飞机头部绕流流场的数值模拟,验证了该文方法的有效性和可行性,并对模型中的气动特性和参数分布规律进行了分析。
