为实现高超声速飞行器姿态自抗扰控制的参数整定,提出一种模糊Q学习算法。首先,采用强化学习中的Q学习算法来实现姿态自抗扰控制参数的离线闭环快速自适应整定;然后,根据模糊控制的思路,将控制参数划分为不同区域,通过设定奖励,不断更新...为实现高超声速飞行器姿态自抗扰控制的参数整定,提出一种模糊Q学习算法。首先,采用强化学习中的Q学习算法来实现姿态自抗扰控制参数的离线闭环快速自适应整定;然后,根据模糊控制的思路,将控制参数划分为不同区域,通过设定奖励,不断更新Q表;最后,将训练好的Q表用于飞行器的控制。仿真结果表明,相对于传统的线性自抗扰控制(linear active disturbance rejection control,LADRC)和滑模控制,基于Q学习的LADRC省去了人工调试参数的繁琐过程,且仍具有良好的跟踪效果。蒙特卡罗仿真测试结果验证了基于Q学习的LADRC的鲁棒性。展开更多
According to the geological structural features, Beijing and the adjacent areas can be divided into two regions of plain in the east and mountain in the west. Among the stations covered by the telemetered digital seis...According to the geological structural features, Beijing and the adjacent areas can be divided into two regions of plain in the east and mountain in the west. Among the stations covered by the telemetered digital seismic station network of Earthquake Administration of Beijing Municipality, the stations in the plain area are all borehole ones and the stations in the western mountainous region are all located on the surface bedrock. In the paper, 511 wave- form data recorded by the network from Oct. 2001 to Oct. 2004 are used in the researches for the entire Beijing region, the western mountainous region and the eastern plain area, respectively. The Q values are calculated for each area by Atkinson′s method and compared with the existed data. The reliability of the Q values and the reasons for the difference in the Q values are also discussed. Then, the source parameters and site response are inverted by the Moya′s method, in which two models are used. The first model uses the Q values, earthquakes and stations in the sub-areas and the second model uses the Q values, earthquakes and stations in the entire Beijing region. The results indicate that the source parameters and site responses obtained by two models are basically consistent with each other. It also indicates that the source parameters obtained by these methods are not affected by the size of station network.展开更多
Based on 49 digital seismograms recorded by 73 seismic stations in the Jiangsu Telemetered Seismic Network,the paper uses Atkinson's method to calculate the inelastic attenuation coefficient of the Jiangsu area. W...Based on 49 digital seismograms recorded by 73 seismic stations in the Jiangsu Telemetered Seismic Network,the paper uses Atkinson's method to calculate the inelastic attenuation coefficient of the Jiangsu area. We find that the frequency-dependent Q in the Jiangsu region is Q( f) = 272. 1·f^(0. 5575). We also use Moya's method to invert the 63 stations' site responses. The results show that the site responses of the 25 stations in Jiangsu are approximately 1 at a range between 1Hz and 20 Hz, which is consistent with their basements on rocks. The response curves of the site responses of the 14 underground stations are similar to each other. Their site responses show an amplification at low frequencies and minimization at high frequencies. The calculation of the Brune model on the waveform data of M_L≥2. 5 earthquakes from Jiangsu Digital Seismic Network between October 2010 and May 2015 in terms of seismic source parameters of 58 seismic waves shows that there are good correlations between seismic magnitude and other source parameters such as seismic moment, source radius and corner frequency, while the correlations between seismic magnitude and stress drop,and stress drop and source radius are not so good.展开更多
Based on 310 horizontal-component digital seismograms recorded at 14 seismic stations in Shanxi Digital Seis-mograph Network, the inelastic attenuation coefficient in Shanxi region is studied. By the methods of Atkins...Based on 310 horizontal-component digital seismograms recorded at 14 seismic stations in Shanxi Digital Seis-mograph Network, the inelastic attenuation coefficient in Shanxi region is studied. By the methods of Atkinson and Moya, the site response of each station and several source parameters are obtained and the inversion results from both methods are compared and analyzed. The frequency-dependent inelastic attenuation coefficient Q is estimated as Q( f )=323.2 f 0.506. The site responses of 14 seismic stations do not show significant amplification, which is consistent with their basement on rock. We also found the dependence of corner frequency on seismic moment, seismic moment on stress drop, source radius on stress drop.展开更多
文摘为实现高超声速飞行器姿态自抗扰控制的参数整定,提出一种模糊Q学习算法。首先,采用强化学习中的Q学习算法来实现姿态自抗扰控制参数的离线闭环快速自适应整定;然后,根据模糊控制的思路,将控制参数划分为不同区域,通过设定奖励,不断更新Q表;最后,将训练好的Q表用于飞行器的控制。仿真结果表明,相对于传统的线性自抗扰控制(linear active disturbance rejection control,LADRC)和滑模控制,基于Q学习的LADRC省去了人工调试参数的繁琐过程,且仍具有良好的跟踪效果。蒙特卡罗仿真测试结果验证了基于Q学习的LADRC的鲁棒性。
基金Project of Center for Analysis and Prediction of CEA, Key Project of State Science and Technology in the Tenth Five-year Plan (02-03-04), Joint Seismological Science Foundation of China (604022), and Natural Science Founda-tion of Beijing Municipality (8022009).
文摘According to the geological structural features, Beijing and the adjacent areas can be divided into two regions of plain in the east and mountain in the west. Among the stations covered by the telemetered digital seismic station network of Earthquake Administration of Beijing Municipality, the stations in the plain area are all borehole ones and the stations in the western mountainous region are all located on the surface bedrock. In the paper, 511 wave- form data recorded by the network from Oct. 2001 to Oct. 2004 are used in the researches for the entire Beijing region, the western mountainous region and the eastern plain area, respectively. The Q values are calculated for each area by Atkinson′s method and compared with the existed data. The reliability of the Q values and the reasons for the difference in the Q values are also discussed. Then, the source parameters and site response are inverted by the Moya′s method, in which two models are used. The first model uses the Q values, earthquakes and stations in the sub-areas and the second model uses the Q values, earthquakes and stations in the entire Beijing region. The results indicate that the source parameters and site responses obtained by two models are basically consistent with each other. It also indicates that the source parameters obtained by these methods are not affected by the size of station network.
基金jointly sponsored by the“Science for Earthquake Resilience(1730801)the Youth Fund Program of Earthquake Administration of Jiangsu Province,China(201405)
文摘Based on 49 digital seismograms recorded by 73 seismic stations in the Jiangsu Telemetered Seismic Network,the paper uses Atkinson's method to calculate the inelastic attenuation coefficient of the Jiangsu area. We find that the frequency-dependent Q in the Jiangsu region is Q( f) = 272. 1·f^(0. 5575). We also use Moya's method to invert the 63 stations' site responses. The results show that the site responses of the 25 stations in Jiangsu are approximately 1 at a range between 1Hz and 20 Hz, which is consistent with their basements on rocks. The response curves of the site responses of the 14 underground stations are similar to each other. Their site responses show an amplification at low frequencies and minimization at high frequencies. The calculation of the Brune model on the waveform data of M_L≥2. 5 earthquakes from Jiangsu Digital Seismic Network between October 2010 and May 2015 in terms of seismic source parameters of 58 seismic waves shows that there are good correlations between seismic magnitude and other source parameters such as seismic moment, source radius and corner frequency, while the correlations between seismic magnitude and stress drop,and stress drop and source radius are not so good.
基金National Scientific Research Target Project of China (2001-BA601B01-01-05) during the tenth Five-year Plan.
文摘Based on 310 horizontal-component digital seismograms recorded at 14 seismic stations in Shanxi Digital Seis-mograph Network, the inelastic attenuation coefficient in Shanxi region is studied. By the methods of Atkinson and Moya, the site response of each station and several source parameters are obtained and the inversion results from both methods are compared and analyzed. The frequency-dependent inelastic attenuation coefficient Q is estimated as Q( f )=323.2 f 0.506. The site responses of 14 seismic stations do not show significant amplification, which is consistent with their basement on rock. We also found the dependence of corner frequency on seismic moment, seismic moment on stress drop, source radius on stress drop.
