智能机器人基于作业子空间划分的自学习自适应控制方法

现有的机器人自适应控制基本上都是在建立机器人线性化的动力学模型的基础上,采用某种显式或隐式参数辨识的方法,在线地修正控制作用.本文针对机器人运动和动力学参数变化的固有特点,提出一种完全不同的自学习自适应方法.这种方法基于智能机器人分级系统中的两级结构,并且在空间域里而不是在时间域里处理机器人参数的变化.把机器人的作业空间划分成子空间,其中包括重力载荷的作用,每个子空间对应一组控制器.规划的轨迹映射到作业空间形成子空间序列.用自学习方法选择与这个序列对应的最佳控制器序列.该方法算法简单,计算量小.避开了通常的自适应方法遇到的一系列困难问题.

华北及周边地区层析成像研究

Our study deals with the high resolution body wave tomography in North china and adjacent areas(30°N-43°N，100°E-130°E), where earthquakes occurred many times in history and has a very complicated geological structure. 6870 events recorded at 273 digital seismic stations from CDSN during 1996－2002 and stations settled by Seislab of IGCAS in Bohai Bay area, including 1382 local earthquakes and 5488 teleseismic earthquakes are used in this study. In the data we used, the average number of received stations is greater than 5, the error of picking up direct arrival time is 0.1－0.5s. Before the inversion, we use Checkerboard method to confirm the reliability of result of Local events; use Restoring Resolution Test to confirm the reliability of result of teleseismic events. We also analyzed the effect of different parameters in the inversion. Based the analysis above, the model used in this paper is divided into small blocks with a dimension of 0.33°in the latitude and longitude directions and 5km、15km、30km in depth, and initial velocity model. Using pseudobending method to calculate the ray traveling path, LSQR algorithm to inverse, finally, we got the body velocity images below 25km and above 480km in this area using Joint- inversion with local events and teleseismic events. We made the conclusion at last: (1)at top zone of the south of Sichuan Basin , there exits low velocity anomalies, below 40km is the high velocity zone extend to 300km; (2) Above the 40km of Ordos block exits low velocity zone, while below 40km until 240km, the high velocity anomalies are interlaced by low velocity anomalies. Below 300km, the anomalies are unclear any more; (3) On the whole, the velocity structure below 400km on the mantle transition zone of Eastern China area shows its changes from low velocity to high velocity.

三种权重赋值方法下项目的权重决策

It is a weighting process to justify the importance of different items. This paper is about how the estimator distributes profiling the weights to different items. This paper is about using design of psychological experiment to figure out the profiles of the weights of different ranked items in three weighting methods, and in the experiment we use five topics as experimental materials. At the same time, we controlled the factors such as the familiarity about the topic and the number of items. Then we use the curve estimation to figure out the exact profiles and use ANOVA to test the topic effects. Curve estimation result shows that there is difference between the profiles of the weights of different ranking items in three weighting methods. To PA, the weighting profile is logarithmic curving style, and to DR, the weighting profile is linear style. ANOVA result shows that there are topic effects to different weighting profiles. But to go ahead, when we use point allocation (PA) to the more important items, the topics seems to have little effects. The methods have relatively more effect than the topics. The result also indicates that to PA methods, there is no significant difference the weight profiles between the fix-sum and the open-sum. This result has a contribution to the basic research to the management science and social science.