FAST索网节点位移的T-S模糊建模

发布时间：2018-05-24 05:58

本文选题：FAST + T-S模糊模型　；参考：《东北大学》2014年硕士论文

【摘要】：天文射电望远镜又被叫做无线电望远镜。与传统天文望远镜相比他具有很多优势(穿透性强、观测距离远、观测范围广),因而在目前天文观测中倍受青睐。500m口径球面射电远镜(FAST)的主动反射面有2400多个节点。只有对2400多个节点实现实时动态定位和控制,FAST主反射面才能顺利地完成观测任务,因此建立合理的节点运动轨迹模型具有十分重要的意义。FAST索网节点运动过程具有多输入、非线性、时变、大惯性等特点,而且节点数量多以及干扰因素多,不便于直接测量。本文以FAST索网50m缩尺模型为背景,利用T-S模糊模型对索网节点的位移建立数学模型。本文将模糊前后件分开辨识。首先,利用FCM算法辨识模糊前件的结构和参数,然后通过最小二乘法辨识模型后件的参数。为了更好地建立模型,提高模型的辨识精度与收敛速度,第四章改进了第三章的辨识方法。针对FCM算法的弊端(对初始值敏感、训练速度慢、容易陷入局部极小等),第四章利用减法聚类辨识前件结构与参数。然后用PSO-BP的混合算法代替最小二乘法来辨识后件参数,最后根据密云50m缩尺模型现场数据进行仿真实验。仿真结果显示,改进后的T-S模糊模型能较好预测节点的位移模型。当模型规则数以及各参数一定时,基于模糊C-均值聚类算法和最小二乘法的T-S模糊模型只能辨识出唯一精度的均方误差。如果想提高精度,就要增加模糊规则数,但是这样就使模型变得复杂。改进的T-S模糊模型利用PSO-BP混合算法在本质上解决了上述问题。本文提出不增加模糊的规则数而将辨识精度提高的模型更适用于实际工程领域。根据预测的节点运动轨迹模型,观测人员就可以及时处理与调控主网,保证主索网变位工作正常运转。因此,改进的T-S模糊模型不仅具有理论研究意义,更具有实际应用价值。
[Abstract]:Astronomical radio telescopes are also called radio telescopes. Compared with the traditional telescope, it has many advantages, such as strong penetration, long observation distance and wide observation range, so it has more than 2400 active reflectors in the present astronomical observation. Only by realtime dynamic positioning and controlling the fast main reflector of more than 2400 nodes can the observation task be successfully completed. Therefore, it is very important to establish a reasonable node motion trajectory model. The motion process of fast cable network node has multiple inputs. Nonlinear, time-varying, large inertia and so on, and the number of nodes and interference factors are many, so it is not easy to measure directly. In this paper, based on the 50m scale model of FAST cable mesh, the T-S fuzzy model is used to establish a mathematical model for the displacement of cable mesh node. In this paper, the fuzzy front and rear parts are identified separately. Firstly, FCM algorithm is used to identify the structure and parameters of the fuzzy front part, and then the parameters of the model are identified by the least square method. In order to establish the model better and improve the accuracy and convergence rate of the model, the fourth chapter improves the identification method in chapter 3. In view of the disadvantages of FCM (sensitive to initial value, slow training speed and easy to fall into local minima), the fourth chapter uses subtraction clustering to identify the structure and parameters of the former parts. Then the hybrid algorithm of PSO-BP is used instead of the least square method to identify the parameters of the second part. Finally, the simulation experiment is carried out according to the field data of the 50m scale model of Miyun. The simulation results show that the improved T-S fuzzy model can better predict the node displacement model. The T-S fuzzy model based on the fuzzy C-means clustering algorithm and the least square method can only identify the unique mean square error when the number of the model rules and parameters are fixed. If we want to improve the precision, we need to increase the number of fuzzy rules, but this makes the model complex. The improved T-S fuzzy model uses PSO-BP hybrid algorithm to solve the above problems in essence. In this paper, it is proposed that the model with improved identification accuracy without increasing the fuzzy rule number is more suitable for the practical engineering field. According to the predicted node motion trajectory model, the observer can handle and regulate the main network in time to ensure the normal operation of the main cable network. Therefore, the improved T-S fuzzy model not only has theoretical significance, but also has practical application value.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TH751

【相似文献】