基于T-S模糊神经网络的变形预报研究  

Deformation prediction research based on T-S fuzzy neural network

作　　者：成枢[1] 荆晓宇 李圳[1,2] CHENG Shu;JING Xiaoyu;LI Zhen(College of Geomatics, Shandong University of Science and Technology, Qingdao 266510, China~ 2. Chinese Academy of Surveying and Mapping, Beijing 100830, China)

机构地区：[1]山东科技大学测绘科学与工程学院,山东青岛266510 [2]中国测绘科学研究院,北京100830

出　　处：《测绘工程》2018年第1期37-41,共5页Engineering of Surveying and Mapping

摘　　要：T-S模糊神经网络模型是根据模糊系统和人工神经网络优缺点具有明显的互补性结合而成。文中基于T-S模糊神经网络,将其应用于变形预测。通过实测数据和仿真数据分析比较了其与BP神经网络、小波神经网络在预测精度、算法稳定性和有效区间3个评价标准上的优劣。结果表明,在变形预测,特别是利用长周期监测数据进行预报时,T-S模糊神经网络具有一定的优势。T-S fuzzy neural network model is based on the fact that the advantages and disadvantages of fuzzy system and artificial neural network have obvious complementarity, which is used in deformation prediction. Through the analysis of measured data and simulation data, comparison on advantages and disadvantages of three evaluation standards between the T-S fuzzy neural network and BP neural network, a wavelet neural network has been carried out. The three evaluation standards include prediction precision, algorithm stability and effective interval. As a result, in the deformation prediction, especially in the forecasting by using long period monitoring data, the T-S fuzzy neural network has some certain advantages.

关 键 词：模糊理论 模糊神经网络 变形预测 精度 稳定性 有效区间 

分 类 号：TU196[建筑科学—建筑理论]