基于T-S模型的混凝土坝变形缺失信息填补方法

T-S model-based method for imputation of missed deformation monitoring information of concrete dam

针对混凝土坝变形监测信息缺失的问题,依据同类测点空间变形上的关联性,提出了T-S模型填补方法。即:首先建立以邻近相关测点变形信息为输入、缺失测点变形信息为输出的T-S模糊模型;然后运用CFSFDP-FCM聚类算法进行模型前件辨识;最后基于交替优化的最小二乘法进行后件参数求解和缺失值填补。实例分析结果表明,该方法可有效挖掘测点间变形的关联关系,填补精度和有效性明显优于常规的反距离加权、多测点回归及BP网络填补方法,且对多测点缺失情况亦具有较好填补效果。

Aiming at the problem of the concrete dam deformation monitoring information missing,a T-S model-based imputation method is proposed herein in accordance with the correlation of the spatial deformations of similar measuring points;for which a T-S fuzzy model is established at first by means of taking the deformation information of the relevant adjacent measuring points as the inputs and the deformation information of the missing measuring points as the outputs,and then the model antecedents identifications are carried out with the CFSFDP-FCM clustering algorithm,finally,both the solution of the consequent parameters and the imputation of the missed values are made as well.The analysis on the actual case shows that the correlative relations among the deformations of the measuring points can be effectively mined with this method,of which the imputation accuracy and effectiveness are obviously better than the imputation methods of inverse distance weighting,multi-point regression and BP network,while the method has a better effect of imputation for the situation of the information missing of multi-measuring points.