Monitoring of long-term static deformation data of Fei-Tsui arch dam using artificial neural network-based approaches

Abstract

SUMMARY The objective of this paper is to develop methods for extracting trends from long‐term static deformation data of a dam and try to set an early warning threshold level on the basis of the results of analyses. The static deformation of a dam is mainly influenced by the water pressure (or water level) of the dam and the temperature distribution of the dam body. The relationship among the static deformation, the water level, and the temperature distribution of the dam body is complex and unknown; therefore, it can be approximated by static neural networks. Although the static deformation almost has no change during a very short time, it changes with time for long‐term continuous observation. Therefore, long‐term static deformation can be approximated dynamically using dynamic neural networks. Moreover, static deformation data is rich, but information is poor. Linear and nonlinear principal component analyses are particularly well suited to deal with this kind of problem. With these reasons, different approaches are applied to extract features of the long‐term daily based static deformations of the Fei‐Tsui arch dam (Taiwan). The methods include the static neural network, the dynamic neural network, principal component analysis, and nonlinear principal component analysis. Discussion of these methods is made. By using these methods, the residual deformation between the estimated and the recorded data are generated, and through statistical analysis, the threshold level of the static deformation of a dam can be determined on the basis of the normality assumption of the residual deformation. Copyright © 2011 John Wiley & Sons, Ltd.

Data-Based Models for the Prediction of Dam Behaviour: A Review and Some Methodological Considerations

“…Kao and Loh [42] proposed a two-step procedure: first, the number of neurons was fixed whereas the optimal amount of iterations was computed. Second, NN models with different number of hidden nodes were trained with the selected amount of iterations, and the final architecture was chosen as the one which provided the lowest error in a validation set.…”

“…If the dependency is non-linear, it may lead to misinterpretation of the results. Non-linear principal component analysis (NPCA) can be an alternative, as showed by Loh et al [52] and Kao and Loh [42], who applied it by means of auto-associative neural networks (AANN) to predict radial displacements in an arch dam.…”

A method for dam health diagnosis based on the improved attribute reduction

“…Targeting at the single measurement point, these models regard the relationship between the monitoring effect size and the environmental variable as a certain cause-and-effect relationship [1,2]. However, modeling efforts combining multiple monitoring measurement points with multiple monitoring effect sizes are insufficient [3,4]. Besides, few scholars take into account uncertainty of health diagnosis, which serves as a basis for dam health diagnosis.…”

Tehnike racunarske inteligencije u modeliranju i identifikaciji indikatora ponasanja brane

“…Kao & Loh [127] користе ауто-асоцијативну неуронску мрежу за моделирање померања бране. У раду је применом поступка параметарске индентификације нелинеарних система и NARX структуре модела формиран модел померања 13 тачака бране.…”

“…У раду је применом поступка параметарске индентификације нелинеарних система и NARX структуре модела формиран модел померања 13 тачака бране. Хидростатичко оптерећење има већи утицај на померања од термичког оптерећења, па се као улази у модел узимају само претходне измерене вредности нивоа воде у акумулацији [127].…”