Given the limitation of traditional univariate analysis method in processing the multicollinearity of dam monitoring data, this paper reconstructs the multivariate response variables by introducing principal component analysis (PCA) method, explores the ways of determining principal components (PCs), and extracts a few PCs that have major influence on data variance. For steady observation series, a control field for the whole observation values has been established based upon PCA; for unsteady observation series that have significant tendency, a control field for the future observation values has been constructed according to PC statistical predication model. These methods have already been applied to an actual project and the results showed that data interpretation method with PCA can not only realize data reduction, lower data redundancy, and reduce noise and false alarm rate, but also be effective to data analysis, having a broad application prospect. dam safety monitoring, multivariate response variables, principal component analysis, data reduction Citation:Yu H, Wu Z R, Bao T F, et al. Multivariate analysis in dam monitoring data with PCA.

Dam behavior is conventionally evaluated with identification models of deformation, seepage, stress, and crack opening. The identification model needs to be described with a complicated and nonlinear function. Wavelet networks based on wavelet frames were used to establish the identification models of dam behavior for the first time. Firstly, time‐frequency analysis for training data was implemented to determine the original structure of the wavelet network. Next, a new method was proposed for iterative elimination of the redundant neurons according to the dependency between the network output and the nodes in the hidden layer. In this method, rough sets theory was used to calculate the dependency. Lastly, this article built the identification models for the displacement and cracks of one concrete arch‐dam with the trained wavelet network. The models can represent the connection between loads and the behavior of the dam. The numerical example shows that the proposed models are reasonable, and the denoising effect of the signal is remarkable.

A large number of hydraulic concrete structures have hidden defects such as cracks, erosion, freeze and thaw, thermal fatigue, carbonization. These hidden defects seriously affect the strength, stability and durability of structures. These problems are studied mainly by single monitoring or diagnosis methods at present. The integration of multiple monitoring and diagnosis methods is not applied widely. Besides, the analysis theory on these problems is not developed very well. The systemic study on the aging mechanism of hydraulic concrete structures, timevariation model and health diagnosis is still not enough. The support for engineering practice is limited. Aimed at these major scientific and technological problems and combined with specific projects, study on detection of hidden defects and health diagnosis of hydraulic concrete structure has been carried out. This study includes the following content: field non-destructive examination of hidden defects of hydraulic concrete structures, seepage detection, the construction of in-situ sensing system, the combination of field detection and in-situ monitoring, the mechanism of crack, freeze and thaw, erosion and carbonization of hydraulic concrete structure, mechanism of combination aging; time-variation model of hydraulic concrete structure, theories and methods for health diagnosis of hydraulic concrete structures.hydraulic concrete structure, hidden trouble detection, health diagnosisTo take full advantage of water resource, more than one hundred thousand dams, water locks, navigation locks and thousands of miles embankment have been built in our country. They produce enormous social and economic benefit and are very important infrastructure. However, the in-

No abstract