Abstract: The original observation signal of dam deformation can be regarded as the superposition of real signal and white noise. A wavelet-cloud prediction model for dam deformation time series analysis is proposed in the present paper by combining wavelet denoising and cloud model to effectively predict dam deformation. Firstly, the multi-resolution analysis of wavelet is used to decompose the original signal into the real signal item and the noise item in the original deformation time series of the dam. Secondly, the cloud model language rules for deformation prediction are created; the principle of maximum membership degree is used to determine the rule predecessor to which the predicted deformation belongs and the corresponding historical cloud which is further combined with the current cloud to generate predictive cloud. The prediction accuracy among traditional statistical model, cloud model, and the proposed wavelet-cloud model is compared with the deformation prediction of a dam as an example. Result demonstrates that the proposed wavelet-cloud prediction model provides more accurate prediction results, offering an effective basis for the safe operation of dam.

Key words: dam deformation prediction, time series analysis, wavelet analysis, cloud model, denoising