Performance-improved TSVR-based DHM model of super high arch dams using measured air temperature

Considering that the displacement of super high arch dams is sensitive to temperature variations and dam temperature field is essentially modulated by the ambient temperature, a measured air temperature-based hydrostatic-thermal-time (HTT) displacement health monitoring (DHM) model of super high arch dams is proposed. To fully explore the complex nonlinearity between dam displacement and its explanatory variables and to improve the predictive accuracy of the model, twin support vector regression (TSVR) is introduced to establish HTT model. Due to the influence of TVSR parameters on model performance, a performance-improved TSVR is proposed by utilizing grey wolf optimizer (GWO) algorithm to determine the optimal TSVR parameters in this study. The availability of the proposed model is tested with measured data of a super high arch dam. Results show that, compared with the harmonic functions in the commonly-used hydrostatic-seasonal-time (HST) models, the detailed variation characteristics of thermal displacement can be better captured with the measured air temperature factors in the proposed HTT model. Meanwhile, the fitting and predicting accuracy of the performance-improved TSVR-based DHM model is significantly improved. And the proposed model has an excellent long-term predictive capability. The conclusion can be drawn that the proposed model is feasible and effective for the analysis and prediction of dam displacement.

Automated summary

A new measured air temperature-based displacement health monitoring model for super high arch dams is proposed in this study, utilizing twin support vector regression and optimized using grey wolf optimizer algorithm. Experimental results demonstrate that the new model can better capture the thermal displacement variations, significantly improving predictive accuracy and showing excellent long-term predictive capability.