Behavior prediction of concrete arch dams requires the interpretation of monitoring data from instrument measurement. Mathematical models based on artificial intelligence algorithms provide an effective approach to interpret the dam behavior from the monitoring data, which can be utilized to model dam displacement as a function of water level, irreversible time effect, and thermal. In our recent study, an improved mathematical model base on kernel-extreme learning machines (KELM) algorithm with long-term daily air temperature monitoring data series has been successfully applied to the displacement prediction modeling of concrete gravity dams. Considering the important contribution of thermal effect on concrete arch dam deformation, the present paper concerns the applicability of the proposed method in the behavior prediction of concrete arch dams. For the sake of improving the efficiency of the mathematical model, the parameters which are crucial to the performance of KELM are selected by the Jaya optimization algorithm. The performance of the proposed automatic parameter optimization-based Jaya-KELM model was compared with the traditional statistical model and other intelligent algorithms. Results demonstrate that the proposed method is a promising and powerful approach for the behavior prediction modeling of concrete arch dams.

混凝土拱坝的行为预测需要从仪器测量中解释监测数据。基于人工智能算法的数学模型提供了一种有效的方法，可以根据监测数据来解释大坝的行为，该模型可用于根据水位，不可逆时间效应和热量来对大坝位移进行建模。在我们最近的研究中，基于核极端学习机（KELM）算法的改进的数学模型与长期的每日气温监测数据系列已成功地应用于混凝土重力坝的位移预测建模。考虑到热效应对混凝土拱坝变形的重要贡献，本文关注该方法在混凝土拱坝性能预测中的适用性。为了提高数学模型的效率，通过Jaya优化算法选择了对KELM性能至关重要的参数。将所提出的基于参数自动优化的Jaya-KELM模型的性能与传统的统计模型和其他智能算法进行了比较。结果表明，该方法对混凝土拱坝的行为预测建模具有广阔的前景。