中文摘要：

混凝土面板堆石坝主要靠上游面板挡水,施工期面板裂缝是影响面板堆石坝质量和安全的关键因素之一.面板混凝土属于准大体积混凝土,尽管其散热面较大、最高水化热温升不是很高,但如果不采取合理的温控防裂措施,仍然会导致面板产生温度裂缝.针对施工期面板混凝土温度裂缝问题,以某在建高混凝土面板堆石坝工程为例,建立了包含地基的三维有限元温度场和徐变温度应力场仿真模型,利用数值方法产生神经网络的学习样本,然后采用遗传算法优化的BP神经网络对所获得的样本进行网络训练,从而获得温控措施优选网络模型,进行已知混凝土热力学材料参数情况下的温控措施优选.由神经网络优选结果可知,本文所采用的面板混凝土温控措施优选方法是合理可行的.

英文摘要：

Concrete face rock-fill dam （ CFRD） depends mainly on the concrete slab located on the upper stream to retain water, so concrete slab crack problem is one of the key factors influencing the quality and safety of CFRD during construction period. Since the slab concrete be-longs to quasi-mass concrete, the radiating surface is big, and the highest hydration heat temperature rise is not very high, but it can still cause temperature crack if do not take reasonable measures to control it. In view of the temperature crack problem of concrete slabs during construction period, this paper taking a high concrete face rock-fill dam project under construction as an example, a three-dimensional finite element simulation model of the temperature field and creep temperature stress field including the foundation was established. Using the nu-merical method to produce neural network learning samples, and using the BP neural network optimized by genetic algorithm to train the sam - ple, thereby it obtained the neural netw-ork optimization model for temperature control measures. Then optimizing of temperature control meas-ures for concrete slabs under the condition that concrete material perfor^nance was fixed. The neural network optimization result show-s that, the optimization method of temperature control measures for concrete slabs that this paper adopted is reasonable and feasible.