基于人工神经网络的混凝土疲劳寿命估算

Fatigue life estimation of concrete based on artificial neural network

把决定混凝土疲劳寿命的应力水平、应力比、加载频率和描述随机性的破坏概率作为神经网络的输入 ,以疲劳寿命作为网络的输出 ,用网络结构描述它们之间的非线性关系。利用实验数据样本完成了网络的训练 ,并对模拟样本进行了估算 ,实例计算结果表明 ,利用已有疲劳试验数据训练好的人工神经网络进行混凝土疲劳寿命估算 ,可部分代替冗长、昂贵的疲劳实验。

Taking the main factors such as stress level, stress ratio, loading frequency and failure probability, which directly affect the fatigue life of concrete, as the input of neural network, and the fatigue life as the output of it, the unlinear relation between them is expressed by the test data. The calculation results of the actual example indicate that the concrete fatigue life estimation through the pre-trained neural network can partly substitute for some long-winded and expensive fatigue tests.