基于进化神经网络的受火钢筋混凝土柱损伤预测

Damage prediction of reinforced concrete columns by fire basing on evolutionary neural network

钢筋混凝土柱在火灾中的损伤情况是建筑消防界的研究热点和难点。考虑钢筋混凝土柱尺寸、混凝土强度、受火时间等影响因素,基于用遗传算法进行优化权值后的神经网络,建立受火钢筋混凝土柱极限承载力的预测模型,进而确定柱子的损伤等级。研究表明该方法的预测结果与实测结果吻合较好,对评价钢筋混凝土柱在火灾中的损伤情况具有重要意义。

Damage degree of reinforced concrete columns by fire is a frequent issue and nut of discussion in the field of architectural fire--protection. Factors, like the size of reinforced concrete columns, the intensity of reinforced concrete and the firing time were taken into consideration; Basing on Neural Network whose weight was optimized by GA, a model to predict the limited bearing capacity of reinforced concrete columns by fire was formed, therefore, the damage hierarchy of reinforced concrete columns can be confirmed. Research findings reflected that the prediction result of this model can be well coincident with the actual data, which shows gear significance in confirming the damage degree of reinforced concrete columns by fire.