基于声发射和BP神经网络的预应力钢筋砼梁损伤过程分析

Prestressed Concrete Beam Damage Process Based on Acoustic Emission and BP Neural Network Analysis

以预应力钢筋混凝土梁三点弯曲试验为基础,绘制了声发射幅值、能量等AE参数相关图,揭示了预应力砼梁的损伤演化过程;借鉴NDIS-2421定量评定标准,对试验梁的损伤程度做出定性的评价;同时,通过AE特征信号Kurtosis指标的计算对演化过程进一步分析,确定4个典型失效阶段后,建立各失效阶段声发射信号特征参数数据,并设计BP神经网络模型进行训练,训练后的定型网络对梁损伤程度有很好的识别能力。可以为工程中进一步开展预应力钢筋砼结构损伤活动性的长期健康监测提供理论基础和指导。

On the basis of the three point bending test of prestressed reinforced concrete beam,mapped the AE amplitude and energy related AE parameters,such as figure,revealed the damage evolution process of prestressed concrete beam.According to quantitative assessment reference NDIS-2421 standard,the qualitative evaluation of the damage degree of the tested beam was made.At the same time,through the calculation of characteristics of AE signal Kurtosis index,we determined four typical failure stage to establish the failure stage characteristics of acoustic emission signal parameters database,and designed the BP neural network model for training.The formed network after training has great ability to recognize the damage degree of the beams.It will provide a theoretical basis for long-term health prediction of prestressed RC beams.