基于自振频率与神经网络方法的简支梁预应力识别

Prestress Identification for Simple Beam Based on Natural Frequency and Neural Network

在桥梁工程领域大量使用预应力梁预制技术,其构件中现有预应力大小的识别对预应力结构和构件的性能检测鉴定起到至关重要的作用。初步探讨了基于自振频率测试与神经网络技术识别简支梁的预应力方法。通过简支梁模型试验,测得不同张拉力条件下的自振频率,面向BP神经网络识别技术,对测试数据进行处理。分别探讨了以部分试验数据构造神经网络训练样本,然后应用构造的BP网络来识别试验结果的方法。研究结果表明,就所试验研究的情况而言,80%的识别结果其精度在可接受范围,对于预应力预制构件的检测具有实用意义。

Precast pre - stressed beams are widely used in bridge engineering. Identification of value of prestress existing in components plays an important role in detection and identification on the performance of prestressed structures and components. A preliminarily study for prestress identification for simple beam based on natural frequency measurement and neural network technology is carried out. With simple beam model test, the natural frequencies under different pre-stressing conditions are measured. The test data are processed based on the BP neural network technology. The methods and results that using some test data to construct a neural network training samples and distinguish test results with BP network are discussed respectively. The results show that for the cases studied, 80% of the identification results are within an acceptable range of accuracy, which has practical significance for detection of precast pre-stressed components.