回弹-超声-拔出综合法检测混凝土强度的研究

Research on Repercussion-ultrasonic and Pull-out Concrete Strength Checking Based on Artificial Neural Network

在试验研究的基础上,建立了混凝土强度回弹-超声-拔出综合法检测的人工神经网络模型。探索并应用了神经网络的改进算法,其中包括附加冲量法,自适应学习算法及S型函数输出限幅算法等,以保证建立的神经网络的快速有效。与传统的回归算法相比,采用人工神经网络模型推测出的混凝土强度具有更高的精度。

The artificial-neural-network model of Repercussion-ultrasonic and pull-out is built up based on considerable data from the experiment.In order to make neural network more efficient,there are some optimized BP algorithms being put forward including momentum algorithms,variable learning rat BP algorithms and limited output of s-function algorithms.The artificial-neural-network model is more efficient than regression analysis.