摘要
将小波母函数嵌入人工神经网络的神经元形成紧致型小波神经网络,将此种网络用于混凝土非破损检测的测强曲线拟合和预测,提升了纯粹的BP神经网络的拟合和预测精度,效果远胜于最小二乘拟合和预测。通过一个算例对小波神经网络的高精准性和非性线逼近能力进行了验证,对实验数据进行了统计分析,结论表明小波神经网络优于BP神经网络。
Inset the small wave generating function into never cell of artificial neural network to form compact small wave neural network. Using this kind of network into strength test curve fitting of no concrete damage testing and forecast, the precision of fitting and forecast of pure BP neural network was improved, the effect is much better than that of least square. Based on a calculate example, the higher precision and ability of no-linear approach of small wave neural network was validated. The experiment data was statistic and analyzed. The result indicate that the small wave neural network is better than BP neural network.
出处
《工程质量》
2011年第12期63-65,68,共4页
Construction Quality
基金
甘肃省教育厅科研资助项目(00330715-01)
关键词
无损检测
测强曲线
小波分析
神经网络
拟合
no damage test
strength test curve
small wave analysis
neural network
fitting