摘要
针对压力容器声发射检测的实际情况 ,选择 5个主要的声发射信号特征参量为研究对象 ,利用人工神经网络 (ANN)信号处理技术 ,对声发射信号进行有效性识别的理论和实验研究 .数据处理结果表明 ,采用改进的多层前向误差反传网络算法和编制的程序 ,可以显著提高声发射检测数据中有效信号的处理速度和识别率 .对声发射实验数据进行有效性分析 ,共得有效数据个数为 14 5 4个 ,近似误判率为 0 .8%
In view of practical experience of AE inspection of pressure vessels, we choose 5 AE parameters as subjects, using artificial neural networks (ANN) signal processing technology to study the theory and experiment of recognizing the availability of AE signal. The result of date processing indicats that using the improved ANN model of Back Propagation (BP) and the program written in this method can obviously raise the rate of signal distinguishing available from AE data. Effective analysis of AE data is carred out, and 1 454 events are obtained. The result shows error rate is 0.8%.
出处
《大庆石油学院学报》
CAS
北大核心
2001年第1期63-66,共4页
Journal of Daqing Petroleum Institute
基金
中国石油天然气集团公司"九五"重点科技攻关项目 !(97611-1)
