低碳钢氢致开裂声发射信号分析与模式识别

Acoustic Emission Analysis and Pattern Recognition on Hydrogen Induced Cracking of Mild Steel

摘要: 应用关联图的声发射信号分析方法, 分析了20R压力容器用低碳钢氢致开裂在不同阶段的声发射信号特点, 进而利用人工神经网络的非线性映射能力, 利用试验所得样本对氢致开裂进行模式识别, 识别出裂纹产生的不同阶段。试验结果表明, 该方法能够检测常规检测方法难以检测出的氢致裂纹, 模式识别结果具有较高的准确度, 为实际工程检测提供了参考方法。

Abstract: The acoustic emission(AE) test method was used and the characteristics of different phases of hydrogen induced cracking (HIC) in 20R mild steel were analyzed. Artificial neural network was used to recognize the pattern of HIC. Practical test showed the above method could test the cracks that common method could not test. The result of pattern recognition was rather accurate, and provides a reference method for practical engineering inspection.