基于零蠕变特征空间投影的P91钢焊接接头蠕变超声检测

为提高P91钢焊接接头蠕变超声检测灵敏度,提出一种相对性的超声检测信号处理方法。该方法使用未发生蠕变的P91钢焊接接头的检测信号生成零蠕变特征空间,将任意检测信号向该空间投影,以投影残差能量与原信号总能量的比例表征检测目标偏离零蠕变状态的程度,比例越高,则表明蠕变损伤越严重。推导了该方法的一般性数学表达,并提出利用盲源分离算法实现该方法的线性化处理。利用不同蠕变程度的P91钢焊接接头试块对该方法进行了验证试验,结果表明,与常规超声方法相比,该方法具有更好的蠕变检测能力。由于该方法为相对性检测,零蠕变特征空间的完备性对检测结果的准确性具有较大影响。该方法以评价检测目标偏离标准特征空间的程度为核心,因此对其他损伤也具有检测能力。

Ultrasonic Inspection for Creep in P91 Steel Weldments Based on Non-creep Feature Space Projection

In order to improve the detection sensitivity of ultrasonic inspection for creep in P91 weldments, a relative inspecting signal processing method is proposed. To apply this method, firstly a non-creep feature space should be generated from all the ultrasonic inspecting signals acquired from no creep specimens; then a residual vector can be calculated by projecting inspecting signal to the non-creep feature space; lastly the deviation ratio can be calculated by dividing the residual vector energy by signal energy. The deviation ratio reflects the distance between inspecting signal and non-creep feature space, higher deviation ratio means worse creep damage in the inspecting target. A general mathematic expression of the non-creep feature space projection method is deduced. For the convenience of signal processing, a simplified linearization algorithm based on blind source separation theory is suggested. To verify this new method, a testing experiment is conducted with P91 steel weldment specimens of different creep status. The results showed that, the new method provides higher sensitivity than usual ultrasonic method. As this method is a relative one, the accuracy of creep evaluation by the method depends on the completeness of non-creep feature space. In addition, as this method is based on evaluating the distance between inspecting target and no-defect feature space, it also can be used for detecting other types of defects.