Recognition of weld flaw based on feature fusion of ultrasonic signal and image

Ultrasonic testing is widely applied to detect the inner flaws of materials,but it is still difficult to recognize the flaw properties.In this paper,a new method for flaw recognition based on feature fusion of ultrasonic signal and image was proposed.The detection data was used to identify the weld flaw by the data fusion of ultrasonic signal feature and morphological feature.The welds containing defects such as hole,slag,crack,lack of penetration and lack of fusion were inspected with the manual ultrasonic testing system.Then the ultrasonic signal features of flaw echo and morphological features of flaw image were extracted respectively.Finally,BP neural network was used to carry out the data fusion of these features.The results show that the multi-class flaws could be identified effectively,and the recognition rate of weld flaws was improved by this method.