Abstract: Visual inspection was an effective method of achieving weld surface quality inspection. However, the black ash generated by the arc welding of aluminum alloy bodies of rail trains caused a large amount of missing data collected by the vision sensors, which affected the accuracy of weld surface quality inspection. To this end, an integrated cleaning-inspection method for aluminum alloy body welds was proposed in the paper, designed the “cleaning-welding-detect” integration device was designed. The laser cleaning technique was used to remove the black ash produced by arc welding, it reduced the missing data rate. The weld surface contour information was collected by line structured light sensor, and the dynamic ideal contour model of weld surface was established by cubic spline interpolation and random sample consensus algorithm. By extracting feature points and comparing the difference between the dynamic ideal contour and the actual contour, the measured weld characteristic dimensions and surface defect detection were realized. The results showed that the missing rate of collected data after laser cleaning was 93.43% lower than that before cleaning. The integrated method of laser cleaning and quality inspection was proposed to realize the measurement of weld characteristic dimensions and the detection of surface defects. The detection accuracy of weld characteristic dimensions reached 0.1 mm, and the detection accuracy of porosity, overlap, lack of fusion was within 0.15 mm.