A multiparameter surface roughness evaluation model of cold-rolled strips using laser speckle images

Abstract

With the wide use of cold-rolled strips, the conventional contact measurement of strip surface roughness cannot meet the requirements of nondestructiveness, high precision, short detection time, and online inspection. Thus, a multiparameter surface roughness evaluation model for the roughness range of cold rolled strips is proposed to address the drawbacks of contact measurement methods and improve the accuracy, stability and robustness of the existing speckle methods. Strip surfaces are used to develop speckle patterns by a laser beam from a light source. Statistical parameters and texture parameters are computed from speckle images. A correlation between multiple speckle parameters and measured surface roughness is developed by using dimension reduction rules and mapping functions; a fitting function is used to establish the surface roughness evaluation model; and the artificial bee colony algorithm (ABC) is used to optimize the model coefficient between the new speckle parameter and surface roughness. Experimental results confirm that the proposed multiparameter evaluation model in exponential form can obtain surface roughness with the best goodness of fit (0.967) and a smaller error than the single parameter model.