An ANN Based Optimization Algorithm for Diffracted Laser Beam Shaping

Li, P.Z.; Zheng, Y.B., and Luo, L., 2020. An ANN based optimization algorithm for diffracted laser beam shaping. In: Guido Aldana, P.A. and Kantamaneni, K. (eds.), Advances in Water Resources, Coastal Management, and Marine Science Technology. Journal of Coastal Research, Special Issue No. 104, pp. 255–260. Coconut Creek (Florida), ISSN 0749-0208.To overcome the lack of flexibility in laser beam shaping in current industrial applications, a new improved artificial neural network algorithm for diffracted laser beam shaping is proposed. Aiming at the existing problems in laser beam shaping, the Unet neural network algorithm is improved from the label image and convolution operation. By clarifying its training and application steps, the improved neural network algorithm is pre-trained firstly and then formally trained (full training). The result shows that the UNet neural network algorithm can gradually realize the laser beam shaping with the spatial light modulator and find the mapping relationship between the input image (phase diagram) and the output image (laser contour diagram).