Fig. 1. Training process of BP neural network

Fig. 2. BP neural network model

Fig. 3. Schematic diagram of simplified mesh technique

Fig. 4. Phase distribution and intensity distribution. (a) Phase distribution of DOE; (b) output intensity distribution on target plane

Fig. 5. Training process and results of BP neural network. (a) Error distribution of prediction results; (b) error descent process; (c) test chart of data gradient and learning times; (d) residual normal test

Fig. 6. Prediction of 20 groups of random parameters in range of training samples

Fig. 7. Light intensity diagrams of analog output of random 4 groups of data (parameters in brackets are values of beam waist radius, beam outer radius, diffraction distance, wavelength, and target surface size). (a) (0.8, 0.92, 310, 633, 2.4); (b) (0.8, 0.95, 300, 633, 2); (c) (0.8, 1, 250, 633, 2.5); (d) (0.8, 1.05, 330, 633, 1.6)

Fig. 8. Uniformity of analog output of random 4 groups of data

Fig. 9. Schematic diagram of expansion of training range

Fig. 10. Range expansion of all parameters. (a) Forward expansion; (b) reverse expansion

Fig. 11. Results of range expansion of single parameter. (a) Range expansion of plane distance f; (b) range expansion of waist external radius R; (c) range expansion of spot size S

Fig. 12. Intensity distributions of target plane at different positions