Fig. 1. Convolutional network structure diagram for fusion of 3D-EEM data and UV-Vis absorption spectral data

Fig. 2. Mask assignment of 3D-EEM and UV-Vis spectra. (a) Mask assignment for 3D-EEM; (b) mask assignment for UV-Vis absorption spectra

Fig. 3. Acquisition of the attention position of the network

Fig. 4. Experimental platform of 3D-EEM and UV-Vis spectra

Fig. 5. Thermal maps of 3D-EEM and corresponding 3D-EEM feature attention matrix. (a) 3D-EEM data; (b) thermal maps of 3D-EEM feature attention matrix

Fig. 6. Attention vector splicing matrix thermal map of UV-Vis spectrum characteristics and UV-Vis spectrum data. (a) Thermal map of attention vector stitching matrix for UV-Vis spectrum characteristics; (b) UV-Vis spectrum data

Fig. 7. Correlation diagrams of COD predicted value and true value. (a) Ten-fold cross-validation results; (b) leave-one-out (LOO) results