Fig. 1. Structure of WNN network

Fig. 2. Flow chart of GA algorithm

Fig. 3. Diagram of experimental set-up

Fig. 4. Experimental samples
(a): Horse blood; (b): Cow blood; (c): Rabit blood;(d): Props fake blood; (e): Red ink

Fig. 5. Time-resolved photoacousic signals of different blood samples at 740 nm

Fig. 6. Photoacoustic peak-to-peak mean spectral of five different blood samples at 700~1 064 nm

Fig. 7. Effects of parameters on the classification and identification of real or fake blood for WNN and WNN-GA
(a): Effect of neuron numbers in hidden layer; (b): Effect of learning factor η; (c): Effect of learning factor γ; (d): Effect of iteration number

Fig. 8. Results of classification and identification of real or fake blood for WNN and WNN-GA
(a): Effect of training time on the mean square error;(b): Results of classification and identification of test bloods

Fig. 9. Effects of parameters on the classification and identification of real or fake blood gor PCA-WNN-GA
(a): Effect of neuron numbers in hidden layer; (b): Effect of learning factor η; (c): Effect of learning factor γ; (d): Effect of iteration number

Fig. 10. Results of classification and identification of real or fake blood for PCA-WNN-GA under different principle components
(a): Effect of training time on the mean square error;(b): Results of classification and identification of test bloods

Fig. 11. Correction rate of classification and identification of real or fake bloods for seven different algorithms