An Adaptive Genetic Algorithm Optimizes Double-Hidden Layer BPNN for Rapid Detection of Moisture Content of Green Tea in Processing

Table 5

The results of PLSR and 1d-BPNN model building with original spectra and different preprocessed spectra.

Model	Pretreatment	Calibration model		Predictive model
			RMSECV		RMSEP
PLSR	None	0.959	2.36	0.959	2.53
	MSC	0.961	2.35	0.949	2.83
	SNV	0.956	2.45	0.957	2.60
	SG	0.961	2.30	0.959	2.53
	DT	0.966	2.15	0.961	2.47
	MSC-DT	0.961	2.33	0.958	2.58
	SNV + DT	0.962	2.31	0.958	2.56
	SG-DT	0.963	2.25	0.962	2.43
1d-BPNN	None	0.961	2.34	0.960	2.47
	MSC	0.951	2.63	0.957	2.60
	SNV	0.951	2.65	0.954	2.75
	SG	0.958	2.42	0.959	2.51
	DT	0.967	2.15	0.965	2.31
	MSC-DT	0.974	1.83	0.961	2.43
	SNV + DT	0.976	1.79	0.965	2.27
	SG-DT	0.977	1.77	0.966	2.26