Employing Artificial Neural Networks to Predict the Performance of Domestic Sewage Treatment Terminals in the Rural Region

Table 6

Error analysis about ANN, ANN-GA, and ANN-PSO.

Model	Input parameters	Predicted parameters	R ^2	RMSE	MAPE	NSEC
ANN	Influent pH,; effluent pH,; influent conductivity,; effluent conductivity,; influent turbidity,; effluent turbidity,; influent NH3-N,; effluent NH3-N.	Effluent TN	0.84 for training;	11.40 for training	18.99% for training;	0.95 for training;
			0.85 for validation	13.49 for validation	19.93% for validation	0.95 for validation
		Effluent COD	0.81 for training;	30.88 for training;	45.42% for training;	0.80 for training;
			0.82 for validation	30.00 for validation	24.54% for validation	0.80 for validation
ANN-GA		Effluent TN	0.90 for training;	9.42 for training;	14.90% for training;	0.97 for training;
			0.84 for validation	13.56 for validation	20.09% for validation	0.95 for validation
		Effluent COD	0.76 for training;	34.31 for training;	53.92% for training;	0.76 for training;
			0.85 for validation	26.25 for validation	26.91% for validation	0.84 for validation
ANN-PSO		Effluent TN	0.90 for training;	9.14 for training;	16.19% for training;	0.97 for training;
			0.90 for validation	11.54 for validation	16.79% for validation	0.97 for validation
		Effluent COD	0.90 for training;	22.10 for training;	34.57% for training;	0.90 for training;
			0.85 for validation	26.57 for validation	22.30% for validation	0.84 for validation