Modeling and Multiresponse Optimization for Anaerobic Codigestion of Oil Refinery Wastewater and Chicken Manure by Using Artificial Neural Network and the Taguchi Method

Table 6

Selection of neurons in the hidden layer.

Number of hidden neurons	Mean square error (MSE)			Correlation coefficient
	Training	Validation	Test	Training	Validation	Test	All
5	315.8403	333.8052	1726.2233	0.9384	0.9589	0.8045	0.9052
10	509.7489	597.2636	871.1606	0.9209	0.9119	0.8124	0.8986
15	366.1776	390.7364	991.7434	0.9317	0.8889	0.8948	0.9169
20	166.2242	882.3070	1253.0458	0.9712	0.9020	0.7213	0.9227
25	181.0968	499.4849	2919.6910	0.9709	0.9287	0.5398	0.8883
30	66.9453	1253.3547	6952.3913	0.9896	0.6727	0.5548	0.8139
50	17.9675	14269.6345	6632.0505	0.9962	0.5130	0.3574	0.6443
16	247.0099	1088.4714	944.7495	0.9557	0.8219	0.8931	0.9175
17	314.0407	519.0463	1295.0456	0.9355	0.9040	0.8693	0.9129
18	187.2529	626.7710	1326.1371	0.9615	0.9039	0.8508	0.9243
19	202.2901	639.2647	919.5157	0.9667	0.9166	0.8661	0.9348