Applying Adaptive Neural Fuzzy Inference System to Improve Concrete Strength Estimation in Ultrasonic Pulse Velocity Tests

Table 2

Artificial neural network model training result.

One-input (single entry)					Two-input (double entry)
Network type	Hidden layer	Number of training	R^2	MAPE (%)	Network type	Hidden layer	Number of training	R^2	MAPE (%)
1-1-1	1	200	0.73	14.79	2-1-1	1	200	0.73	12.85
1-1-1	1	500	0.74	11.37	2-1-1	1	500	0.74	10.63
1-1-1	1	1000	0.74	14.52	2-1-1	1	1000	0.74	12.43
1-3-1	1	2000	0.73	10.48	2-3-1	1	2000	0.75	10.46
1-5-1	1	2000	0.75	10.98	2-2-1	1	2000	0.74	10.58
1-1-1	1	2000	0.74	10.93	2-7-1	1	2000	0.73	10.96
1-1-1	1	3000	0.74	12.91	2-1-1	1	3000	0.74	10.61
1-1-1	1	5000	0.74	13.49	2-1-1	1	5000	0.73	14.79
1-1-1	1	7500	0.74	13.08	2-1-1	1	7500	0.74	12.84
1-1-1	1	15000	0.74	14.31	2-1-1	1	15000	0.75	13.23
1-1-1	1	20000	0.74	11.45	2-1-1	1	20000	0.74	11.01