Prediction of Compressive Strength of Concrete Using the Spearman and PCA-Based BP Neural Network

Table 5

Comparison of results of prediction of compressive strength of concrete by two BP-based models.

Group number (no.)	Actual value/MPa	The Spearman and PCA optimized BP model		Nonoptimized BP model
		Predicted value (MPa)	Relative error [%]	Predicted value (MPa)	Relative error (%)
21	46.9	47.82	1.96	48.11	2.58
22	48.9	49.22	0.66	48.70	-0.41
23	50.4	50.95	1.08	49.37	-2.04
24	51.5	50.18	-2.57	48.80	-5.24
25	46.6	48.99	5.12	48.53	4.14
26	45.7	47.61	4.19	48.17	5.40
27	45.7	44.21	-3.26	50.69	10.92
28	45.6	49.63	8.84	48.25	5.81
29	46.6	46.62	0.04	52.96	13.65
30	48.7	47.92	-1.60	47.40	-2.67
31	45.1	47.85	6.10	48.05	6.54
32	48.8	49.04	0.49	49.96	2.38
33	49.7	48.17	-3.08	49.58	-0.24
34	46.3	47.49	2.58	46.41	0.24
35	48.2	47.79	-0.86	47.01	-2.47
36	51.0	49.22	-3.49	56.49	10.76
37	51.9	49.98	-3.69	55.25	6.45
38	52.5	52.73	0.43	55.06	4.88
Mean absolute value	48.3	49.93	4.82	48.63	2.78