Conjugate Cellular Automata and Neural Network Approach: Failure Load Prediction of Masonry Panels

Table 7

Comparison between proposed NN and GSED methods (two panels (1173 and 1172) are used in the GSED method as the base on which the prediction is built).

Panel		Failure load (kN/m^2)		Error (%)
	Expt.	GSED (1173 based)	NN	GSED (1173 based)	NN
1172	20.9	20.01	20.372	4.26	2.53
1169	9.5	12.53	10.059	31.89	5.89
1244	8	6.42	8.112	19.75	1.40
1150	10.7	10.36	10.727	2.9	0.26
1153	6.7	7.03	5.973	4.93	10.86
1237	6.5	5.3	6.406	18.46	1.44
	Expt.	GSED (1172 based)	NN	GSED (1172 based)	NN
1169	9.5	13.6	10.059	43.16	5.89
1244	8	7.13	8.112	10.88	1.40
1150	10.7	11.23	10.727	4.95	0.26
1153	6.7	7.73	5.973	15.37	10.86
1237	6.5	5.9	6.406	9.23	1.44