Damage Identification of a Steel Frame Based on Integration of Time Series and Neural Network under Varying Temperatures

Table 11

The identification results of damage severities when two braces are damaged and contaminated with noises.

Damaged braces	aj (%)	T (°C)	Target output		Noise intensity (1%)				Noise intensity (3%)
					Actual output		Relative error		Actual output		Relative error
1, 4	50	20	0.5	0.5	0.50956	0.50752	1.91	1.50	0.52651	0.52599	5.30	5.20
		30	0.5	0.5	0.5244	0.52215	4.88	4.43	0.54715	0.54677	9.43	9.35
	70	20	0.7	0.7	0.71206	0.71533	1.72	2.19	0.76501	0.76407	9.29	9.15
		30	0.7	0.7	0.74194	0.74332	5.99	6.19	0.79502	0.79349	13.57	13.36
1, 5	50	20	0.5	0.5	0.41313	0.41148	17.37	17.70	0.47940	0.47879	4.12	4.24
		30	0.5	0.5	0.45802	0.45692	8.40	8.62	0.49173	0.49111	1.65	1.78
	70	20	0.7	0.7	0.61038	0.61457	12.80	12.20	0.5861	0.58614	16.27	16.26
		30	0.7	0.7	0.73215	0.73307	4.59	4.72	0.62806	0.62811	10.28	10.27
4, 5	50	20	0.5	0.5	0.39151	0.3929	21.70	21.42	0.45318	0.45306	9.36	9.39
		30	0.5	0.5	0.45248	0.4539	9.50	9.22	0.43403	0.43389	13.19	13.22
	70	20	0.7	0.7	0.6481	0.64835	7.41	7.38	0.72292	0.72244	3.28	3.21
		30	0.7	0.7	0.68631	0.68668	1.96	1.90	0.72519	0.72474	3.60	3.53