Research Article

Damage Identification for Large Span Structure Based on Multiscale Inputs to Artificial Neural Networks

Table 15

The identification errors of neural networks based on strain damage parameters.

Training functionsSetsNeuronsNoise levelsMaximum error
0.000.010.020.040.080.10

T_LM192 (13,6)0.0570.1950.2640.1800.1560.1640.264
T_LM192(15,6)0.1180.3650.2020.1820.2310.2300.365
T_LM192(15,7)0.0640.2560.1850.1720.7140.1170.714
T_LM192(17,8)0.0660.3460.2520.1530.1720.1980.346
T_LM240(13,6)0.0400.2020.1510.1860.1610.1620.202
T_LM240(15,6)0.0680.1560.1510.1340.2790.1270.279
T_LM240(15,7)0.0460.1820.1350.9250.1720.1500.925
T_LM240(17,8)0.0770.3430.1940.2460.1650.1570.343
T_LM264(13,6)0.0860.4000.1820.5090.2280.1640.509
T_LM264(15,6)0.2390.3800.1740.7500.1490.1600.750
T_LM264(15,7)0.0820.1640.1760.1510.1970.1920.197
T_LM264(17,8)0.0410.2450.2220.1630.3090.1720.309
T_LM288(13,6)0.0440.1360.3900.1450.1120.1650.390
T_LM288(15,6)0.1440.1570.1960.1420.1220.1490.196
T_LM288(15,7)0.0670.2420.2240.1780.4040.2570.404
T_LM288(17,8)0.0950.1580.2680.1090.1280.1410.268
T_LM336(13,6)0.0820.1110.1250.1030.2550.1140.255
T_LM336(15,6)0.0390.1490.2450.1260.2730.1930.273
T_LM336(15,7)0.0440.1690.1300.0810.1280.4840.484
T_LM336(17,8)0.0400.3220.1960.1490.0670.0550.322