Back analysis of the thermal parameters of an multiple arch dam using BP neural network algorithm

The backpropagation (BP) neural network algorithm was applied to calibrate thermal parameters of a multiple arch concrete dam. Samples were obtained from forward analysis using the three dimensionalfinite element method. The obtained samples were then used to train the network, and the measuredtemperature and thermal input parameters were applied to back analysis of thermal parameters. Different temperature fields were calculated and analysed by achieved network, giving different thermal parameters as input. Best input parameters were selected according to the least root mean square error (RMSE), Comparing network computational results with the thermometer's measured values. Results indicated acceptable conformity error and good compliance with engineering requirements. Furthermore, the improved BP neural network algorithm has fine stability and convergence. The resultant network could be used as a substitute of the finite element method to estimate temperature fields and also a powerful tool for back analysis of dam thermal parameters.

کلیدواژه ها:

improved BP neural network algorithm ، three dimensional finite element method ، temperature field ، back analysis ، thermal parameters