Application of BP Neural Network Fusion Forecast for Deep Excavation Deformation Monitoring
YU Chun-ping · 2008
This paper proposed a method based on Error Back Propagation Neural Network to syncretize the multi-sensor deformation data to precise forecast the deformation,and to obtain the observed value's precise weight through the Error Back Propagation Neural Network's train to avoid subjective influence.This method could precisely find the broken sensors by across forecast of multi-sensor's data,to make sure the data continuity.It could also use for longitudinal forecast.In the deep excavation monitoring engineering in Guangzhou,the ordinal forecast result of multi-sensor's data was consistent with the real deformation,achieved good effect.