Parameters Identification Of High Rock-filled Embankment Constitutive Model Based On BP Neural Network And Application Research

With the development of Chinese highway construction toward the mountainous areacomplex geological and terrain condition, the high rock-filled embankment and diggingtunnels have become inevitable. In the actual construction process, high-quality tunnel slagstone is often used as roadbed filler, which can not only solve the gravel packing problem ofthe mountain, but also reduce the damage to the ecological environment and geologicaldisasters along the line, so the high rock-filled embankment will become the types ofembankment structure which is more common, economic, and environmental protection.The settlement of the high rock-filled embankment, particularly the calculation andprediction of the post-construction settlement have become the key to the problem whichdepends on the success of engineering or not, but due to its complexity, usually need tochoose or establish a suitable constitutive model for simulation calculation, however themodel parameters become the "bottleneck" of restricted the problem analysis. Theconventional engineering analogy method and the test method have their defective, peopleturn to study the model theory and method of the parameters which are based on back analysisof field measurements, to keep the existing theory and calculations can be used in engineeringpractice. In this paper, based on the typical cross-section embankment of Hurongxi highway,according to the measured data and the dynamic construction simulation, the stress and strainlaw of the embankment was revealed, and creep model of three parameters was established.Finally, the method that finite element was combined with the BP neural network,which canidentify parameters of the elastic-plastic model, the Duncan-Zhang (EB) model and the creepmodel, and the identified parameters were analyzed and calculated. Through the comparisonof the calculated values and the measured values found that they were good agreement andhigher reliability.Therefore, the method that finite element was combined with the BP neural network cannot only more accurate calculate and predict the embankment settlement, but also providescientific decision-making with the construction of information technology, avoid the unevensettlement of embankment and large post-construction settlement disasters, so it is veryimportant significance to maintain the smooth flow of traffic, protect the vehicle andpedestrian safety, promote the development of the national economy.