Application Of Artificial Neural Network In Settlement Prediction Of Highway Soft Foundation

With the rapid development of economy, the highways have been builtcontinuously in our country. Most of the highways are built in coastland where softclay is distributing. Controlling the settlement is the key problem to build highwayson soft ground. Therefore, in order to construct properly and save the engineeringinvestment, it is important to predict the settlement of soft ground precisely.Recently, the technology of artificial neural network is presented which cansimulate the learning function of human brains and deal with the informationcollaterally. It has the strong nonlinear mapping ability and the good adaptability. Byfar, it has been practiced in many different engineering areas due to its particularmerits.A new method for evaluating settlement of embankment for highways ispresented in this paper, by using of the strong nonlinear mapping and learning abilityof artificial neural network. This method can model highly complex and nonlinearstructures directly based on real samples.Two different modeling methods of neural network are used in the paper fordifferent conditions and requirements in practical engineering. One is using implicitformula to express the connection between factors and settlement. The currentsettlement can be concluded by known factors when predicting. The other is notconsidering the factors of settlement, but establishing the neural network model forthe connection between current settlement and the past. Then the late settlement canbe predicted by highly nonlinear curvefitting.By testing the real data of several engineering and contrasting with sometraditional methods, it is proved that the neural network method can avoid themistakes due to factiousness in traditional methods and can simulate engineeringprecisely, widely and easily. Therefore, the method has a bright future in practicalengineering.