Study On Prediction Of Ram-compacted Piles With Bearing Base Quality Based On BP Neural Network

Ram-compacted piles with bearing base are new piles and widely used in pile foundation, how to get the single pile's quality is one of the most important problems at present. It is difficult to detect the actual quality of every pile accurately, conveniently, economically in practice projects for the limitation of empirical formula method and field test method. Therefore, it has important significance for engineering managers to inquiry a new prediction method with scientific and rationality according to practical needs and corresponding theory. Because there are lots of complex influencing factors of ram-compacted piles with bearing base with massive unascertained information, it is difficult to get the result using accurate mathematical formula. However, the BP Neural Network has great learning ability and non-linear massive simultaneous managing capacity; this paper will solve how to forecast quality of Ram-compacted piles with bearing base.Firstly, this paper describes the load bearing mechanism of ram-compacted piles with bearing base, analysis all factors with regard to quality of the single pile from five respects that bearing stratum, pile body, construction machinery, time and spatial effect, meanwhile, establishes prediction mode for bearing capacity characteristic value of the single pile and pile integrity. Secondly, compiles the program with the neural network toolbox of MATLAB, making use of data of test pile on Handan area to train the Network, then compares the predictive result with the results of measured value, verifying the accuracy and reliability. Finally, this paper provides a new prediction method for quality of the single pile of Ram-compacted piles with bearing base.