Application Research On Artificial Neural Network In Pile Foundation Engineering Of High-rise Building

Because of their strong comprehensive and involving many uncertain factors, it is difficult tosolve the problems in pile engineering by the traditional regression analysis, which is one of themathematical programming methods, on the foundation of building determined mathematical model.The artificial neural network (ANN) based on the theory of the knowledge discovery in databaseand intelligent decision system have the ability of processing uncertain information, the strongproperty of learning and the function of nonlinear relationship mapping and prediction, which issuitable to solve the problems that are difficult to be solved by a determined mathematical modeland their training samples are easily collected. At present, the application researches of the theory ofthe artificial intelligent assistant decision system on pile foundation engineering in high rise buildinghome and abroad are less, engineering examples and experiment data are collected and trainingexample sets are set up in this paper, and decision-making of pile foundation type and determinationof ultimate bearing capacity of single pile are researched by ANN. In addition, the primarygeotechnical soil behavior parameters, which effect the pile shaft resistance and pile tip resistance,are used for reference in this paper. These parameters are used for study on ultimate bearing capacityof pile foundation by domestic experts and scholars in geotechnical engineering field. These factorsare used as the main influencing factors in predicting ultimate bearing capacity. But ultimate tipresistance of pile end soil and rigidity of pile body are not considered in process of predictingbearing capacity.The related theories of pile foundation home and abroad and application research results ofANN on decision-making of pile foundation type and determination of ultimate bearing capacity ofsingle pile are used for reference, and the training example sets of ANN used for pile foundationselection are set up on foundation of analysis of the respective influencing factors of pile foundationtype through collecting and arranging plenty of built high rise building examples. The ANN modelused for predicting pile foundation selection and ultimate bearing capacity of single pile isestablished after finishing optimization of neural network’s number of the hidden layer node, transferfunction and training function. When ultimate bearing capacity of super long pile is determined inthis paper, the primary geotechnical soil behavior parameters, which effect the pile shaft resistanceand pile tip resistance, are considered, and the ultimate tip resistance of pile end soil and rigidity ofpile body are also taken into consideration in process of predicting bearing capacity. In this basis, themodel of ANN used for predicting ultimate bearing capacity of super long pile is established.The results of prediction show that the predictive accuracy of the prediction model of ANN ishigh and the prediction error is small, which can meet the need of practical projects. The prediction results has guiding significance in pile foundation selection in high rise building and designingultimate bearing capacity of single pile, and laid a foundation for pile foundation reasonableselection and further study of ultimate bearing capacity of single pile. The model is able to providevaluable reference data for geological prospecting and design department, and guide the design ofultimate bearing capacity of super long pile.