Study On Property Of AM Construction Method Belled Pile Quality Based On BP Neural Network

The AM construction method belled pile often used as a groundwater-rich areas andpile of soft foundation construction, whole process of rotary drilling reaming.Currently how to determine the AM pile construction method is on of the issues need tobe resolved in the works. Therefore, this paper based understanding the quality of AMpile construction method and mechanism of neural networks, proposed neural networkmodel for quality prediction. Using this method to solve AM pile construction methodaspects of the problem in the quality prediction.In this thesis,by affecting AM engineering methods Piles exist various factorsdetailed and thorough analysis and evaluation,for including soil, pile structure,construction technology, time and spatial effects and other considerations,establishingaffect AM engineering method spiles Quality Factors the index system.Then discussesthe BP neural network method pile construction method in the AM on the issue ofquality prediction feasibility and superiority,on the basis of establishing the pile, soil,construction and other aspects of a total of seven factors as the network input layer,pilevertical bearing capacity,pile integrity as the network output layer BP neural networkquality prediction model.Finally, the collection of AM in Tianjin pile constructionmethod test data processing,using Matlab software neural network toolbox BP neuralnetwork training, and then the test samples pile AM engineering methods to predict theresults of the analysis test network reliability.Through analyzing the training results on the test sample, the measured values withthe site to do comparison, verification BP neural network to solve such problems in thereliability and superiority, as well as the quality of AM pile construction methodprovides a higher prediction accuracy.