Settlement Prediction of Model Piles Embedded in Sandy Soil Using the Levenberg–Marquardt (LM) Training Algorithm

Abstract

This investigation aimed to examine the load carrying capacity of model piles embedded in sand soil and to develop a predictive model to simulate pile settlement using a new artificial neural network (ANN) approach. A series of experimental pile load tests were carried out on model concrete piles, comprised of three piles with slenderness ratios of 12, 17 and 25. This was to provide an initial dataset to establish the ANN model, in attempt at making current, in situ pile-load test methods unnecessary. Evolutionary Levenberg-Marquardt (LM) MATLAB algorithms, enhanced by T-tests and F-tests, were developed and applied in this process. The model piles were embedded in a calibration chamber in three densities of sand; loose, medium and dense. According to the statistical analysis and the relative importance study, pile lengths, applied load, pile flexural rigidity, pile aspects ratio, and sand-pile friction angle were found to play a key role in pile settlement at different contribution levels, following the order: P > δ > lc/d > lc > EA. The results revealed that the optimum model of the LM training algorithm can be used to characterize pile settlement with good degree of accuracy. There was also close agreement between the experimental and predicted data with a root mean square error, (RMSE) and correlation coefficient (R) of 0.0025192 and 0.988, respectively.

Research on the Prediction Method of Centrifugal Pump Performance Based on a Double Hidden Layer BP Neural Network

“…Here, ∆x t , H e , and g t represent the updated value of the weight, Hessian matrix, and current gradient, respectively. H e and g t can be obtained from Equations (11) and (12), where J and r(x) represent the Jacobian matrix and the error, respectively:…”

“…In recent years, the LM algorithm has been favored by domestic and overseas scholars [9][10][11][12]. It is a classical nonlinear numerical optimization algorithm that combines the advantages of the gradient descent method and the Gauss-Newton method.…”

Artificial neural network (ANN) approach for modelling of pile settlement of open-ended steel piles subjected to compression load

“…The feasibility of the LM training algorithm has recently been underlined as an efficient prediction tool in many engineering sectors and is now gaining growing attention in engineering research (Deo and Şahin, 2015;Juncai et al, 2015;Jebur et al, 2018b). These technical papers have reported the outstanding performance of the LM algorithm over the classical artificial neural networks methods.…”

“…Unlike conventional training methods, the trained LM has several distinctive merits in that it is self-tuning training (doesn't require user dependent parameters after each application), it is 10 to 100 times faster without being trapped in local minima, less vulnerable to overfitting issues, has the ability to determine the optimum solution during the learning process, and it is extremely recommended as the first choice of supervised algorithm. (Abdellatif, 2013;Alrashydah and Abo-Qudais, 2018;Jebur et al, 2018b).…”

Field and numerical analysis on effect of axial load on the lateral response of a pile

“…It was reported that for a pile having length to diameter ratio 40 and constructed in dry dense sand, the bending moment and deformation increase when input motion is severe. Jebur et al (2018) develop a predictive model to determine pile settlement using a novel artificial intelligence method considering Levenberg-Marquardt MATLAB algorithms. Based on the statistical analysis, they found that pile length, applied load, pile flexural rigidity, pile aspect ratio and sand-pile friction angle have significant influence in pile settlement.…”