Fuzzy and neural network models for analyses of piles

Piles have been used as a foundation for both inland and offshore structures. The evaluation of the load carrying capacity of a pile, setup, and its drivability are important problems of pile design. First, the best way to estimate the ultimate capacity of pile is a static load test. However, static load tests can not be done routinely because of high cost involved. The most frequently used method of estimating the load capacity of driven piles is to use dynamic driving formula. But these formulae cannot explain time-dependent events, and cannot cover a variety of pile driving systems. Second, an evaluation of ultimate load capacity considering pile setup may lead to more economical pile design leading to reductions in pile lengths, pile sections, and the size of driving equipment. The commonly used relationships do not provide reliable predictions for use in practice because the methods used to estimate pile setup are highly empirical and their predictive abilities are limited by the corresponding data sets from which they were derived. Third, in practice driving criteria are provided to the contractor for the piles to have a required load bearing capacity and to be driven without being overstressed. A wave equation based computer program is used to generate the pile driving criteria for every project. This process takes a significant amount of time and requires the usage of several procedures as PILECAP, GRLWEAP, PDA, and CAPWAP. This current practice requires significant training skills, and can be very time consuming since a good deal of effort is devoted to the analyses.; The essence of modeling is prediction, which is obtained by mapping a set of variables in input space to a set of response variables in output space through a model. The conventional modeling of the underlying systems, often tends to become quite intractable and very difficult. Recently an alternative approach to modeling has emerged under the rubric of 'soft computing' with 'neural network' and 'fuzzy logic' as its main constituents. These are 'observational models' developed on the basis of available sets of data representing a mapping between input and output variables. The general nature of geotechnical problems and the consequent role engineering judgment plays in their treatment, make them ideally amenable to this approach of modeling. The development of these models however requires a set of data. Fortunately, for many problems such data are available. Fuzzy systems and neural networks are both model-free numerical estimators. They share the ability to improve the predictive capability of a system working in uncertain, imprecise, and noisy environments. Fuzzy logic and neural networks are complementary technologies. In order to utilize the strengths of both, fuzzy systems and neural networks may be combined into an integrated system. The integrated system then has the advantages of both neural networks (e.g., learning abilities, optimization abilities, and connectionist structure) and fuzzy systems (e.g., humanlike 'if-then' rules, and ease of incorporating expert knowledge available in linguistic terms).; In this study, Back Propagation Neural Network (BPNN) models and Adaptive Neuro Fuzzy Inference System (ANFIS) models are developed for: (i) Ultimate pile capacity, (ii) Pile setup, and (iii) Pile drivability (blows per foot (BPF), Maximum compressive stress, and Maximum tension stress). A database for ultimate pile capacity and pile setup has been developed from a comprehensive literature review. Predictions for the above are made using BPNNs as well as commonly used empirical methods, and they are also compared with actual measurements for each application. For the pile drivability analysis, a database of a number (3,283) of HP piles is developed from the data on HP piles from 57 projects in North Carolina (with both GRLWEAP data and soil profile information and without PDA and CAPWAP analyses). All of the programs are developed within MATLAB...