Estimating Ultimate Load Carrying Capacity of Shell Foundation: Neural Network model and Sensitivity Analysis

Abstract

The use of computational intelligence techniques is becoming popular across several disciplines. One of the key criteria for examining an existing algorithm is to apply it on different data sets. This paper is an application oriented work that uses a computational intelligence technique to analyze an important civil engineering problem which evaluates the suitability of neural network in estimating the ultimate load of shell foundations. In addition, to understand the relative importance of input parameters, sensitivity analysis using various methods are presented. Neural interpretation diagrams are drawn to know the relation between inputs and the output. An empirical equation developed using the connection weight and biases of the trained ANN model with reasonable accuracy.