Predicting Compression Index Using Artificial Neural Networks: A Case Study from Dalian Artificial Island

Compression index is very important in the design of geotechnical engineering such as consolidation settlement prediction and construction design. However, measuring compression index is very complex and time-consuming. In addition, it is very difficult to collect unbroken core samples from underground. Artificial neural network has been adopted in some geotechnical applications and has achieved some success. In this paper, artificial neural network (ANN) models are developed for estimating compression index by basic soil parameters based on 2859 soil test data. All of the marine soil samples, which are divided into three subsets to train the optimum model, are collected from Dalian Artificial Island and compression index and other parameters are measured in soil mechanical laboratory as well. At last, the optimized ANN model structure and suitable inputs are determined followed by the comparison between empirical formulas predictions and ANN models output. It is revealed that ANN models perform better than empirical formulas with respect to the accuracy of compression index prediction.