Application of the Richards Model for Settlement Prediction Based on a Bidirectional Difference-Weighted Least-Squares Method

Abstract

Predictions of subgrade settlement are important in engineering construction. However, current settlement prediction methods are often unsatisfactory; therefore, the Richards model, which is widely used in the biological field, is introduced in this paper to predict subgrade settlement. To improve the prediction accuracy of soft soil subgrade settlement, the settlement properties of soft soil subgrade and the characteristics of the Richards model are analyzed, and a form of the Richards model based on the bidirectional difference-weighted least-squares (BD-WLS) method is proposed. The performance of the BD-WLS method is verified via comparison with traditional methods of parameter estimation. Moreover, a case study is used to evaluate the predictive effectiveness of the Richards model based on the optimal parameter estimation method. The results show that the Richards model can reduce the mean absolute error by 2.15% on average and improve the prediction accuracy by 22.14% on average compared to other available models. Thus, the Richards model is considered suitable for simulating and predicting the settlement of soft soil subgrade, and it has excellent application potential for settlement prediction.