A novel back-analysis method for stope displacements using gradient boosted regression tree and firefly algorithm

New paper published on Journal of Computing in Civil Engineering

Paper Title: A novel back-analysis method for stope displacements using gradient boosted regression tree and firefly algorithm

Authors: Chong-chong Qi*, Andy Fourie & Xu Zhao

Abstract: It is essential to determine the properties of the rock mass surrounding underground excavations to facilitate stability analysis and engineering design. In this paper, a novel displacement back-analysis method is proposed based on gradient boosted regression tree (GBRT) and firefly algorithm (FA). The proposed method, the GBRT-FA, utilized GBRT as an instance-based learning approach to substitute numerical modelling. Furthermore, FA was used for the hyper-parameters tuning and the rock mass properties searching. The input variables in the numerical modelling were chosen to be deformation modulus, Poisson’s ratio, cohesion and internal friction angle, which were back-analysed using the GBRT-FA. A total of 13,310 numerical models were conducted to provide the dataset for the training and testing of GBRT models. A parametric study of back-analysis performance was also conducted. The results show that FA was efficient in the hyper-parameters tuning of GBRT with stabilized results being obtained within six iterations. The average median absolute percentage error (APE) between displacement values from numerical modelling and the optimum GBRT model was 5.4%, denoting that numerical modelling could be well substituted by the optimum GBRT model. The overall performance of the GBRT-FA was reasonably good, with the average APE value for all input variables being 6.3%. The substitution performance of GBRT models, the dataset size, and the number of displacement measurements were found to have a significant influence on the performance of the displacement back-analysis method. Suggestions for the engineering applications of back-analysis methods were made based on the results, which have a guiding significance for underground mines.

Keywords: Mining stopes; Displacement back-analysis; Gradient boosted regression tree; Firefly algorithm; Numerical modelling.

Comments from reviews:

Reviewer 1: This paper is reported a novel method to do back-analysis using gradient boosted regression tree (GBRT), firefly algorithm (FA). The capability of the proposed method has been validated and a parametric study was performed. This topic is interesting and within the scope of Journal of Computing in Civil Engineering. The article is well written with figures being professionally drawn, which will contribute to the existing literature. Therefore, I would like to recommend to publishing this paper subjected to minor revisions as requested below.

Reviewer 2: The study proposed a back-analysis method for stope displacements using gradient boosted regression tree (GBRT) and firefly algorithm. GBRT is a relatively novel machine learning technique for classification and regression. It has not been founded to be adopted in engineering inverse analysis problems. Therefore this paper makes some progress on GBRT based system identification. The following comments are presented to improve the paper.

Representative figure from this paper:

Back-analysis performance on the testing set.

Full paper can be obtained through Email.