A new auto-tuning model for predicting the rock fragmentation: a cat swarm optimization algorithm

The main focus of the present work is to offer an auto-tuning model, called cat swarm optimization (CSO), to predict rock fragmentation. This population-based method has a stochastic formation involving exploration and exploitation phases. CSO is a robust and powerful meta-heuristic algorithm inspired by the behaviors of cats; it is composed of two search modes: seeking and tracing, which can be joined by mixture ratio parameter. CSO is applied to large-scale optimization problems like rock fragmentation to have good forecasting parameters in D-80 formulas (D-80 is a common descriptor that evaluates rock fragmentation). To evaluate the efficiency of the proposed CSO model, its obtained results were compared to those of the particle swarm optimization (PSO) algorithm. In the modeling, two forms of CSO and PSO models, including power and linear forms, were developed. The comparative results showed that CSO models outperformed the rival in terms of the task defined. The precision of the proposed models was computed using statistical evaluation criteria. Comparison results concluded that CSO-power model with the root mean square error (RMSE) of 0.847 was more computationally efficient with better predictive ability compared to CSO-linear, PSO-linear and PSO-power models with the RMSE of 1.314, 1.545 and 2.307, respectively. Furthermore, the sensitivity analysis revealed the effect of the stemming parameter upon D-80 in comparison with other input parameters.

Automated summary

The main objective of this study is to provide an auto-tuning model called cat swarm optimization (CSO) for predicting rock fragmentation. The results of the study show that the CSO model outperforms the particle swarm optimization (PSO) algorithm in terms of its predictive ability in the D-80 formula.