A novel TS Fuzzy-GMDH model optimized by PSO to determine the deformation values of rock material

Since determining the rock deformation directly in the laboratory is costly and time consuming, it is important to reliably determine/estimate this parameter through the use of several simple rock index tests. This study develops a new hybrid intelligent technique according to Takagi-Sugeno Fuzzy Inference System-Group Method of Data Handling optimized by the particle swarm optimization, called TS Fuzzy-GMDH-PSO for prediction of the rock deformation. The PSO role in this advanced system is to optimize the membership functions of TS Fuzzy-GMDH model for enhancing the level of prediction capacity. In this research, four rock index tests including Schmidt hammer, p-wave velocity, porosity and point load were selected and conducted in laboratory in order to establish a suitable database for prediction purposes. To demonstrate the feasibility and applicability of the advanced hybrid model, two base models of TS Fuzzy and GMDH were also modeled to forecast rock deformation. After conducting several sensitivity analyses on the mentioned models to get the highest performance capacity, their prediction levels were evaluated using some statistical indices, such as root mean square error and correlation coefficient (R). The comparative results confirmed the superiority of the TS Fuzzy-GMDH-PSO over other two models, namely TS Fuzzy and GMDH in terms of both train and test phases. It can be concluded that the TS Fuzzy-GMDH-PSO can be recommended as a powerful, capable and new model to solve the problems related to rock strength and deformation.