Probabilistic analysis of gravity retaining wall using ANFIS-based optimization techniques
Rashid Mustafa,Pijush Samui,Sunita Kumari
DOI: https://doi.org/10.1007/s41062-022-00962-1
2022-10-23
Innovative Infrastructure Solutions
Abstract:The destruction of gravity retaining wall affected by inadequate design and improper construction is the most common failure in the field of geotechnical engineering. Based on hybrid adaptive neuro-fuzzy inference system (ANFIS), probabilistic analysis of gravity retaining wall was established. Novel hybrid approach is proposed for predicting of stability of gravity retaining wall. The proposed approach is established on the blend of computational model like adaptive neuro-fuzzy inference system, and meta-heuristic optimization techniques like particle swarm optimization (PSO), Genetic algorithm (GA), Firefly algorithm (FFA), Bio-geography-based optimization (BBO) and Grey wolf optimization (GWO) are used. In this approach, the PSO, GA, FFA, BBO and GWO are used to improve the performance of ANFIS. The models are trained and tested based on 100 random numbers generated for input parameter like cohesion (c), unit weight of soil (γ), angle of shearing resistance (φ) and friction between soil and wall (µ) to compute factor of safety against sliding. To determine the accuracy of the developed models, various statistical parameters like coefficient of determination (R2), root mean square error (RMSE), a-20 index, Kling Gupta efficiency (KGE), etc., are computed for training and testing data for the described model.