Searching for soil models’ parameters using metaheuristics

Abstract

Grounding systems are an important part of protection systems which protect people and devices in case of defects in electro energetic systems or lightning strikes. The Finite Element Method (FEM) is often used for proper dimensioning of the grounding systems. Often data about the soil in the surroundings of the grounding system are obtained using measurements. Soil parameters can be determined using analytical soil models, and the determination of the soil models’ parameters, which are based on the measured data, is an optimization problem. In this paper, different soil models are tested on different measured data and compared with each other. Different metaheuristics are used and tested for the determination of soil parameters: A Genetic Algorithm, Differential Evolution with two different strategies, Teaching-Learning Based Optimization, Artificial Bee Colony and Dynamic Optimization using Self-Adaptive Differential Evolution. Based on the test results, we improved the most appropriate method. As a result, the most appropriate soil model among those tested is selected, and a method for parameter determination is presented which combines Artificial Bee Colony and Teaching-Learning Based Optimization. The presented solution is appropriate to be used with, or as a part of, FEM calculation software.