An effective parameter estimation on thermoelectric devices for power generation based on multiverse optimization algorithm
Luis Fernando Grisales-Noreña, Vanessa Botero-Gómez, Rubén Iván Bolaños, Faustino Moreno-Gamboa, Daniel Sanín-Villa
Abstract
This study explores using the Multiverse Optimization Algorithm to estimate the material properties of thermoelectric generators used for power generation . The modeling of these modules is complex due to nonlinear differential equations and the challenge of determining the actual values of thermoelectric properties like electrical resistivity, thermal conductivity , and one of the most relevant parameters, the Seebeck coefficient . The study compares the performance of the Multiverse Optimization Algorithm with six other metaheuristic techniques reported in the literature for estimating these parameters. The parameters were estimated by conducting experimental power measurements and comparing the results with the real ones predicted by the equations model using the root mean square error . The Multiverse Optimization Algorithm demonstrates superior performance compared to other algorithms presented in previous works for this case. It achieves a minimum root mean square error of 0.001803 and a mean root mean square error of 0.001874 with a standard deviation equal to 3.013% over 1000 simulations. These results improve over those obtained by the Ant Lion Optimizer, which reported a minimum root mean square error of 0.001804 and a root mean square error of 0.001896 on average with a standard deviation of 3.788%. Additionally, the study highlights the efficiency of the Multiverse Optimization Algorithm in processing time, with an average execution time of 223.65 seconds. The statistical analysis was conducted using box-and-whisker figures to assess the potential of the new method. This evaluation considered various metrics, including the best solution of each algorithm, the mean solution, the standard deviation, and the time consumed on average on each run.