Previsão do estado de carga de bateria aplicada a veículos elétricos usando modelos lineares auto-ajustados através de algoritmos de otimização

Abstract

In the last decades, electric vehicles have gained great popularity by their performance and efficiency. Investment in the development of this new technology is justified by increased consciousness of the environmental impacts caused by combustion vehicles such as greenhouse gas emissions that have contributed to global warming as well as the depletion of non-oil renewable energy source. Lithium-ion batteries are the most promising have been widely used for their advantageous features, forecasts such as high energy density, a large number of cycles, and low self-discharge. One of the critical factors for the correct operation of an electric vehicle is the estimation of the battery charge state. This work presents a comparison of the load state estimation (SOC), tested in four different conduction profiles, which was performed using three distinct models: Generalized Linear Model (GLM), GLM with Poisson regression and Linear interpolation (SPLINE) calibrated by 3 different optimization techniques: Genetic Algorithm (GA), Differential Evolution (DE) and Particle Swarm Optimization (PSO). The computational results showed that for the SOC prediction the most suitable model is the GLM calibrated by the PSO optimization algorithm.