Modelo de degradación de batería de li-ion para aplicaciones en microrredes eléctricas

Abstract

Non-conventional sources of energy, such as photovoltaic energy and wind energy, constitute the beginning of the transformation of the traditional energy system, moving from centralized generation away from consumption centers, to a system with distributed generation, which is closer to final loads, and a more diverse energy matrix that reduces dependence on a few energy sources. One of the ways in which non-conventional sources of energy can be integrated into the traditional system of generation-transmission/distribution of electrical energy are electrical microgrids. Electrical microgrids are controllable and manageable systems where non-conventional energy sources, loads and storage systems interact through interfaces made up of power electronics devices. Storage systems in electricity microgrids are key to the balance between energy generation and consumption, stability, system autonomy, storage and subsequent sale of surplus generation, among others. Depending on the technology used to store the energy, storage systems can be divided into electrical, mechanical, electrochemical, and chemical. Batteries are a form of electrochemical storage system, there are different types of batteries used in electrical microgrids, such as lead acid batteries and lithium-ion (Li-ion) batteries, the latter offering high power and energy density characteristics compared to other types of batteries. One of the great challenges of electrochemical systems is the degradation they suffer over time, this variable directly impacts electrical microgrids in terms of reliability and stability. In this research project a Li-ion battery degradation model is proposed in the context of an electrical microgrid, the model is based on a NARX neural network which was trained with a set of data available in the literature, the same set of data allows to determine the circuit parameters of a battery circuit model that serves as a connection between the microgrid model and the degradation model. The proposed Li-ion battery degradation model was evaluated by finding an error of less than 2%, considering the MAE (mean absolute error) and RMSE (root mean square error).