System identification of a lead-acid battery based on experimental data

Abstract

Listen

Translate article

This paper aims to identify a battery model using experimental data obtained from the discharge of a lead-acid battery. In particular, obtaining a representative mathematical model for the system is important to accurately determine the electrical characteristics of the battery and analyze its discharge behavior during use. Additionally, the model is extremely useful for the design of the Battery Management System (BMS), which consists of an electronic circuit that manages the rechargeable batteries and at the same time protects the circuit from operating outside its safe operating area. Regarding the functioning of lead-acid batteries, although there are complex chemical reactions and laws from physics that can be employed to describe the system dynamics, mathematical models obtained from identification techniques can be an alternative to modeling and understanding their behavior. Even though there are simple models to more sophisticated models, the degree of complexity of the model will be determined by the application for which it will be used. In this sense, this work presents the identification of an electrical model for a lead-acid battery from data gathered. Jackey's model was adopted to represent the battery dynamics, since this model, in addition to having the parameters modeled as resistive and capacitive elements and being well suited to the experimental data, offers the advantage of not presenting a high level of complexity. Furthermore, it is intended to extend the methodology applied to obtain a Matlab/Simulink model that simulates the energy generation and storage system from a photovoltaic system. In this work, experimental data are sampled for different battery discharge rates using the Arduino. Then, the optimization tools of Matlab environment were used for system identification. The objective is to apply the optimization techniques to obtain the parameters of Jackey's model. Finally, according to the results obtained, the mathematical model identified is representative of the system, and, therefore, the tool is a viable alternative for the mathematical modeling of lead-acid batteries.