Comparison of the performances of heuristic optimization algorithms PSO, ABC and GA for parameter estimation in the discharge processes of Li-NMC battery

Abstract

The effects of the studies performed for the development of cells, which are the fundamental components of electrochemical battery units are felt in many different areas such as electric rail transportation systems, battery-based energy storage systems, battery units in electric vehicles, and energy storage units for individual use. For this goal, studies conducted by other searchers in the similar field have been investigated. In this paper, optimization techniques are used to guess the model parameters with major righteousness using the electrical equivalent circuit model of the battery. The discharge processes of the 18650 cylindrical type 2000 mAh Li-NCM battery cell with 1 A pulsed constant current at 25 ºC have been investigated. The real parameter values obtained have been transferred to the electrical equivalent circuit model. The open circuit voltage is determined as a functional term depending on the state of current supply level by using the curve fitting method in the Matlab. Studies have been carried out on particle swarm optimization algorithm, artificial bee colony algorithm, and genetic algorithm to estimate the battery output terminal voltage by using the open circuit voltage. Comparisons have been made and differences have been analyzed between the technics by using different statistical methods of true error values, the correct prediction ability, and response speed. As a result, the optimization method that makes the most accurate estimation has been determined.

Online estimation of state-of-charge using auxiliary load

“…Therefore, several models were investigated to imitate the dynamic properties of the battery such as: (i) Electrical equivalent models, where single or multiple parallel pairs of resistance, capacitance, and additional circuit components are used to build an electric circuit that mimics the dynamic features of Lithium-ion batteries [12]. Moreover, PSO (Particle Swarm Optimization), GA (Genetic Algorithms), and ABC (Artificial Bee Colony) have been studied to predict the potential difference in the battery's output poles using the OCV method [13]. Based on the comparisons presented in [13], the ABC algorithm has demonstrated the best performance in OCV prediction operations.…”

“…Moreover, PSO (Particle Swarm Optimization), GA (Genetic Algorithms), and ABC (Artificial Bee Colony) have been studied to predict the potential difference in the battery's output poles using the OCV method [13]. Based on the comparisons presented in [13], the ABC algorithm has demonstrated the best performance in OCV prediction operations. After ABC, the PSO algorithm demonstrated satisfactory outcomes.…”

Comparative Analysis of Lithium-Ion Batteries for Urban Electric/Hybrid Electric Vehicles

“…where i-the measured current depending on the gross balancing time t; C n -the internal capacity of the battery. An evaluation of the SOC is carried out according to the dynamic CC method (Coulomb counting-∆Q) [15,32,34]. For example, if the initial state is SOC 0 , using 1 A current to charge the battery for t hours, they will add I•t Ah of charge to the battery.…”