CPSO-Based Parameter-Identification Method for the Fractional-Order Modeling of Lithium-Ion Batteries

For battery equivalent circuit model parameter identification, the fractional-order modeling and the bionic algorithm are two excellent techniques. The former can describe the impedance characteristics of batteries accurately, while the latter has natural advantages in solving some nonlinear problems. However, the high computational cost limits their application. In this article, a parameter-identification method for a battery fractional-order model based on the coevolutionary particle swarm optimization (CPSO) is proposed. In this algorithm, a large number of optimization calculations are dispersed between the adjacent sampling times in the form of evolutionary steps by CPSO, so the algorithm can run in real time with the sampling process. In addition, the simplified fractional approximation further reduces the computational cost. By conducting tests under various algorithm conditions, we evaluate the main factors affecting the algorithm performance in detail. Our results show that compared with the integer-order model, the fractional-order model can track the optimal value more effectively in a wider optimization space, CPSO can track the time-varying battery parameters in real time by continuous evolution, and computational costs can be effectively reduced by using a fixed-order fractional-order model and appropriately compressing the length of the historical data required for fractional-order computation.

Automated summary

This article proposes a battery fractional-order model parameter identification method based on cooperative evolution particle swarm optimization, which can run in real time and reduce computational costs. Tests show that the fractional-order model is more effective in a wider optimization space, CPSO can track time-varying battery parameters in real time.