An Improved Sparrow Search Algorithm-Based Double Reservoir Open-Loop Echo State Network Model for Lithium-Ion Battery SOC Estimation Considering Wide Temperature Range Effects

Accurate estimation of the state-of-charge of lithium-ion batteries is an important prerequisite for ensuring safe and stable battery operation. To address the issue of difficulty in determining hyperparameters when estimating battery state-of-charge using traditional echo state network, this paper proposes an improved sparrow search algorithm-based double reservoir open-loop echo state network to predict the state-of-charge of the batteries using measurable variables. The model uses the sparrow search algorithm to optimize network hyperparameters, enabling the model to accurately adjust structural parameters and extract input features under different temperatures and operating conditions. The extracted information is then sent to the echo state network, where iterative training is used to gradually reduce prediction errors. Furthermore, by increasing the reservoir and removing output feedback, the algorithm complexity is reduced while improving model stability. Finally, the model is tested under different temperatures and operating conditions and compared with traditional echo state network and other models.