An Optimization Algorithm with Hausdorff-like Derivative in GRU for SOC Estimation of Lithium-ion Battery

This paper proposes an optimization algorithm in the gate recurrent unit (GRU) for state of charge (SOC) estimation. To introduce the error into the optimization algorithm, the Hausdorff-like derivative with the cost function is first defined. The Hausdorff-like derivative as one type of fractal derivatives is applied to the back propagation in the GRU, and the Adam algorithm with Hausdorff-like derivative are proposed to optimize GRU networks. An adaptive method of the order is proposed to tune the parameters of GRU networks. The SOC estimation experiments are carried out with different working conditions, and the experimental results are given to verify the effectiveness of proposed algorithm.