Accurate estimation of the state of charge (SOC) of lithium-ion batteries relies on precise model parameters. When using the forgetting factor recursive least square (FFRLS) algorithm for parameter identification of the equivalent circuit model of lithium-ion batteries, improper selection of initial iterative values can lead to low identification accuracy and slow convergence speed. To address this issue, circuit analysis is combined with the FFRLS algorithm, and then an improved initial value-FFRLS (IIV-FFRLS) algorithm is proposed. Firstly, offline identification is performed to obtain the equivalent circuit model parameters corresponding to various SOC points, which are then fitted using a polynomial function. Secondly, the initial SOC is obtained using the initial open circuit voltage (OCV) and the OCV-SOC curve, which is then substituted into the parameter fitting function to obtain the initial parameters. Finally, these initial parameters are used in the recursive formula to obtain the initial iterative values for the IIV-FFRLS algorithm. Parameter identification is performed for four operating conditions of lithium-ion batteries, and the results show that compared with traditional methods, the IIV-FFRLS algorithm reduces the average relative error by more than 58% and the convergence time by more than 23%. The IIV-FFRLS algorithm exhibits higher identification accuracy and faster convergence speed.