In this paper, a collaborative online algorithm is proposed to estimate the state of charge (SOC) and state of health (SOH) of lead-carbon batteries that participate in frequency regulation of a power system with a high proportion of renewable energy. The algorithm addresses the inaccurate estimation of energy storage battery states caused by continuous and alternating charging and discharging over a short period. Analysis of lead-carbon battery chemistry and materials reveals that the resistance of the diaphragm is the most influential factor in battery aging. In addition, the hysteresis characteristics of an energy storage battery vary significantly between the charging and discharging stages. A second-order RC equivalent circuit model is proposed that considers the contact and diaphragm resistances, and hysteresis characteristics. Based on this, models for constant current charging interaction, constant voltage charging interaction, and dynamic discharging interaction are developed. The adaptive forgetting factor recursive leassquare (AFF-RLS) method is used to identify the parameters of the interactive models. Then an interactive multiple model with the embedded unscented Kalmanfilter (UKF) is used to estimate the SOC of the energy storage battery. The membrane and contact resistances identified by the interactive multi-model (IMM) are used to estimate the SOH, and online collaborative optimization of the SOC and SOH is achieved. The error of the proposed SOC estimation method is experimentally verified to be within 2%, which is less than 5% of the standard value, and the error of SOH estimation is within 0.5%, demonstrating the high accuracy of the proposed method.