The accurate estimation of the solid concentration distribution in anode and cathode, and state-of-charge (SOC) for a Li-ion battery cell is significantly important for developing the real-time monitoring algorithm of the Li-ion cell's working operation, and further establishing an efficient and reliable advanced battery management system (BMS). Firstly, according to the porous electrode theory and concentration theory, in this article we present a systematic optimized model and a method of identifying the key internal state parameters based on a Li-ion cell's enhanced single-particle-model (ESPM), in which, an appropriate parameter vector is identified in the typical hybrid-pulse-power-characterization (HPPC) operation scenario by using the parameter sensitivity analysis method, and then this parameter optimization problem is evaluated by genetic algorithm. It is verified that the maximum relative errors of the cell's output voltage for ESPM are 1.92%, 3.18% and 2.86% under HPPC, 1C-discharge and urban dynamometer driving schedule (UDDS) current profiles, respectively. Secondly, by introducing some assumptions and reduction techniques, the battery ESPM is further reduced and then a novel interconnected state observer is proposed through the combination of the reduced ESPM and H∞ robust control theory framework, which can realize the concurrent estimation of solid concentration and SOC in anode and cathode. Finally, the comparative validation and analysis study are conducted by using the experimental data acquired in HPPC and UDDS condition to demonstrate the effectiveness and feasibility of the proposed interconnected observer. The results show that the maximum relative errors of output voltage for the ESPM, the single-electrode concentration observer (Obsv-1) and the proposed interconnected observer (Obsv-2) of Li-ion cell are 2.0%, 3.8% and 2.6%, respectively, under HPPC operation at 23 ℃; under the same input current profile and operating condition, the maximum relative errors of SOC estimation are 2.4%, 4.7% and 3.4%, respectively. Moreover, the maximum relative errors of cell's output voltage for ESPM, Obsv-1 and Obsv-2 model are 1.9%, 3.2% and 2.1%, respectively, and the maximum relative errors of SOC estimation values for these three mathematical models are 2.1%, 4.4% and 3.2%, respectively. It is concluded that the proposed robust observer for a Li-ion cell can accurately predict the output voltage and SOC, and can also improve the dynamic performance and robust stability of Li-ion cell, which provides a solid theoretical foundation for developing the BMS.