SOC estimation of lithium-ion batteries based on the condition of vessels

The battery management system (BMS) of electric ships plays an essential role in ensuring such ships work smoothly and safely. Its vital function is to estimate the State of Charge (SOC) that can show helmsmen remaining mileage, improve the efficiency of power distribution and prolong life expectancy of batteries. With the aim of SOC estimation, four methods are usually adopted, including ampere-hour integral method, open-circuit voltage method, data-driven method, and model-based method. Due to the integral action, ampere-hour integral method is beset with the problem of error accumulation ^[1] . As for open-circuit voltage method, real-time estimation cannot be carried out, because the open-circuit voltage cannot be obtained in real time. Making use of the black-box model to simulate battery characteristics through data training, data-driven method can estimate accurately. Nevertheless, a massive amount of data about battery is required for training. The key of model-based method lies in the accuracy of the established battery model. Such model can be classified into two categories: the electrochemical model and the equivalent circuit model. The former is not suitable for practical application, since it involves the specific chemical reaction inside the battery and there are plenty of differential equations need to be solved. In contrast, the latter has the advantages of high precision and small calculation when it describes external characteristics of the battery through the combination of circuit components. Based on Hai Gong You 31's navigation condition and the equivalent circuit model, this paper applies PSO algorithm to make a parameter identification in the dynamic working condition of ternary lithium-ion batteries at different temperatures and generates the three-dimensional map (SOC-temperatures-model parameters) ^[2] . On this basis, unscented Kalman filter (UKF) algorithm is employed to estimate SOC ^[3] .