The Research Of State Of Charge Estimation Method For Ternary Li-ion Battery

Ternary Li-ion battery has higher energy density, mature technology and lower cost than Li-ion phosphate battery or the others, which is a better choice to solve EV range problem. Panasonic NCR18650A is a typical LiNiCoAlO2 ternary Li-ion battery. Assembling many cells into packs is suitable for EV power source and solar energy storage. The estimation of SoC is the most significant and essential task for battery management system. The mature SoC estimation method is employing intelligent algorithm combined with battery model. The parameters of battery model can be extracted from elaborated experiment. After a survey of current works, there are some common issues undiscussed, e.g. the data processing method is unclear, the data acquisition sampling rate is ambiguous, the classical definition of SoC is not suitable for dynamic operation working condition, the influence on SoC precision of different standing duration is unknown, extraction of parameters under charge and discharge respectively is worth or not, the number of RC network nodes for cell polarization modeling is uncertain.In this paper, a battery test system is constructed. SoC definition is calibrated by using official data from Panasonic. The temperature, current rate etc. are accounted during the calibration. Based on the calibrated SoC and feature test data, battery model is constructed and parameters are extracted in charge and discharge conditions respectively. The square-root unscented Kalman filter is selected to estimate SoC aiming at solving the negative covariance problem of unscented Kalman filter. The main work of this paper is listed as follows.First, capacity test, feature test and simulated operating condition test are conducted by using self-made battery testing system. The data processing method is given. Voltage sampling rate is set to 1 OHz to measure the Rohm voltage drop of the zero-state response.Second, SoC definition is calibrated by taking charge or discharge rate, temperature, cycle times and charging efficiency. The test shows the calibrated definition fits well with the real data and can be used for Ah counting in dynamic operating condition.Third, the discrete equivalent circuit model is built. Model parameters are extracted by analyzing potential and over potential features of battery. Hysteresis effect and standing duration are taken into account for potential feature analysis. Over potential feature is studied by calculating the ohmic internal resistance and degree of polarization in charge and discharge condition respectively at different SoC.Finally, square-root unscented Kalman filter is used to estimate the SoC. This SoC estimation proved its validity with a high precision. Furthermore it had good robustness for initial value offset.In this paper, a system has been provided for battery test and the parameters of NCR18650A model are obtained. The SoC estimation method is achieved by using SR-UKF combined with equivalent circuit battery model. This paper proves its value for application by solving the mentioned common problems of SoC estimation.