Model-Based Battery SOC Estimation Based on GA-UKF Algorithm
A model-based State-of-Charge (SOC) estimation approach of Lithium-Ion battery is put forward based on parameter identification using Genetic Algorithm (GA) and state prediction using an Unscented Kalman Filter (UKF) in this paper. Firstly, the second-ord
- PDF / 883,561 Bytes
- 9 Pages / 439.37 x 666.142 pts Page_size
- 22 Downloads / 250 Views
ds: SOC estimation Kalman Filter
1
· Genetical algorithm · Unscented
Introduction
Recently, Lithium-Ion batteries have gradually become the research focus in the field of energy storage due to its excellent performance [1]. However, The batteries show obvious aging phenomenon with service time increasing, leading to the inaccurate state estimation such as Remaining-Useful-Lifetime (RUL) and SOC [2]. For the sake of safe operation, an accurate state estimation method is required to efficiently manage the battery pack. Among all the states of Lithium-ion battery, SOC is the most significant one. The SOC is defined as the following formula: SOC =
Qf Qr × 100% = (1 − ) × 100% Qn Qn
(1)
where Qr indicates the current remaining capacity of the battery; Qn denotes the rated capacity of the battery, and Qf indicates the current consumed capacity of the battery. It is difficult to measure the SOC of Lithium-ion battery directly through an electronic sensor [3]. Therefore, the estimation of SOC is implemented with the assistance of other measurable variables like terminal voltage and current. However, precise estimation of SOC is a challenging task because of the complex c The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 Y. Jia et al. (Eds.): CISC 2020, LNEE 706, pp. 298–306, 2021. https://doi.org/10.1007/978-981-15-8458-9_32
Model-Based Battery SOC Estimation
299
internal chemical dynamics of Lithium-ion battery. The common SOC estimation consists of the ampere-hour integration method, the open circuit voltage method and the model-based method [4,5]. Among all the available methods, the modelbased method is the most suitable approach for estimation. The overall process of model-based method include battery modelling and algorithm implementation. The battery modeling research can be seperated into electrochemical model, equivalent circuit model and black box model [6–9]. Parameters of selected models can be offline identified through different kinds of algorithms, such as GA [10]. As to algorithm implementation, adaptive filter algorithm represented by Kalman Filter (KF) family algorithm has been put into widespread research. The fundamental principle of KF family algorithm is to recursively estimate the state of current time step based on the state estimated previously in the last time step and the current measurable variables [11–15], providing accurate SOC estimates for the model. The rest of the paper is listed as follows: In Sect. 2, the ECM is established. Also, the Uocv -SOC relationship curve is calibrated on the basis of constant current charging and discharging experiment, and parameter identification is thus committed using GA. In Sect. 3, UKF algorithm is utilized to perform accurate SOC estimation. In Sect. 4, the Main Discharge Cycle Condition (MDCC) is applied for experimental verification of the proposed GA-UKF algorithm. The estimation results are shown to validate the effectiveness. Finally, the conclusion is given in Sect. 5.
2
Lithium-Ion Ba
Data Loading...
