Research On Equalization Algorithm Based On LiFePO₄ Cell State Of Charge

Electric vehicles have the advantages of low noise and nearly zero emissions, so it is a good alternative to solve the problems of energy shortage and environment pollution.Battery pack which is important for hybrid electric vehicles(HEVs) or electric vehicles(EVs) manipulates the performance of vehicles to a great extent. Due to high power density and high cell voltage, lithium-ion battery becomes more and more popular for HEVs or EVs. However, the battery inconsistency is inherent and inevitable during production. Moreover, vehicles usually operate in various conditions such as starting/stopping, acceleration climbing or deceleration braking according to demand. In such case, the charging/discharging mode, current magnitude and duration of battery pack are uncertain and thus the internal heat distribution is uneven, which also aggravates inconsistency especially for the high-capacity battery pack connected in series. In this condition, the battery cycle-life would rapidly decay or even failure. To achieve more wide application for lithium-ion battery, an effective equalization scheme is essential to improve battery performance and extend battery life-time. Accordingly, the main research contains:1) Referring to the problems of energy loss for dissipative circuit and complex structure and low efficiency for non-dissipative circuit, taking the characteristics of lithium-ion battery system into consideration, an integrated equalizer based on both charging and discharging is proposed. Furthermore, parameters of discharge resistance and DC/DC converter are designed and achieved good performance for equalization optimization.2) Experimental research on constant current charge-discharge characteristics, pulse charge-discharge characteristics, cycle-life characteristics, internal resistance, hysteresis characteristics, open circuit voltage(OCV) characteristics and efficiency characteristics of lithium-ion battery are developed by Digatron test equipment and the regression models are established. Under this condition, the physical performance, mathematical expression and evolution rules related to characteristics of lithium-ion battery are achievedcomprehensively for battery model, SOC estimation and equalization optimization.3) Considering that the voltage equalization leads to over equalization while the capacity equalization is not available for on-line application, the SOC equalization is utilized accordingly. Referring to the problems of poor adaptability and low accuracy for battery SOC estimation in literature, studying the electrochemical mechanism and characteristics of dynamic state, an adaptive weighted feedback integration algorithm which is composed of weighted feedback model and dynamic modified model is proposed for on-line SOC estimation. The weighted feedback model based on Ah counting method and equivalent circuit model(ECM) is established by genetic algorithm-fuzzy logic controller(GA-FLC) algorithm. While the dynamic modified model is a modification for weighted feedback model by battery current, temperature, state of health(SOH),self-discharge rate and self-recovery factor through quantum-behaved particle swarm optimization-back propagation(QPSO-BP) network. To validate the adaptive weighted feedback integration algorithm, experiments are executed on condition of different initial SOC, different temperature and different SOH and an accurate battery SOC estimation is obtained accordingly, which is essential for equalization.4) Referring to problems of over equalization, energy loss and time consumption,addressing the issues of time Alignment and noise elimination, an cell SOC equalization optimization based on adaptive genetic algorithm is proposed. In order to improve the real-time capability, a parameter identification technique based driven-data is employed to realize on-line equalization. Furthermore, equalization is studied by comparison with mean-difference method and fuzzy logic controller(FLC). To validate the SOC equalization based on adaptive genetic algorithm, simulations and experiments are both implemented and the results show that the adaptive genetic algorithm not only improves the battery inconsistency greatly but also considerably improves the time efficiency and energy efficiency.