Data Pieces-based Parameter Identification For Lithium-ion Battery Of Electric Vehicles

Facing energy crisis and air pollution problems,electric vehicle is one of the most promising methods to reduce energy consumption and emissions worldwide.Sales of electric cars increase rapidly in recent years.However,traction battery is still the bottleneck of electric vehicle industry,because of its high upfront cost and immaturity of management technology.Battery consistency problem exists in battery package,and many situations such as over charge/discharge,high current rate,or charging at low temperature may cause permanently nonreversible damage to lithium-ion batteries.So that it is of paramount importance to build a reliable battery management system to ensure safe,reliable and efficient battery operation.Model-based battery management is attracting increasing attentions from academic and industry.However it comes with a big problem,model parameters such as bettery capacity and impedance change with temperature and battery aging.Then how to accurately identify battery parameters under complex battery operating conditions is a key issue for BMS to ensure precise management.The main objective of this dissertation is to develop a method to identify battery parameters such as capacity,impedance-Ah relationship and OCV-Ah relationship etc.only based on battery operation data.For this purpose,a data pieces-based battery parameter identification method is proposed.The whole dissertation is working on theory development and validation of this method,and tries to implement this method on real vehicles.The research content includes:1.A data pieces-based battery parameter identification method for lithium battery is proposed.The key philosophy of this method is to cut the battery training data into many small data pieces,and then an offline optimization method is applied to identify parameters for each small data piece,then the identified parameters are connected piece by piece,and finally a whole impedance-Ah and OCV-Ah relationships can be achieved.Parameters of two types of lithium-ion batteries at different aging levels and different temperatures are identified using this method.A comprehensive parameter comparison is conducted for different battery cells at the same aging level,and parameter comparison is also conducted for one battery cell at different aging levels.The result validates the effectiveness and adaptability of the proposed method.2.A battery capacity identification method is proposed based on multiple incomplete data pieces.Most times the battery operation data recorded from vehicle daily driving are incomplete data pieces.In order to identify battery capacity,first parameters of the data pieces can be identified,and then a method is proposed to connect and merge the identified OCVAh pieces together,so that a long OCV-Ah curve can be achieved.Finally an OCV-Ah aging database matching process can be conducted to identify capacity of the battery.3.Improvement of battery model accuracy for aged battery.After battery aging,simulation accuracy of battery equivalent circuit model becomes poor.Three approaches are tried to improve model accuracy for aged battery.First,six different lumped battery model structures are compared systematically.Then we try to shorten the length of training data pieces to improve accuracy.Third,the parameter identification method is improved.And finally an SOC estimation approach for the aged battery is conducted based on 2-RC model and 1-RC model to validate the improvement of SOC estimation brought by a more accurate battery model.4.Vehicle field test and battery parameter identification for the test vehicle.Vehicle field test is conducted to collect battery operation data.Based on the test data,battery capacity,impedance-Ah and OCV-Ah relationships are identified.Then adaptive extended Kalman filter is employed to estimate battery SOC for a real driving cycle based on the identified battery parameters.The result shows good accuracy and effectiveness of the identified parameters.5.Consistency analysis of battery cells and battery equalization capacity analysis.Parameters of all 97 battery cells of the test vehicle are identified.For the battery cells cannot be fully charged or discharged,a method is proposed to fill up the incomplete OCV-Ah curves,so that capacity of all battery cells can be achieved.Then a comprehensive comparison of battery OCV,capacity and internal resistance is conducted.The result shows good consistency of the battery package.Then an equalization analysis is conducted for the battery package,and two methods are proposed to calculate the bleeding capacity for each battery cell to achieve equalization,thus providing accurate reference information for battery active equalization control.