Research On SOC Estimation And Active Equalization Of Power Battery

The problems of the continuous depletion of non renewable energy and serious environmental pollution are becoming more and more prominent in today's society.As a widely used modern transport,the number of automobile is increasing continuously.This has been fuelling a revival of research,development and manufacture of electric vehicle.In recent years,the lithium-ion battery has been widely used in electric vehicle field.To prevent over-charging and over-discharging of lithium-ion battery,and provide the accurate information about the endurance mileage for drivers,the battery management system needs provide the real-time state of charge information.Meanwhile,in order to prolong the battery life and extend the endurance mileage,the equalization on battery cells which have different state of charge is necessary.Therefore,the research on battery state of charge estimation and battery equalization is carried out in this thesis.The state of charge estimation is easily affected by the noise signal in battey voltage,current signal,etc.This thesis introduces an algorithm based on the wavelet transform matrix and adaptive extended kalman filter.It can de-noise the battery voltage,current signal effectively,and increase the accuracy of state of charge estimation under the electric vehicle working condition.This thesis analyses the construction process of battery model,construction method of wavelet transform matrix and the process of signal decomposition,coefficients adjustment and signal reconstruction.It also achieves the accurate estimation of state of charge based on the wavelet transform matrix and adaptive extended kalman filter.The accuracy of battery pack model and the battery equalization circuit have effects on battery pack state of charge estimation.This paper analyses the battery pack state of charge and introduces a new definition of battery pack state of charge which is suitable for the practical application of electric vehicle.Meanwhile,the effects of the equalization current in the equalization circuit are also analysed carefully in this paper.The equalization current is considered carefully in the battery pack state of charge estimation,and the estimaton accuracy is increased.Finally,the existing battery pack active equalization circuit has the disadvantages of large volume,slow equalization speed.An active equalization circuit which is based on switch matrix and bidirectional multiport flyback converter is proposed in this paper.The switching frequency is 1 MHz and the GaN device is applied in the flyback converter.It can reduce the volume and increase the equalization speed effectively.The test bench of battery pack state of charge estimation inc ludes the battery management system,10 series-connected Li?NiCoMn?O₂ batteries?each battery cell has the capacity of 200 Ah and the voltage range is from 3.6V to 4.2V?,DC source and electronic load,etc.Co mpared to the maximu m state of charge estimation error without de-noising,an estimation error reduction of 1.5%is achieved after signal de-noising.Co mpared to the traditional state of charge estimation based on the battery pack model,the state of charge estimation erro r reduction of 2.5%and 1%can be achieved during the charging and discharging process respectively.Co mpared to the method without considering balancing current,an error reduction of 1.4%can be achieved by the approach proposed in this thesis.The proposed battery pack active equalization circuit is based on the switching matrix and the bidirectional multiport flyback converter.Compared to the existing active equalization circuit,it has the advantages of small volume and fast equalization speed.