| Confronted with the growing energy crisis and environmental pollution,the rapid development of new energy rail transit vehicles is emerged as the vital approach to ensure the sustainable development of cities.In particular,lithium titanate batteries(also called LTO batteries)have been widely utilized in the energy storage system,owing to their attractive performance and environmentally-friendly character.Thus operation efficiency and safety of the whole vehicle is heavily dependent on the estimation accuracy of battery states.Unfortunately,most of existing methods related to battery state estimation focus on the commercial lithium-ion batteries with graphite anodes,yet their research conclusions cannot be directly extended to LTO batteries.Considering the high rate characteristics and wide temperature range of onboard application in the rail transit vehicle,the accurate state estimation of LTO batteries becomes a highly challenging issue.Based on numerous research literatures and international testing standards,a test bench of commercial LTO batteries is established,also massive data including reference performance and cyclic aging behavior are collected.Regarding the state estimation of LTO batteries used in the new energy rail transit vehicles,the research work accomplished in this paper mainly includes model-driven state estimation method of LTO batteries on short time scale,and aging mechanisms based state estimation method of LTO batteries on long time scale,which can be further split into the following four parts:(1)To improve the precision of traditional equivalent circuit model at high rate and harsh temperature conditions,which is further used in battery state estimation on short time scale,the current and temperature dependencies of model components are comprehensively analyzed,such as open circuit voltage-state of charge mapping,ohmic resistance and polarization phenomenon.Then it is concluded that the decrease of model accuracy can be attributed to the lack of electrochemical polarization model.Accordingly,a simplified form of Butler-Volmer(BV)equation is carefully derived from the original form by the electrochemical dynamics theory.Moreover,the electrochemical equation based equivalent circuit model for LTO batteries is proposed,which takes the same structure with the first-order RC model but embeds a simplified form of BV equation to predict the polarization voltage drop.In addition,the proposed model is verified against different conditions including material composition,loading profile,and temperature.(2)To enhance the robustness of traditional estimation method concerning battery state of charge(SOC)and state of power(SOP)at various temperature conditions,state of useful charge(SOU)is introduced to separate the coupled properties existed in the SOC definition.And new voltage constraints considering the open circuit voltage are proposed for battery SOP prediction,which originates from the nonlinear characteristics during charge and discharge process of LTO batteries.Based on the proposed equivalent circuit model and state observers,different model-driven SOC estimation methods are presented.According to results comparison among four observers in terms of absolute error median,absolute error maximum and response time,on one hand,it can be found that proportional-integral observer is easy to be implemented in engineering but requires a longer time to be convergent.On the other hand,filtering methods based on Bayesian optimal theories,which are capable of correcting the error rapidly,show low sensitivity to temperature variation.Combing the new voltage constraints and pulse response derived from the proposed model,a model-driven SOP prediction method is presented and verified at three ambient temperatures.(3)To reveal the aging mechanisms of LTO batteries when cycled at high rate and different temperatures,which is further used in battery capacity estimation on long time scale,fade trends of LTO batteries performance in terms of capacity retention,direct current(DC)internal resistance and other parameters are deeply analyzed,indicating the battery degradation is primarily resulted from the capacity loss.Furthermore,high current dependency of battery resistance is observed during the entire life especially at charge condition.According to the voltage characteristics of spinel-structured LTO,the regional division of the typical quasi open circuit voltage curve is demonstrated.Utilizing incremental capacity analysis(ICA)and differential voltage analysis(DVA),it could be detected that capacity degradation of LTO batteries is primarily caused by loss of active material in the positive electrode,yet pure LLI is detectable when cycling at lower currents.Among the four region capacities in the DV curve,the degradation of region capacity located in the low SOC follows a uniform pattern,which is not influenced by other regions.(4)To establish the selection rule of features used in the capacity estimation method related to phase transformation,the consistency between features extracted from IC and DV curves when determing battery capacity is proved by utilizing different spectral functions.According to the SOC distribution of onboard applications and aging mechanisms of LTO batteries,the optimal selection rule of features related to phase transformation is presented.Based on the linear relation between the chosen feature and battery capacity,an online estimation method for LTO batteries capacity is proposed,which is built on partial DV curve.In addition,the proposed method of capacity estimation is verified against different rates and temperatures.Overall,this study proposes an electrochemical equation based equivalent circuit model,a model-driven SOC estimation and SOP prediction method,and an estimation method of battery capacity based on partial DV curve.The research listed above successfully solves the state estimation issues of LTO batteries at high rates and different temperatures,which is of great values in the engineering field and future popularization. |
PDF Full Text Request