| Compared with traditional primary batteries,lithium-ion batteries,as a new type of recyclable power supply energy sources,have attracted great attention from countries all over the world because of their long battery life,stable discharge current,and long cycle life.It is widely used in civil and military fields.The state-of-charge and health status of lithium-ion batteries are the core indicators for monitoring and evaluating the performance of batteries.The state-of-charge(SOC,State of Charge)of power lithium-ion batteries characterizes the remaining discharge of the battery under the current charge and discharge cycle.The ratio of the capacity to the maximum dischargeable capacity.Accurately predicting the state of charge of the battery is of great significance for improving the driving performance of the vehicle and monitoring the remaining mileage of the vehicle in real time.The battery health status is an evaluation based on the calculation of the correlation between the rated capacity of the lithium-ion battery and the remaining useful life(RUL,Remaining Useful Life).The rated capacity of the battery refers to the amount of power that the battery can discharge under certain discharge conditions(a certain degree of saturation,a certain current density and a termination voltage).RUL characterizes the battery discharge performance(or the maximum dischargeable capacity of the battery).SOC and SOH(State of Health)are important parameters in the battery management system.The two cooperate with each other to make the battery state prediction more accurate.Common analysis methods include: mathematical model method,intelligent algorithm led by neural network,direct measurement method,etc.Most of these methods have the advantages of higher prediction accuracy and the prediction model established by some algorithms is relatively simple,and the program runs fast,but there are still some problems: electric vehicles are easily affected by factors such as current fluctuations and nonlinear operating conditions during SOC prediction.There is a problem of large errors.When estimating the health status of a lithium-ion battery under actual operating conditions,the mathematical modeling is difficult,the internal parameter measurement is complex,and the model adaptability is poor.For the above problems,the main research contents of this article are as follows:(1)The working state and internal chemical reaction mechanism of vehicle-mounted lithium ion are introduced in detail.Combining the commonly used battery equivalent models with the prediction difficulties of SOC and SOH and analyzing the advantages and disadvantages of each model,at the same time comprehensively considering the external environment and the influence of the battery’s own factors,exploring the relationship between the two.From the perspective of influencing factors,the influencing factors of SOC and SOH are roughly the same,and are susceptible to environmental factors such as ambient temperature,charge and discharge current,voltage,etc.However,as one of the influencing factors of SOC,SOH has a direct impact on the prediction accuracy of SOC In order to avoid double counting,the SOH should be analyzed first.(2)Due to the complexity of the factors that affect battery life,in order to make the overall SOH prediction of the battery more accurate,the factors that affect the rated capacity of the battery are divided into internal degradation factors and characteristic influence(external)factors according to the degree of influence.According to the degree of influence on the SOH prediction error of the battery,the characteristic influencing factors are divided into two using the Principal Component Analysis(KPCA,Kernel Principal Component Analysis): the main influencing factors(temperature,charge and discharge rate)and incidental factors(collector error,humidity).Due to the dynamic and nonlinear characteristics of this type of influence factors,it is necessary to combine the more accurate RUL prediction data calculated by the particle swarm improved radial basis neural network algorithm mentioned in this article,and combine the predicted RUL data with harsh engineering.Under the condition of battery discharge performance degradation(temperature too cold,hot,overcharge,discharge,overcurrent,etc.),correct it.The modified RUL value is used as the input terminal,combined with the incidental factors and characteristic influencing factors,and substituted into the particle swarm optimization radial basis neural network algorithm(PSO-RBF)to make the rated capacity prediction according to the current driving environment.So as to achieve fast and accurate prediction of dynamic rated capacity.(3)By using the more accurate SOH data obtained above,the SOC is further predicted.Considering the advantages and disadvantages of various algorithms comprehensively,an optimized radial basis neural network algorithm is selected to estimate the SOC.SOC calculation accuracy is mainly affected by factors such as SOC initial value(SOC0),battery calibration capacity(CN),temperature and other factors.Combined with battery charging and discharging characteristics,the battery is divided into three states: charging,standing,and discharging to reduce calculation time and improve prediction Precision.According to the different working states of the battery when the car is driving,corresponding strategies are adopted to predict the SOC.In the battery discharge phase,a radial basis neural network improved by particle swarm is used to dynamically predict the SOC by collecting data such as the open circuit voltage,temperature,discharge rate,and cycle number of the battery;when the battery is in a static state and a charged state,its location The environment is relatively stable,and the current and temperature do not change much.To reduce the amount of calculation,the open circuit voltage curve considering temperature drift and the current node mutation curve during charging are made into a two-dimensional array table.The proportional search method in the binary search method is used.The SOC prediction value obtained by dynamic measurement while the car is running is corrected.The experiment uses the remaining battery life data and SOC measurement data measured by the provincial key laboratory for simulation.At the same time,the battery data published by the National Aeronautics and Space Administration is used for comparative experiments.The results show that the composition is analyzed and optimized by the KPCA algorithm.The PSO-RBF algorithm predicts an average error of remaining service life of 1.9% and a rated capacity prediction error of 2.3%.The results prove that the SOH prediction model has certain feasibility.At the same time,the obtained battery health status data is input into the SOC prediction model,combined with the SOC change curve and temperature,voltage and other environmental change data obtained by simulating road driving,and finally the maximum error of SOC predicted by the step-by-step table lookup method is finally obtained Is 2.1%.Overall,the model has high prediction accuracy,and the calculation time is fast,which has certain feasibility. |
