| The dual pressures of energy crisis and environmental protection have promoted the rapid development of electric vehicles and battery energy storage system.As the common power source of electric vehicles and battery energy storage system,lithium-ion batteries are usually equipped with a battery management system(BMS)to ensure reliable and stable operation.Therefore,the efficient and accurate testing of BMS is important.Due to the problems of poor repeatability,time consuming,peripheral devices required,and safety hazards in testing BMS with real batteries,there is an urgent need for the battery emulator,i.e.,the "virtual battery" involved in this paper.The application of virtual battery is significantly limited by the shortcoming of the existing battery models with low accuracy,large computational effort and poor applicability.Meanwhile,how to ensure that each cell in the virtual battery has the same parameter inconsistency as the real battery is the focus of current research.In addition,the use of virtual battery packs to test BMS lacks a mature solution for technical difficulties such as test profile extraction,reference value acquisition and error analysis of the key parameters.In order to solve the above problems,the following specific work is carried out in this paper:(1)Aiming at the problem of low accuracy,large amount of calculation and poor applicability of the existing model for virtual battery,a variable-order equivalent circuit model for lithium-ion battery with multi-time scale unified and multi-factor couplied is established.Based on the analysis of the battery impedance characteristics in the time-frequency domain,the boundary between the short time scale and the long-time scale was clarified,and fractional-order and integer-order models were used to describe the output characteristics of the short and long time scales,respectively.On the basis of considering the evolution law of model parameters with polarization state and current amplitude,a variable-order equivalent circuit model of lithium-ion battery is established by integrating fractional-order and integer-order models and considering different time scales.At the same time,breaking through the conventional methods and considering the transient characteristics of long time scales,a parameter identification method based on continuous short-time pulses is proposed.It not only improves the simulation accuracy of model parameters,but also reduces the computational effort of the identification method.(2)Aiming at the problem that the existing virtual battery pack model cannot accurately simulate the inconsistency of real battery packs,considering the distribution characteristics and correlation of 49 battery pack parameters,two battery pack inconsistency modeling methods for different data dimensions are proposed and compared in terms of the applicability.Multivariate kernel density estimation is used to improve the shortcoming of the traditional Monte Carlo method that is difficult to describe the joint probability distribution of multidimensional parameters.A battery pack inconsistency modeling method based on multivariate kernel density estimation and Monte Carlo sampling(MKDE-MC)is proposed.In addition,an improved variational autoencoder(VAE)is used to implement the battery pack inconsistency modeling with 49 parameter dimensions and any number of battery series connections.The simulation results show that the proposed methods are able to reconstruct the parameters of the virtual battery pack with the same distribution as the real data with high accuracy.The results of the applicability comparison analysis show that the MKDE-MC method is suitable for scenarios with small dimensions and large sample sizes,while the VAE method is suitable for scenarios with large dimensions and small samples.(3)Aiming at technical difficulties such as extraction of test profile and error analysis of key parameters,key studies of SOC\SOP estimation accuracy test and temperature rise test of BMS with virtual battery were carried out.Taking the common estimation methods of SOC and SOP as the research object,considering the mediating effect and the moderating effect,the quantitative analysis method of estimation error with multi-factor based on path analysis is proposed.It is revealed that the most significant factors affecting the estimation accuracy of SOC are the current and voltage acquisition errors.The indirect effects of different error sources on SOP estimation accuracy through SOC estimation accuracy are analyzed.For the extraction of BMS temperature measurement testing profile,based on cluster analysis and Elastic Net regression method,the standard profile with high volatility is simplified according to the same temperature rise characteristics under the same heat dissipation conditions,and the temperature rise test profiles that can be easily implemented by high accuracy simulation are obtained to improve the testing capability of the virtual battery pack.(4)Aiming at the problems of poor repeatability,time-consuming and safety hazards of traditional BMS test methods,a BMS test platform with strong operability and high flexibility was established based on virtual battery technology,and the method of obtaining reference values of key parameters was proposed and relevant tests were conducted.Based on the BMS test requirements,a BMS test platform based on the virtual battery is built to realize the custom download of the battery model and the automatic test of the BMS function.The test methods for the BMS key parameters(SOC and SOP estimation accuracy)are studied,and the influence of test profiles on SOC estimation accuracy is clarified.The method of obtaining reference values of SOP under different scenarios is realized by multi-boundary feature chart method and battery model calculation method.Based on the built test platform,the different functions of BMS are tested and analyzed.The virtual battery cell model,the battery pack inconsistency model,the quantitative analysis method of battery SOC and SOP estimation errors,the battery pack temperature rise test profile extraction method,and the established BMS test platform proposed in this paper break through limitation of the virtual battery application and realize the accurate,fast,efficient and safe BMS test,which is of great significance to improve the safe and reliable operation of the battery system. |
