Electric car power battery SOC estimation method

Abstract

The invention discloses an electric car power battery SOC estimation method. The method comprises the steps that s1, association features being capable of representing the SOC are extracted from a BMS; s2, a deep neural network model is trained by means of the association features; and s3, on-line SOC estimation is conducted on the basis of the trained deep neural network model. The method startsfrom data, and dependency on various types of SOC estimation approximation models is avoided; and meanwhile, compared with other data-driven methods, the method is more suitable for processing a largeamount of BMS sample data and can achieve the higher estimation precision.

Description

Technical field [0001] This application relates to the technical field of new energy vehicles, in particular to a method for estimating the SOC of an electric vehicle power battery. Background technique [0002] Correctly estimating the SOC (State of Charge, battery state of charge) of the lithium battery is the basis for improving the utilization of the battery and prolonging the service life of the battery pack in the energy management of the vehicle. SOC varies significantly under different conditions such as temperature, rate, charge and discharge efficiency; battery working temperature has a significant impact on SOC, too high or too low will cause the battery's usable capacity to decrease; battery aging and self-discharge and other factors cause The accurate estimation of SOC is more difficult. In addition, there is a large inconsistency in the attenuation of the monomer capacity in the battery pack. In the actual operation of the electric vehicle, the estimation accuracy ...

AI Technical Summary

Problems solved by technology

[0005] 2) The operating state of the lithium battery is complex, and the time of turning off or turning on the lithium battery is random, which brings considerable difficulties to the estimation work

[0006] 3) The current of electric vehicles is unstable, the working environment is complex, and complex factors such as ambient temperature and battery self-discharge add many difficulties to the estimation

[0008] (1) The ampere-hour integration method needs the initial SOC value to give accurate estimation results. At the same time, in the actual operation of electric vehicles, due to the discreteness of BMS (BATTERY MANAGEMENT SYSTEM, battery management system) system sampling and the data transmission process Inevitable errors and data loss in the process make the error of the Anshi method often too large, which will eventually lead to deviations in the estimated SOC

[0009] (2) The open circuit voltage method records voltage and SOC data through discharge experiments, and predicts the value of SOC according to the size of the voltage data, but this method does not support dynamic online detection

However, it is often difficult to obtain accurate SOC models in practice. Most of the existing SOC models are polynomial models or exponential models obtained by data fitting, with large deviations.

Images