Kalman filtering and data driving fusion battery SOC estimation method

Abstract

The invention discloses a Kalman filtering and data driving fusion battery SOC estimation method belonging to the battery SOC estimation method technology field. The invention provides a varied variance Kalman filter least square support vector machine (VVKF LSSVM) fusion method. Based on two equations of a KF, a noise variance, which is adapted to a current system state to the greatest extent, is set during every iteration, and a problem of declined precision caused by Kalman filtering noise variance initial value relied on artificial experience setting is solved. A least square support vector machine (LS SVM) is selected as the measurement equation of the KF, and by starting from a data angle, the SOC estimation method suitable for various batteries is completed by establishing a simple sample library, and the estimation precision is improved by dynamic modeling. A part of data in an NASA lithium battery data set and a CACLE lithium battery data set is used for experimenting to prove the superiority of the VVKF by comparing with the KF, and the validity of the whole method on the lithium battery SOC estimation.

Description

technical field [0001] The invention relates to a method for estimating battery SOC combined with Kalman filtering and data driving, and belongs to the technical field of battery SOC estimating methods. Background technique [0002] People usually only observe the terminal voltage to avoid overcharging and overdischarging of the battery. As people's requirements for battery usage increase, the state of charge (State of Charge, SOC) is the main parameter of battery management just like the oil gauge of a car. . A battery is a device that converts chemical energy into electrical energy, so its state of charge is unmeasurable. Usually, parameters such as terminal voltage, operating current, and internal resistance correspond to SOC, so that SOC can be estimated through these physical quantities. At present, there are many battery SOC estimation methods, which can be divided into open circuit voltage method, ampere-hour method, internal resistance method, model-based method, da...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is in order to solve the problem that above-mentioned prior art exists, namely for the parameter setting that widely used Kalman filter (Kalman Filter, KF) and its derivation method exist at present have no specific standard, its model changes with working condition environment The problem of poor adaptability, and then provide a battery SOC estimation method based on Kalman filtering and data-driven fusion

Images