A kind of SOC estimation method of lithium iron phosphate battery pack

Abstract

The invention provides a method for estimating SOC of a lithium iron phosphate battery pack using a BP neural network. The BP neural network adopts a hidden layer, three inputs, and one output networkstructure, and the number of nodes of the hidden layer is 11. The estimation method comprises the following steps: adjusting a connection weight adjustment link of the BP neural network, inserting aninertia coefficient, using smooth weighted calculation to improve the connection weight adjustment, increasing an original connection weight item when performing the connection weight adjustment calculation, and giving a heavier weight. The method uses the current integral correction method to change the parameters of the BP neural network, which reduces the influence of the capacity decay of thelithium iron phosphate battery pack on the SOC estimation accuracy.

Description

technical field [0001] The application belongs to the technical field of electrochemical energy storage monitoring, and relates to a method for estimating the SOC of a lithium iron phosphate battery pack, in particular to a method for estimating the SOC of a lithium iron phosphate battery pack based on a BP neural network. Background technique [0002] During use, the remaining power of the lithium iron phosphate battery pack directly affects the decision-making of energy dispatch management, while overcharging or overdischarging directly affects the service life of the lithium iron phosphate battery pack. ofCharge, SOC) has become an important reference index for power management. Since the SOC estimation of the lithium iron phosphate battery pack involves complex electrochemical reactions, the voltage and current of the energy storage lithium iron phosphate battery pack are constantly changing under working conditions, and the accurate estimation of the SOC has become a he...

AI Technical Summary

Problems solved by technology

[0008] These prediction methods in the prior art do not take into account the complex nonlinear characteristics of the battery, and need to train a large number of sample data. The efficiency is low in the area where the SOC changes relatively flat, and the accuracy is low in the area where the SOC changes.

Images