A Power Battery SOC Estimation Method Based on Robust Unscented Kalman Filter Against Outliers

Abstract

The invention discloses a method for estimating the SOC of a power battery with robust unscented Kalman filtering against outliers, belonging to the technical field of power batteries, comprising the following steps: designing the state and observation equation of the power battery by using a composite model method combined with an ampere-time method, Determine the model equation of the on-board battery, establish the battery equivalent model; carry out model parameter identification, recursive least squares method to identify the relevant parameters of the battery model observation test equation, the input of the system is continuous excitation, and the number of identification iterations makes the final result converge and tends to be stable; the battery SOC is estimated by using an improved anti-outlier robust unscented Kalman filter algorithm. The present invention corrects the measurement error model to a normalized contaminated normal distribution model, and uses Bayesian theorem to calculate the posterior probability of occurrence of outliers as a weighting coefficient to adaptively adjust the measurement prediction correlation variance and gain matrix, which can effectively overcome outliers The problem of interference.

Description

Technical field[0001]The present invention relates to a power battery SOC estimation method, in particular, to a power battery SOC estimation method of anti-wild rod elsewhere, belonging, belonging to the field of power battery technology.Background technique[0002]In the past, the power battery-loaded state SOC estimation typically assumes that the measurement noise is a normal random sequence, but in practical applications, observation sequences due to errors or environmental interference in the measurement device itself or the data transmission process. Include some erroneous observations, the engineering domain is called wild values. SOC cannot be measured directly, and can only be indirectly estimated by measuring other status quantities of the battery. In this case, if there is a wad value to produce a more serious impact on the system, the accuracy and stability of the filter will be significantly dropped, and it is likely to cause the filter to be diverged when it is consecut...

AI Technical Summary

Problems solved by technology

[0004] The main purpose of the present invention is to provide a robust unscented Kalman filtering power battery SOC estimation method against outliers, which combines the normalized contaminated normal distribution model, Bayes' theorem and the introduction of suboptimal fading factors The power battery SOC estimation of the tracked unscented Kalman filter algorithm mainly solves the problem of battery SOC estimation outlier interference

Images