Lithium ion battery electric quantity prediction method considering aging condition

Abstract

The invention discloses a lithium ion battery electric quantity prediction method considering an aging condition. Input data comprises charge-discharge cycle times and internal resistance which are aging indexes of a battery, a five-dimensional data set is established, battery data are preprocessed through empirical wavelet decomposition, and empirical wavelet decomposition is used for separatingoverall degradation and local regeneration of battery capacity; a CNN-GRU network of each decomposition component is designed and trained, a relationship among five-dimensional data is extracted by the CNN, a historical relationship before and after the data is extracted by the GRU, and feature capture of the data is completed; weight allocation is completed by using an attention mechanism, more weights are allocated to the key features, and interference is reduced; and finally, the prediction results of all the models are combined and reconstructed to obtain a final electric quantity prediction result. According to the invention, the electric quantity of the lithium ion battery can be predicted, and the aging condition of the battery is considered, so that the network does not need to befrequently changed and retrained to adapt to the performance change of the battery after being used for a period of time.

Description

technical field [0001] The invention relates to a prediction method based on a CNN-GRU-ATTENTION network structure, especially considering battery aging. Background technique [0002] Energy shortage and environmental pollution are the major problems facing this century. Traditional gasoline vehicles and diesel vehicles have large exhaust emissions and pollute the environment, and the fuels they need are also non-renewable resources. Therefore, people have begun to pay attention to the electric vehicle industry. Countries such as the United Kingdom and France have announced that they will stop producing gasoline vehicles between 2025 and 2040. The power source of electric vehicles is the battery pack, and in order to ensure the driving performance of the car, the battery needs high performance. Lithium-ion batteries have the advantages of high energy density, low self-discharge rate, and long cycle life, and have been widely used in the field of electric vehicles. The sta...

AI Technical Summary

Problems solved by technology

SOC cannot be obtained by actual measurement, but can only be estimated by other measurable values ​​(such as voltage and current)

But in the actual discharge process, the discharge current is very unstable, and there are chemical reactions inside the battery that are difficult to predict, so it is a practical work to accurately estimate the SOC of the battery

[0003] At present, the prediction of battery SOC is almost always based on new batteries as an example, but battery aging is an inevitable process. Battery aging will cause the previous prediction model to fail to capture the discharge characteristics of the aging battery, and the prediction results will have large deviations.

However, SOC is an indicator of power, and inaccurate power indicators will lead to overcharging and overdischarging of the battery, which will further damage the performance of the battery and reduce the life of the battery.

Images