Summary:

The invention discloses a method for assessing the state of a battery based on autoencoder, comprising S1 obtaining an online operating parameters of the battery, and obtaining a voltage-capacity curve of the battery based on the operating parameters;S2 selects N capacity points in the voltage interval corresponding to the voltage-capacity curve of the battery as the input features of the model according to the pre-set voltage interval for feature extraction;S3 builds an autoencoder model based on the input characteristics of the battery and trains the autoencoder model using all the data of the battery from the initial state to the current state; S4 obtains a reconstruction error based on the inputs and outputs of the trained autoencoder model and evaluates the aging state of the battery based on the magnitude of said reconstruction error.This invention does not require a complete charge/discharge cycle of the battery to determine capacity. On-line testing of the battery voltage and current, selecting a feature input model, and calculating the reconstruction error of the model is all that is required to assess the aging state of the battery.

Link: http://epub.cnipa.gov.cn/patent/CN113076689A?8kt2YOWWXQBD=1709537553940