Power battery inconsistency quantification method based on battery monomer voltage value

Abstract

The application discloses a power battery inconsistency quantification method based on battery monomer voltage value, comprising the following steps: obtaining real-time reporting real vehicle operation data of electric vehicles stored in the cloud; extracting three types of feature data of vehicle motion state features, power driving system features and battery electric signal features from the real vehicle operation data, obtaining measured battery monomer voltage true value, and selecting a discharge data segment; establishing a neural network fusion model, determining a standard discharge data segment, training the neural network fusion model by using the three types of feature data and the battery monomer voltage true value until the model is well trained; dividing the to-be-tested discharge data segment into multiple subsegments, inputting the three types of feature data into the neural network fusion model, and calculating battery monomer voltage estimated values of the subsegments; comparing the voltage estimated values with the voltage true values, constructing a battery monomer inconsistency scoring matrix, and quantitatively evaluating the power battery inconsistency by using the inconsistency scoring matrix.

Images