Method and device for predicting powder battery SOC on basis of improved I-ELM

The invention discloses a method and device for predicting the powder battery SOC on the basis of improved I-ELM. The method includes the steps of collecting training samples, normalizing charge and discharge data and SOC data of a power battery, inputting the training samples to an improved I-ELM network to be trained, and determining network model parameters; establishing an improved I-ELM network model according to the network model parameters, normalizing the charge and discharge data, collected on site, of the power battery, inputting normalized data to the improved I-ELM network model, and determining the charge state of the power battery, wherein the improved I-ELM network is a network where bias is added to a hidden layer output matrix on the basis of the I-ELM network. The method has the advantages of being high in learning speed and higher in prediction accuracy. The battery SOC can be accurately predicted, battery loss caused by repeated charge and discharge is reduced, and the service life of the battery is prolonged.