The fractional-order model of lithium-ion battery is used for battery aging research because it contains parameters that characterize the aging mechanism, and it's expected to realize the exploration of online battery aging process. The online identification of as many aging parameters as possible will promote the online applications of this model under limited operating conditions and product-level testing conditions of the battery. In this paper, a multi-parameter online identification method based on back propagation neural network is proposed. First, the parameter set of online identification is determined by analyzing the parameter sensitivity under typical operating conditions. Then, the network and training algorithm are designed based on the battery aging law to improve the identification speed and accuracy. At the same time, a verification method is designed to ensure the convergence of identification parameters. Finally, simulation and experimental results verified the identification speed and accuracy of the proposed online identification method.