In view of an improvement of the accuracy of remaining useful life (RUL) prediction of lithium-ion batteries, a prediction model of lithium-ion batteries has thus been proposed based on the improved particle swarm optimization (IPSO) as well as gated recurrent unit (GRU) neural network. Firstly, the optimization ability of PSO algorithm is improved by changing the inertia weight and the update rules of learning factors. Next, the parameter selection of GRU neural network is optimized by IPSO algorithm, with an IPSO-GRU model built. Finally, the accuracy of IPSO-GRU model is to be verified by using the experimental data of lithium-ion battery published by NASA. The experimental results show that compared with the single GRU model, the proposed IPSO-GRU model helps to reduce the capacity prediction error and effectively improves the RUL prediction accuracy of lithium-ion batteries.