To solve the problem of high cost and long measurement time for low-frequency impedance spectrum measurement of lithium batteries, this paper proposes a low-frequency impedance segment prediction technology based on the grey wolf optimization deep neural network model (GWO-DNN). The low-frequency impedance segment is predicted by using the easy-to-measure medium-frequency Electrochemical Impedance Spectroscopy data, and the impedance segment is obtained quickly. The effects of different cycle numbers, temperatures, and state of charge on the system’s prediction results are discussed. Then the prediction effects of different algorithms are compared, with a maximum error of only 2.61%. Finally, the equivalent circuit parameter identification effects of the actual impedance spectrum and the predicted impedance spectrum are compared, and the results prove that the proposed prediction model has the advantages of high accuracy and strong applicability.