To reduce hydrogen consumption by hydrogen fuel cell vehicles (HFCVs), an adaptive power-following control strategy based on gated recurrent unit (GRU) neural network operating condition recognition was proposed. The future vehicle speed was predicted based on a GRU neural network and a driving cycle condition recognition model was established based on k-means cluster analysis. By predicting the speed over a specific time horizon, feature parameters were extracted and compared with those of typical operating conditions to determine the categories of the parameters, thus the adjustment of the power-following control strategy was realized. The simulation results indicate that the proposed control strategy reduces hydrogen consumption by hydrogen fuel cell vehicles (HFCVs) by 16.6% with the CLTC-P driving cycle and by 4.7% with the NEDC driving cycle, compared to the conventional power-following control strategy. Additionally, the proposed strategy effectively stabilizes the battery’s state of charge (SOC).