Low inertia is becoming an issue that cannot be avoided as renewable energy sources, represented by wind energy, become increasingly integrated into the power systems. Wind farms have taken on the role of supplying frequency support during system disturbances. In this case, the source of the measured signal for the wind farm frequency support becomes a critical factor for the control performance. The Center of Inertia (COI) frequency can better reflect the dynamic change of the system frequency after disturbance compared with the Point of Common Coupling (PCC) frequency. However, due to the time delay in signal communication and calculation, the direct utilization of the COI frequency may bring a detrimental influence on the performance of frequency support control, while the PCC frequency is measured locally with rare time delay. Therefore, how to combine both advantages of the PCC frequency and COI frequency into the auxiliary frequency control signal of wind farms requires specific attention. In this paper, a PCC-COI frequency based auxiliary frequency control strategy is proposed for the sake of enhance the frequency control performance of the wind farm by combining both the advantages of PCC and COI frequency signal. Firstly, the control effect of the auxiliary frequency of wind farms with either PCC or COI frequency as the control signal is compared, then the effect of lag time on wind farm auxiliary frequency control signals is investigated. Based on the analysis results, a delay correction method of the COI frequency signal is developed using the long and short-term (LSTM) neural network to minimize the detrimental influence of time lag in COI frequency. Finally, a new control strategy, which combines both PCC and the time-delay corrected COI frequency as its control signal, is proposed. In the modified IEEE 39 bus system, the efficacy of the suggested control strategy is validated.