For the problem that asynchronous sampling causes spectral leakage and reactive power calculation error when the power signal frequency fluctuates, this paper proposes a reactive power calculation method by Hilbert transform based on data preprocessing. It is that firstly, the improved Rife algorithm is used to estimate the real-time frequency, and then the interpolation operation is used to synchronize data, finally the Hilbert transform is used to get the accurate reactive power. Based on the measuring of signal frequency, for interpolation method, the Hermite interpolation algorithm achieves the ideal data synchronization effect by simulation comparison. Then the reactive power measurement method based on Hilbert is analyzed and arc furnace load simulation model is built as a reactive power source. Simulation results show that when the power signal fluctuates and contains harmonics the calculation accuracy of reactive power is mostly improved by using Hilbert method based on interpolation synchronization compared to conventional reactive power calculation method.