There is a problem that the fault traveling wave signal of the secondary side of a voltage traveling wave sensor cannot truly reflect the characteristics of the primary traveling wave of the power grid. Thus this paper proposes a high-precision detection method for a voltage traveling wave based on L1 regularization inversion. First, the nonideal transmission characteristics of the traveling wave sensor are analyzed. Then a method of multi-scale decomposition of the observed signal using wavelet packet transform and inversion of the signals in each frequency band is proposed to reduce the distortion error caused by the traveling wave sensor. Second, an L1 regularization constraint is introduced into the inversion model to describe its sparsity, so that the inversion results can better reflect the characteristics of real fault waveforms. Finally, FISTA is used to solve the problem iteratively, and the fault traveling wave signal can be obtained by linear superposition of the inversion waveforms of each component. The outcome of the simulations and experiments demonstrate that, compared with direct inversion, this method can achieve high-precision real measurement of a fault traveling wave in the time and frequency domains, and can obtain more accurate inversion results even in a weak fault and noise environment. This offers value for engineering application.