Since the global energy crisis is getting worse, obtaining energy from the ocean is now a trend. The marine current turbine (MCT) is the key to converting marine current energy to electric power. However, the operation of the MCT is threatened by seabed creatures (e.g., fish et al.) It is necessary to detect the impact of underwater MCT blades in time. Nevertheless, weak impact faults will be submerged in solid noise due to interference from turbulence and surges. Aim to this problem, and this paper proposes an impact fault detection method based on envelope spectrum analysis and PCA model. First, the single-phase stator current is collected, and Hilbert demodulates the current signal to obtain the envelope, which is then divided into a sample matrix. Therefore, PCA is used to reduce the dimension of the frequency-domain samples and calculate the corresponding fault detection limit according to the statistics. Finally, a 0. 23kW MCT prototype was built as an experimental platform to verify the effectiveness of the algorithm proposed in this paper and provides a new direction for the state monitoring and maintenance of MCTs.