Compared with conventional HVDC systems, the hybrid multiterminal HVDC (Hybrid-MTDC) system has a more sophisticated network topology composed of line commutated converter (LCC) and modular multilevel converter (MMC) stations, which leads to dc fault discrimination being difficult and ambiguous. Furthermore, for long transmission lines, high fault resistances, and different converter impedance characteristics, the reliable discrimination of external and internal faults located at the remote end and T areas is particularly demanding. Therefore, considering the frequency characteristics of dc boundaries and equivalent converter impedances and dc line models, this article constructs a time-domain convolution power. The performance of the convolution power in various fault scenarios is mathematically elaborated. Due to its good decomposition ability for short-time nonstationary signals, wavelet decomposition is utilized to extract the specific frequency bands (FBs) of traveling waves; thus, the convolution power in different FBs can be calculated. On this basis, a convolution power-based protection scheme is proposed for identifying external and internal faults, locating a dc bus fault, and selecting faulty poles. PSCAD/EMTDC simulation results show that various dc faults can be correctly identified within 3 ms, and the proposed scheme has high reliability and sensitivity, which is beneficial for the protection of Hybrid-MTDC systems.