Fast and reliable DC fault identification is key to ensuring the safe and stable operation of multi-terminal flexible DC grids. To address the insufficient speed and reliability of existing protection methods under high transition resistance

lightning interference

and long transmission distances

we proposed a longitudinal protection method based on complementary ensemble empirical mode decomposition and weighted energy entropy. By extracting high-frequency features of transient fault current

a weighted energy entropy-based start-up criterion was constructed. A ratio-type differential regional criterion was developed by incorporating directional current energy to achieve rapid fault identification and section location. Considering the characteristics of bipolar DC faults and the susceptibility of overhead lines to lightning

we introduced the pole selection and lightning disturbance identification criteria. Moreover

time-frequency analysis was first conducted on internal and external fault traveling waves

followed by energy variation analysis before and after faults. A four-terminal flexible DC grid model was built in PSCAD/EMTDC for validation. Results show the method can be adopted to identify faults within 1.5 ms. Under a 300 Ω transition resistance and 20 dB noise

the criterion differences exceed 65.8% and 73.33%

effectively distinguishing 50 kA lightning. High reliability is also maintained under asynchronization of 500 μs data.