Abstract: To address the increased difficulty of fault identification and fault location in four-terminal VSC-HVDC hybrid overhead-line/cable transmission lines caused by section parameter differences, reflection and refraction at junctions, and variations in fault resistance, a multi-model fusion-based fault identification and location method is proposed. The proposed method constructs feature sets for fault identification and fault location using only the positive- and negative-pole voltage and current waveforms at the sending end of line L1. In the fault identification stage, a cascaded fault identification model is constructed to identify the faulted line and classify the fault type; in the fault location stage, four base fault location models are established separately for the three fault types, and K-means is employed to divide the waveform feature regions. Dynamic fusion weights are then determined according to the regional errors on the validation set, thereby realizing dynamic-weight multi-model fusion fault location. Simulation results show that the fault identification accuracy is 100% under noise-free sample training and remains no lower than 98.67% under different test conditions after noise-augmented training. Under noise-free sample training, the MAEs of the dynamic-weight fault location model are 0.18 km, 0.15 km, and 0.19 km, respectively, outperforming the best single model and several global fixed-weight fusion methods. After noise-augmented training, under a signal-to-noise ratio of 20 dB, the mean absolute errors of the three fault types are all no greater than 0.30 km, indicating that the proposed method achieves high fault identification accuracy, high fault location accuracy, and good noise robustness.