The main reducer, a critical component in helicopter transmission systems, plays a pivotal role in transferring engine power to the rotor system while ensuring stable torque output. Its failure can lead to catastrophic accidents, substantial maintenance costs, and operational downtime, underscoring the necessity for robust health management strategies. To address the challenge, this paper developed a fault feature extraction method based on the Complex-Entropy Causality Plane (CECP) for the helicopter main reducers fault diagnosis. The CECP method could quantify nonlinear interactions and causal dependencies within vibration signals through entropy-based complexity metrics and phase space reconstruction. Utilizing fault-simulated experimental data and dynamic model simulation data across different health states of the main reducer, the fault features extracted from vibration signals exhibited obvious clustering patterns, verifying the effectiveness of this method in fault diagnosis and providing an efficacious method and tool for the health management and maintenance of the main reducer system.