The Adaptive Disturbance Rejection Control (ADRC) is widely used in nonlinear systems such as Permanent Magnet Synchronous Motors (PMSM) due to its adaptability, strong robustness, and ease of implementation. However, traditional ADRC performs poorly in high-order nonlinear systems, as its performance mainly depends on the nonlinear function representing the total disturbance. To address this issue, this paper proposes an improved ADRC based on the Compensation Function Observer (CFO). The CFO belongs to the type III system, exhibiting significantly improved signal processing performance compared to the type I Extended State Observer (ESO), and it can attenuate the influence of the nonlinear function in the compensation stage. Additionally, in order to alleviate the difficulty of parameter tuning in ADRC, fuzzy control is introduced in the error feedback loop. The feasibility of the proposed method is verified by establishing a Matlab/Simulink simulation model. The simulation results demonstrate that, compared to traditional ADRC, the new ADRC exhibits better response speed and disturbance rejection for common signals.