Aiming to address the challenges of strong nonlinearity and model uncertainties in the air intake system of a high-altitude flight environment simulation experimental facility during engine transient states — which are further complicated by unknown disturbances and measurement noise, making it difficult to achieve rapid, stable, and precise control — this paper proposes an active disturbance rejection control (ADRC) method based on the fusion of an improved sliding-mode extended state observer (ESO) with a Kalman filter. First, a variable-power reaching law is designed to accelerate state convergence. Second, a non-singular fast convergent sliding mode surface is constructed to enhance tracking error convergence, and its corresponding nonlinear convergence function is derived. Crucially, a collaborative estimation framework is established through this fusion. In this framework, the ESO actively estimates and compensates for the total disturbance, while the Kalman filter suppresses measurement noise in the feedback signal. Their synergistic operation enables the dual attenuation of both disturbances and noise. Based on this framework, the ADRC is designed. Complete stability analysis and simulation results are provided. The simulation experiment shows that the ADRC based on an improved sliding-mode ESO achieves higher tracking accuracy and faster convergence speed than the conventional ESO based ADRC under measurement noise. After fusion with the Kalman filter, the controller maintains its tracking performance while effectively suppressing control valve chattering in the air intake system.