Parameter uncertainties, disturbance variations, and mis-matched disturbances significantly degrade the control performance of electro-hydrostatic actuators (EHAs), leading to impaired aircraft control surface effectiveness and even inducing surface flutter phenomena. These issues severely compromise flight maneuverability and safety while substantially increasing the challenges in controller parameter tuning for engineers. To address these limitations, the paper proposes a novel fuzzy Q-learning enhanced active disturbance rejection control (ADRC) algorithm. The methodology begins with model-based selection of control gains. A reduced-order linear extended state observer (LESO) is then implemented to estimate and compensate for aggregated system disturbances through feedforward compensation. Subsequently, a reinforcement learning framework employing Q-learning enables offline closed-loop adaptive tuning of position and velocity control parameters. The optimized Q-matrix is ultimately deployed for precision position control of the EHA. Simulation results demonstrate that, compared to conventional ADRC and sliding mode control (SMC), the proposed algorithm achieves more precise tracking under time-varying load conditions, with faster response, smaller overshoot, and eliminates the cumbersome manual parameter-tuning process.