<p>This paper proposes a reduced-order active disturbance rejection control (RADRC) framework integrated with an extremum seeking (ES) algorithm for diesel engine speed regulation under unknown time-varying moment of inertia and nonlinear uncertainties. The ES algorithm is introduced to learn the optimal control input gain online by minimizing tracking error, thereby circumventing the need to measure the actual moment of inertia. Based on the estimated optimal input gain from the ES algorithm, the RADRC composed of a reduced-order extended state observer is established, which estimates and compensates for the total disturbance including the coupled uncertainties caused by the ES algorithm. Moreover, the closed-loop stability of the proposed method is rigorously analyzed, accounting for the coupling between the ES dynamics and state-dependent nonlinear disturbances. The experimental results demonstrate the effectiveness of the proposed method.</p>