This paper proposes an improved model-free adaptive control scheme for the unmanned operation of excavators, taking into account that the hydraulic system of excavators has strong nonlinearity, parameter uncertainty, time-varying load and high energy consumption of the excavator during operation. Firstly, the trajectory tracking problem of the bucket tooth tip of the excavator is transformed into a hydraulic cylinder piston rod expansion and contraction tracking problem by analyzing the mapping relationship between the drive space and the task space? Then, a quintic five times NURBS curve interpolation method is used to ensure continuity and smoothness of the trajectory, reduce computational complexity and minimise system instability during startup and shutdown. Furthermore, the improved full-format dynamic linearized model-free adaptive control algorithm is designed to realize the unmanned operation of the excavator and in order to reduce the energy consumption, a penalty term of the energy function is introduced into the input criterion function to realize the optimal energy consumption while ensuring the good tracking accuracy. Finally, an example simulation of the proposed scheme is carried out to verify the feasibility and superiority of the scheme.