Abstract: Hybrid energy storage microgrids can efficiently utilize distributed energy, but the power quality often lacks guarantee under perturbations. To this end, a deviation optimization strategy that considers both the higher-order disturbance components and the estimated model underachievement components is proposed to modify the existing observation process. First, based on the description extracted from the state space, two-dimensional variables are introduced in the observer to describe the disturbance, while the remaining disturbances are handed over to the cascade architecture. Subsequently, based on the derived internal disturbance model, a control law with all non-ideal components is established to achieve an accurate reconstruction of the overall disturbance. Afterwards, the intrinsic basis of the optimized strategy to improve the control quality and the typical relationship between adjustable parameters and operation characteristics are pointed out. Meanwhile, the controllability of the load-side converter in a given stability domain is demonstrated by Lyapunov theory. Finally, the effectiveness of the proposed strategy is verified by the multi-case curves obtained from the experimental platform with the capacity of 1.5 kW.