The existing Grey Wolf Optimizer (GWO) suffers from slow convergence and susceptibility to local optima in error compensation for logistics sorting encoders. This paper proposes an improved Chaotic Differential-Lévy Grey Wolf Optimizer (CDL-GWO) to address these issues. Firstly, a hybrid initialization strategy integrating Logistic chaotic mapping and Gaussian perturbation is introduced to enhance the population diversity in the solution space, thereby mitigating the local clustering caused by traditional random initialization. Secondly, a nonlinear hybrid decay convergence factor is designed by combining exponential decay’s global rapid convergence characteristics with the local fine-tuning capability of cosine decay, effectively balancing the exploration and exploitation of parameters in the error compensation model. Furthermore, a dual enhancement mechanism incorporating differential evolution and Lévy flight is implemented: The mutation-crossover operations of differential evolution improve the algorithm’s ability to escape local optima. In contrast, an adaptive step-size Lévy flight strategy dynamically adjusts the search radius, enabling the compensation model to adapt to multimodal error characteristics in complex sorting scenarios.