Real-time payload identification remains a critical challenge in robotic dynamic control despite advancements in identification accuracy. This paper proposes a wavelet approximation-based method to enhance the computational efficiency of inertial parameter identification for serial manipulators while maintaining high precision. Focusing on the UR5 industrial robot, the study integrates one-dimensional wavelet approximation with conventional dynamic modeling to develop a lightweight identification framework. The methodology encompasses optimized excitation trajectory design, wavelet-based parameter decoupling, and adaptive data processing strategies. Experimental validation demonstrates that the proposed approach achieves comparable accuracy to traditional least-squares-based methods (within 2.3% deviation in mass estimation) while reducing computational latency by approximately 40%. By preserving the critical dynamic characteristics through wavelet decomposition, the method effectively balances model fidelity and computational efficiency. These improvements address the pressing need for real-time parameter adaptation in applications requiring rapid environmental interaction, such as collaborative assembly and precision force control. The results substantiate the feasibility of wavelet-based approximations as a viable pathway for enhancing real-time performance in robotic dynamic identification systems, offering practical implications for next-generation industrial automation.