<p>Environmental changes in dynamic multi-objective optimization problems (DMOPs) require higher demands on an algorithm to adapt and search more effectively. This study presents a dynamic multi-objective evolutionary algorithm based on multi-level prediction and elite individual mutation strategy (MPEIS) to quickly track the Pareto Set (PS) and Pareto Front (PF) under new environments. This algorithm includes components for detecting environmental change, integrating response mechanisms, and static multi-objective optimization algorithms. First, this algorithm employs fast non-dominated sorting to divide the population into multiple levels and performs hierarchical predictions for different population levels. This multi-level prediction strategy allows the algorithm to adapt to environmental changes and track the latest PF quickly. Second, the algorithm introduces a diversity maintenance mechanism based on elite individuals to address the problem of maintaining population diversity. The algorithm balances population convergence and diversity by mutating specific points like Knee and boundary points and introducing elite individuals from previous environments through a memory mechanism. This proposed mechanism aims to keep the population distributed across multiple regions of the solution space, thereby enhancing the algorithm’s global search capability. The effectiveness of the algorithm is compared with five advanced algorithms. Extensive qualitative and quantitative experiments demonstrate that MPEIS can quickly track the PF under new environments while balancing population convergence and diversity. The algorithm was tested on the dynamic welding beam design problem to verify its effectiveness further, leading to superior performances compared with the other five algorithms. Based on multi-level prediction and elite individual mutation, the dynamic multi-objective evolutionary algorithm effectively solves complex DMOPs in real-world applications and holds broad prospects and practical significance.</p>