The paper introduces the Multi-Strategy Improved Dung Beetle Optimization Algorithm (MDBO) to enhance the performance of the original Dung Beetle Optimization (DBO) algorithm. The DBO algorithm, inspired by the behaviors of dung beetles, is known for its robust optimization capability and fast convergence speed but suffers from low population diversity, susceptibility to local optima, and unsatisfactory convergence speed in complex optimization problems. To address these issues, the MDBO algorithm incorporates several improvements: 1. **Latin Hypercube Sampling (LHS)**: This method is used for better population initialization, ensuring a more uniform distribution of the initial population in the solution space. 2. **Mean Differential Variation**: A novel differential variation strategy is introduced to enhance the algorithm's ability to escape local optima by maintaining population diversity. 3. **Lens Imaging Reverse Learning and Dimension-by-Dimension Optimization**: A strategy combining lens imaging reverse learning with dimension-by-dimension optimization is applied to the current optimal solution to improve its quality. The effectiveness of the MDBO algorithm is evaluated through comprehensive performance testing on standard benchmark functions from CEC2017 and CEC2020. The results show that MDBO outperforms other classical metaheuristic optimization algorithms in terms of optimization accuracy, stability, and convergence speed. Additionally, the algorithm is validated through three real-world engineering application scenarios: extension/compression spring design, reducer design, and welded beam design, demonstrating its superior capability in solving complex engineering problems.The paper introduces the Multi-Strategy Improved Dung Beetle Optimization Algorithm (MDBO) to enhance the performance of the original Dung Beetle Optimization (DBO) algorithm. The DBO algorithm, inspired by the behaviors of dung beetles, is known for its robust optimization capability and fast convergence speed but suffers from low population diversity, susceptibility to local optima, and unsatisfactory convergence speed in complex optimization problems. To address these issues, the MDBO algorithm incorporates several improvements: 1. **Latin Hypercube Sampling (LHS)**: This method is used for better population initialization, ensuring a more uniform distribution of the initial population in the solution space. 2. **Mean Differential Variation**: A novel differential variation strategy is introduced to enhance the algorithm's ability to escape local optima by maintaining population diversity. 3. **Lens Imaging Reverse Learning and Dimension-by-Dimension Optimization**: A strategy combining lens imaging reverse learning with dimension-by-dimension optimization is applied to the current optimal solution to improve its quality. The effectiveness of the MDBO algorithm is evaluated through comprehensive performance testing on standard benchmark functions from CEC2017 and CEC2020. The results show that MDBO outperforms other classical metaheuristic optimization algorithms in terms of optimization accuracy, stability, and convergence speed. Additionally, the algorithm is validated through three real-world engineering application scenarios: extension/compression spring design, reducer design, and welded beam design, demonstrating its superior capability in solving complex engineering problems.