Multi-objective Flexible flow shop scheduling with limited buffers based on model-driven differential evolution algorithm

The multi-objective flexible flow shop scheduling problem with limited buffers (MO-FFSS-LB) is widely encountered in modern manufacturing systems. However, the existence of limited buffers greatly increases the complexity of job scheduling and needs to consider job completion time, replacement costs of different recipes, and dynamic buffer balancing simultaneously. To address the challenge of MO-FFSS-LB, this paper proposes a model-driven differential evolution (MDE) algorithm. Firstly, it designs a multi-layer coding mechanism to represent the solution of the problem according to the characteristics of the scheduling problem and then introduces an initialization mechanism to improve the quality of the initial solution and the solving efficiency. Moreover, the proposed MDE utilizes a machine learning-based model trained on historical scheduling data to approximate fitness, enabling rapid population evolution. To further improve the convergence speed and solution quality, a neighborhood search-based local optimization strategy and dynamically adjusted crossover rates and mutation factors strategy are introduced, enhancing the algorithm’s adaptability to complex scheduling problems. Experimental results demonstrate that the MDE algorithm can generate high-quality solutions for the MO-FFSS-LB and outperforms existing traditional scheduling algorithms in several performance metrics. This research not only provides an effective optimization tool for solving complex scheduling problems with limited buffer constraints but also offers new insights and methods for theoretical research and practical applications in related fields.