Multi-dimensional assignment model and its algorithm for multi-features decision-making problems

This paper proposes the multi-dimensional assignment model to address complex decision problems involving multi-agent features, multi-stages, resources, constraints, and more. Compared to traditional two-dimensional (2D) assignment models, this model can effectively describe assignment problems ranging from 2D to N-dimensional with multi-feature constraints. To tackle this model, we develop a dimensionality-reducible “Virtual Matching Algorithm” (VMA) based on ideas from Roulette-wheel selection, Particle Swarm Optimization (PSO) and Genetic Algorithm (GA). Taking organizational assessment and guavas production as real cases, the corresponding model is constructed and the performance of VMA is tested and optimized. In contrast to PSO, Binary PSO (Bi-PSO), GA, and hybrid GA-PSO, VMA demonstrates improved solving speed and the ability to find better solutions. These results exhibit the universality and effectiveness of our proposed model and algorithm.