Application of multi-agent co-evolution algorithm in dynamic matching system between vocational education and industrial demand

The Multi-Agent Co-Evolution Algorithm (MACEA) models enterprises, schools, and students as autonomous agents, each characterized by heterogeneous states, objective functions, and local adjustment rules. A dynamic co-evolutionary process is established through stable matching, fitness-triggered updates, and cross-agent market signals. The simulation experiment comprised 50 enterprise agents, 10 school agents, and 1,000 student agents, with each model subjected to 30 independent runs under uniform initial conditions. MACEA achieved an overall matching degree of 0.82 ± 0.03, outperforming the Static Matching Model (0.48 ± 0.05), the Genetic Algorithm-Based Global Optimization Model (0.71 ± 0.04), and the Decentralized Adaptive Baseline (0.67 ± 0.04). MACEA stabilized after approximately 78 rounds and demonstrated rapid recovery in matching following an external skill demand shock. Abandonment, parameter sensitivity, and agent-scale expansion experiments confirmed the stability of this performance pattern within the defined simulation range.