Abstract: The mental health challenges faced by college students are becoming increasingly complex, while the traditional college psychological counseling model has certain limitations. This study presents a novel technical framework that integrates psychology and artificial intelligence: by incorporating psychological counseling domain knowledge and data into the base model, a multi-agent collaborative system composed of three types of psychological counseling intelligent agents (evaluator, counselor, supervisor) and the college student intelligent agent is constructed. The system adopts a "double-loop" mode of "internal loop training - external loop service". In the "internal loop training" stage, the evaluator, counselor intelligent agents and the college student intelligent agent simulate the real counseling process through virtual scene interaction, and use the feedback from the supervisor intelligent agent to optimize the service strategy, accumulate personalized counseling files and multi-school intervention experience; in the "external loop service" stage, the psychological counseling intelligent agents provide professional and precise psychological assessment and intervention services for real college students based on the results of "internal loop training". The system is expected to become an effective auxiliary tool for college student psychological counseling and help the college mental health service system achieve intelligent upgrading.

Key words: intelligent agents, large language models, psychological counseling, college students