Research on Higher Education Policy Effect Prediction Based on Multi-Agent Simulation Technology

Higher education policy formulation has long relied on empirical judgment and static data analysis, which is difficult to cope with the dynamic complexity of education ecology in the AI era. This study proposes a policy effect prediction framework based on multi-agent simulation technology, and simulates the multi-agent interaction and system evolution under policy intervention by constructing a higher education ecosystem model including students, teachers, departments, social needs and other agents. Based on the intelligent education practice data of Zhejiang University, Wuhan University and other universities, the policy causal reasoning algorithm is developed to realize the dynamic deduction of major adjustment, enrollment scale regulation and other policies. Compared to traditional expert evaluation methods that exhibit 2.3-year policy lag (MOE 2024 data), our framework reduces trial-and-error costs for discipline adjustments by 64% through digital twin campuses. The research provides interpretable and iterative decision support tools for higher education governance, and helps policy making shift from “post-correction” to “foresight”