This study implements a fairness-aware algorithmic framework for machine learning-enhanced quota allocation that addresses systemic inequities in China's higher education admissions. The research formulates quota allocation as a constrained multi-objective optimization problem optimizing both efficiency and equity through demographic parity and equal opportunity constraints. The framework integrates ensemble learning (Random Forest, Gradient Boosting Trees, Neural Networks, and Support Vector Machines) with adaptive fairness optimization mechanisms and real-time corrective feedback loops. Empirical testing on 50,000 student records from 15 Chinese universities demonstrates remarkable performance improvements. The framework achieves 91.3% allocation accuracy compared to 72.4% for traditional quota methods and 84.7% for baseline machine learning approaches, while attaining a composite fairness score of 0.857. Regional inequities were reduced by over 50% across all geographical regions. Comprehensive ablation analysis validates the necessity of both ensemble learning and adaptive constraints for optimal performance. SHAP-based interpretability analysis reveals transparent decision-making patterns with academic performance contributing 58.5% to allocation decisions while providing systematic equity adjustments for underrepresented groups. Longitudinal analysis spanning 2019-2022 demonstrates sustained convergence toward equitable outcomes. Stakeholder satisfaction exceeds 85% across government, university, and student groups. Implementation case studies at four institutions show significant improvements in diversity measures and operational efficiency. This work establishes methodological foundations for algorithmic equity in education while providing evidence-based frameworks for policymakers pursuing educational equity advancement.