针对现有元启发式算法求解矿井风量调控无约束优化数学模型存在收敛速度较慢的问题，提出了一种基于改进人工蜂群算法（ABC）的矿井风量按需调控智能决策方法。以矿井调节分支风阻为决策变量、各分支实际风量与需风量相符合为约束条件，以目标用风分支风量与理想风量差距最小为目标，建立了矿井风量按需调控智能决策模型；运用拉格朗日松弛方法优化模型的约束条件，采用冲突数方法优化模型的目标函数，利用随机搜索方法和启发式算法优化模型的搜索策略。针对人工蜂群算法（ABC）利用能力不足的问题，提出了一种改进ABC算法，并将其用于求解矿井风量按需调控智能决策模型。该算法在采蜜蜂局部寻优时引入群体历史最优解引导采蜜行为，并利用一般反向学习策略保存侦查蜂的搜索经验，良好地平衡了算法的探索和利用能力。实验结果表明：与粒子群优化（PSO）算法、ABC算法、基于全局最优的人工蜂群（GABC）算法和基于一般反向学习的人工蜂群（GABC−GOBL）算法相比，改进ABC算法能更加快速、稳定地求解出矿井风量按需调控最优方案，且风量调控精度可达0.49 m³/s。

To address the issue of slow convergence speed in solving the unconstrained optimization mathematical model of mine airflow control using existing metaheuristic algorithms, an intelligent decision-making method for mine airflow on demand based on an improved Artificial Bee Colony (ABC) algorithm was proposed. The decision variable was the mine ventilation branch resistance adjustment, with the constraint that the actual airflow in each branch matched the required airflow. The objective was to minimize the difference between the target branch airflow and the ideal airflow. A mine airflow on-demand control intelligent decision-making model was established. The Lagrange relaxation method was used to optimize the model's constraint conditions, the conflict count method was used to optimize the objective function of the model, and random search methods combined with heuristic algorithms were used to optimize the search strategy of the model. To address the issue of insufficient utilization capability in the ABC algorithm, an improved ABC algorithm was proposed and applied to solve the mine airflow on-demand control intelligent decision-making model. The algorithm introduced the population's historical optimal solution to guide the foraging behavior of the honeybees during local optimization and used a general reverse learning strategy to preserve the scout bees' search experience, which effectively balanced the exploration and exploitation capabilities of the algorithm. Experimental results showed that, compared with the Particle Swarm Optimization (PSO) algorithm, the ABC algorithm, the Gbest-guided artificial bee colony (GABC) algorithm, and artificial bee colony with generalized opposition-based learning (ABC-GOBL) algorithm, the improved ABC algorithm could solve the optimal solution of mine airflow on-demand control more quickly and stably, with an airflow control accuracy of up to 0.49 m³/s.