【摘要】 为实现连铸坯有效定重预测，避免在连铸生产中产生质量缺陷和资源浪费，根据真实方坯生产数据，提出一种基于改进粒子群算法优化XGBoost的连铸定重预测模型。针对粒子群算法全局寻优精度较低等缺点进行改进，主要采用反向学习策略优化种群初始分布提高算法优化效率，根据进化状态自适应调整惯性权重，匹配粒子当前搜索状态，同时使用突变策略使粒子突破局部收敛。数值实验结果表明，改进算法收敛速度更快，精度更高，验证了良好性能。将其应用到XGBoost中优化模型超参数，连铸坯定重预测精度获得提升。更多还原

【Abstract】 To realize effective weight prediction of continuous casting slabs and avoid quality defects and waste of resources in continuous casting production, a continuous casting fixed weight prediction model based on improved particle swarm optimization algorithm optimized XGBoost was proposed using real billet production data. Among them, to improve the shortcomings of the particle swarm optimization algorithm such as low global optimization accuracy, the reverse learning strategy was mainly used to optimize the initial distribution of the population to improve the optimization efficiency of the algorithm, and the inertia weight was adaptively adjusted according to the evolution state to match the current search state of the particles. The mutation strategy made particles break out of local convergence. Numerical experiments show that the improved algorithm has higher convergence speed and higher precision, which verifies the good performance. It is applied to XGBoost to optimize the hyperparameters of the model, and the prediction accuracy of continuous casting slab fixed weight is improved.更多还原