以传统的物料平衡和热量平衡为基础 ,对炼铜转炉中最新的 30 0炉历史数据进行动态仿真 ,计算出吹炼过程中每隔 1min的熔体主要元素组成和熔体温度 ,然后对之进行四次曲线拟合 ,将拟合出的系数作为人工神经网络的输出 ,优化模块中优化出的操作参数和部分原始数据作为神经网络的输入 ,经过训练 ,使之能模拟吹炼过程中的炉况变化 ,达到炉况监控的目的。可以通过不断增加样本数 ,使网络进行自学习 ,并组合成为炼铜转炉炉况实时预报系统。该系统成功地应用于某厂的炼铜转炉中 ,2 0 0 0年 6～ 10月经过四个多月的试运行 ,炉况实时预报系统使操作人员按系统的预报值指挥生产 ,各项生产指标显著提高 ,粗铜产量提高 6 0 % ,冷料处理量提高 8% ,平均炉寿从原来的 2 13炉至少提高到 2 35炉 ,并为炼铜转炉实现在线控制提供了新思路 Based on the traditional material balance and heat balance, the historical data of the latest 300 furnaces in the copper smelting converter were dynamically simulated, and the main elemental composition of the melt and the melt temperature were calculated every 1 min in the blowing process, and then the The curve fitting is carried out four times, the fitted coefficient is taken as the output of the artificial neural network, the optimized operation parameter and some original data in the optimization module are input as the neural network, and are trained to simulate the process of the blowing process Changes in furnace conditions, to achieve the purpose of monitoring furnace conditions. By continuously increasing the number of samples, the network can be self-learning and combined into a real-time forecasting system for copper converter furnace status. The system has been successfully applied to a copper smelter converter in a factory. After more than four months of trial operation from June to October in 2000, the real-time temperature forecasting system enables operators to direct production according to the predicted value of the system. The production index increased significantly, the output of blister copper increased by 60%, the throughput of cold material increased by 8%, the average furnace life increased from at least 213 furnaces to 235 furnaces, and a new idea was provided for on-line control of copper smelters