基于AdaBoost.RS算法的LF炉钢水温度预报分析

LF炉钢水温度的精准控制有利于缩短钢的冶炼时间,从而节约其生产成本。而获得准确的LF炉钢水温度预报是钢水温度控制的先决条件。通过分析LF炉冶炼过程对钢水温度的影响因素,提出一种适用于LF炉钢水温度预报同时具有增量学习功能的AdaBoost.RS集成建模算法。该算法引入松弛变量和遗忘因子2个参数,在提高预测精度的同时,可以克服大噪声数据带来的干扰,同时增量学习可以降低早期生产数据对模型的影响。以福建三钢有限责任公司100tLF炉为研究对象,采用5个测试函数验证算法的抗噪性能,分别用静态数据和动态数据对钢水出站的终点温度进行预报。实验结果表明,预测的绝对误差小于10℃的样本数量超过了样本总数的90%,算法精度较高,有利于实际生产应用。

Analysis of temperature prediction of molten steel for LF based on AdaBoost.RS

The accuracy control of molten steel temperature is benefit to save cost and time during ladle furnace (LF)refining.Thus the prediction of temperature on LF is the precondition for temperature control.A new algo-rithm with AdaBoost.RS function is proposed for LF temperature predition by analyzing the influence factors on molten steel temperature.The slack variable and forgetting factor are introduced to increase the accuracy of the soft sensor model and to fit the noisy industrial data.At the same time,the incremental learning can reduce the impact of historical data on the model.Five testing function are used to test the anti-noise ability of the algorithm. In addition,the dynamic and static data which collected from Fujian Sangang′s 100 t LF are used to predict the temperature of molten steel.The experiments demonstrate that more than ninety percent of the sample data′s ab-solute error is less than 10℃.The algorithm can be used in practical production with the high precision.