高炉透气性指数的核超限学习机预测模型

Prediction model of kernel extreme learning machine for permeability index of blast furnace

高炉透气性指数反映了高炉内煤气流运动受到阻碍的大小,是操作人员判断高炉运行状态的重要依据.本文针对超限学习机的缺点,提出了基于核超限学习机的高炉透气性指数预测模型.首先选取了适当的高炉参数作为模型的输入.其次采用小波变换对生产数据降噪处理.然后建立基于核超限学习机的高炉透气性指数预测模型.在建模过程中,探索了不同的核函数对模型性能的影响,并对相关参数寻优.最后进行仿真实验,同其他算法对比.实验结果表明,相比于传统算法,基于核超限学习机的高炉透气性指数预测模型训练速度更快,预测精度更高,预测结果更稳定.

The permeability index of blast furnace reflects the strength of the obstruction of the gas flow in the blast furnace,and is an important basis for the operator to judge the operating status of the blast furnace.Aiming at the shortcomings of the extreme learning machine,this paper proposes a prediction model for permeability index of blast furnace based on the kernel extreme learning machine.First,this paper selects appropriate parameters of blast furnace as the input of the model.Second,the wavelet transform is used to reduce the noise of production data.Then,the prediction model for permeability index of blast furnace based on the kernel extreme learning machine is established.During the modeling process,the influence of different kernel functions on the performance of the model is explored,and the relevant parameters are optimized.Finally,a simulation experiment is carried out to compare with other algorithms.The experimental results show that compared with the traditional algorithms,the prediction model for permeability index of the blast furnace based on the kernel extreme learning machine has faster training speed,higher prediction accuracy and more stable prediction results.