基于核偏最小二乘的锌层重量预测模型

Forecasting Model for Zinc Coating Weights Based on Kernel Partial Least Squares

为了给带钢热镀锌生产的质量控制提供必要的决策支持和分析手段,针对气刀对锌层重量的控制工艺,提出了基于核偏最小二乘回归的锌层重量预测模型。利用核函数将低维空间的非线性回归转化为高维空间的线性回归,克服了实际生产工艺中非线性因素对预测模型的不利影响。应用鞍山钢铁集团公司带钢热镀锌的生产实际数据进行验证,结果表明,基于核偏最小二乘的锌层重量预测方法与线性偏最小二乘、BP神经网络等方法相比,具有更好的预测精度。

To provide the necessary decision supports and analysis tools for quality control of strip hot-dip galvanizing, using the airknffe parameters, a zinc coating weights forecasting model based on kernel partial least squares is proposed. In the forecasting model, kernel function is introduced to transfom, the nonlinear regression problem in low-dimensional space into the linear regression in a high-dimensional space, so as to avoid the negative intluence of nonlinearity. The real field data from strip hot-dip galvanizing production of Anshan Iron and Steel Corporation are used for validation. The results show that the regression model based on kernel partial least square has higher prediction precision than methods of linear partial least square or back propagation neural network.