铁水含硅量预测系统中数据预处理技术的研究

Study of a Data Preprocessing Used in a Mathematical Model Forecasting Silicon Content in Hot Metal

通过使用自适应预报模型与时差方法相结合的方法对铁水含硅量进行预测,模型所需的工艺参数的原始数据经采集后需使用OPC技术标准进行传输以存入数据库系统中;硅预测程序通过对数据库查询得到所需的操作数据,由于生产环境对数据的干扰,需对其作均值滤波和插值运算的一次处理,然后进行求平均值、梯度和标准差统计学运算的二次处理,之后运用模糊理论将各个工艺参数归一化以提取参数特征值,作为铁水含硅量预测模型的输入参数,在预测过程中使用多元线性回归不断修正模型中的权重系数,以提高数学模型预测的准确率。

Adaptive prediction model combined with the time difference method is used in the model of hot metal silicon content prediction.Required original data of model parameters collected after using OPC technology standard for transmission to store in the database system.Due to the interference of production environment,computing the data for the mean filtering and interpolation on the first process is needed.Then average,and standard deviation statistics gradient computation is needed as the secondary process。Using fuzzy theory to various process parameters normalization parameter extraction complex eigenvalues as iron the silicon content of the model input parameters.While the process,continuously correcting the weight coefficient through multiple linear regression is necessary.The operating result in improving the mathematical model.