Determinant of nonlinear trait of silicon content in blast furnace hot metal

With thetime series (containing 2000 data points) of silicon content in No.7 blast furnace (BF) at Han Iron and Steel Group Co. as sample space, a quantitative surrogate data technique was used to examine the nonlinear characteristic in the time series. For the shortcoming of the quantitative method that data are vulnerable to the system noise and the autocorrelation degree of timing sequence, a further qualitative nonlinear discriminant was proposed according to the images of information-theoretic functional-redundancies (linear and nonlinear forms). The redundant image comparison between silicon constant and two typical data series (linear regression model and chaos Lorenz model) shows that the nonlinearity of silicon constant is not ordinary nonlinearity caused by noise, but the intrinsic nonlinearity decided by the blast furnace smelting inherent mechanism. The analysis provides firm rationale for the nonlinear prediction and control of furnace temperature.