In order to reduce the energy consumption of cement clinker firing process and achieve green energy-saving production, a high-precision, strong and stable process model is urgently needed. However, in the process of modeling cement clinker firing process with data-driven modeling method, in addition to the objective difficulties of multivariable, nonlinear, large delay and strong coupling of the firing process itself, and the data collected from cement production sites can also make modeling more difficult due to its complexity, repeatability, and incompleteness., in order to make more effective use of the information contained in the data and obtain the cement clinker firing process model with higher accuracy and stability. This paper proposes a series of data preprocessing steps for the raw data, which can remove the redundant information in the data and make the modeling process more efficient and accurate. Data preprocessing includes time domain unification, abnormal data elimination, data filtering and principal component analysis.
Keywords Data preprocessing, abnormal data elimination, data filtering, principal component analysis, process modeling