Long Term Prediction For Generation Amount Of Converter Gas Based On Steelmaking Production Status Estimation

With the crisis of fossil energy resource, reasonable and effective usage of the byproduct energy is very important to reduce the operational cost and energy waste in steel industry. Linz Donawiz converter Gas (LDG) generated from steel making process is a kind of significant byproduct energy, which can be further used as the secondary fuel for several production processes, such as hot rolling, cold rolling and power generation, etc. However, the scheduling problem of the gas system is rather difficult since its generation amount exhibits frequent and large fluctuations with respect to the production status of steel making. Thus, a long-term prediction for generation amount of LDG that can provide effective guidance and save operational cost plays a very important role in reducing energy cost and environmental degradation.In this paper, a long term prediction approach for the generation amount of Converter gas based on steelmaking production status estimation is proposed, in which the steelmaking production status estimation consists of two stages, the feature extraction and the feature fusion. At the first stage, the generation flow data of LDG is divided into some data segments with same length, and then a template matching is used to extract the time and frequency domain characteristics of steelmaking production status. At the second stage, an improved version of fuzzy C-mean clustering method, which integrates the prior process knowledge and the clustering objective function, is developed for feature fusion. Specifically, the characteristics extracted from different data segments are assembled to obtain a universal feature of steelmaking production status. Finally, the universal feature is used to reconstruct the generation flow data of LDG.To verify the effectiveness of the proposed method, the real-world generation amount data from a steel plant are employed. Considering the industrial situation of missing data, the experiments of the long term prediction based on complete data and missing data are conducted respectively. The experimental results demonstrate that the proposed method exhibits high accuracy and robustness and can provide an effective guidance for balancing and scheduling the byproduct gas.