An Optimal Scheduling Method for Converter Gas System Based on Dynamic Causal Network and Fuzzy C-means Clustering

Linz Donawitz converter gas (LDG) is a very important secondary energy in the process of steel production, which is of great significance to make full use of, so as to save energy and reduce waste in steel enterprises. In this study, a dynamic causal network and Fuzzy C-Means (CN-FCM) based scheduling method is proposed. To identify the causal relationship among the generation units, the storage ones and the consumption ones under different operating conditions, a dynamic causal network is established based on the transfer entropy (TE) of the process knowledge. Fuzzy C-means clustering algorithm is used to classify the scheduling solutions under different operating conditions, and the causal matrix is calculated as the input of the clustering model to improve the recognition efficiency of the model. Considering that the solutions might be affected by the change of the equipment loss, temperature and process environment, the feedback update strategy of the historical scheduling solutions is then introduced to ensure that the current scheduling solutions is more matching the actual working condition. Comparative experiments are conducted by employing the actual production data from a steel enterprise, and the results show that the proposed method is effective when facing the scheduling problem of the LDG system.