LONG-TERM INTERVAL PREDICTION FOR STEEL COAL GAS SYSTEM AND STRUCTURE LEARNING METHOD THEREFOR

摘要

Long-term interval prediction for steel coal gas system and structure learning method therefor, belonging to the field of information technology. The method employs real industry data: first, constructing a multi-layer information granularity unequal length distribution structure, and establishing a corresponding optimization model; next, considering the importance of model structure to prediction accuracy, carrying out strengthening learning on structural parameters of the multi-layer model by means of a Monte Carlo method; finally, computing a structure on the basis of optimal multi-layer granularity, and using a parallel computing policy to obtain a long term interval prediction result for the yield and consumption of coal gas. The accuracy of a result obtained by the present method is relatively high, and the computing efficiency meets actual application requirements; in addition, the method may be popularized and applied in the field of steel for other energy medium systems.