基于LSTM的复杂炼化过程报警预测

In recent years, with the rapid development of the chemical industry and information combination, the data produced in the chemical refining system presents explosive growth. Alarm system is a kind of transmitting equipment abnormal state information to the operator of the system, but if the design is not reasonable, the equipment under abnormal state process may produce a large number of alarm and alarm saturation phenomenon, the serious influence the operator's information processing ability, thus increasing the probability of all kinds of industrial accidents. Therefore, how to mine useful information from the massive alarm logs is very important, and use the mined information to give positive guidance to the complex refining process. Deep learning is a method that can automatically learn and extract features from data. It does not require manual construction of complex and accurate physical and mathematical models, so it has been widely applied and paid attention to in the field of data prediction and classification.