A neural network-based system and process for prediction of power consumption in an air separation plant

摘要

The present invention relates to a system and process for prediction of power consumption in an air separation plant. The system (100)comprises at least one Air Separation Unit (102) for separating air in the air separation plant; at least one data means (104)for controlling, monitoring and gathering information of the air separation plant; and providing real-time control of operation parameters in the Air Separation Unit (102); and at least one Long Short Term Memory neural network module (106) for predicting total power consumption in the air separation plant. In particular, the Long Short Term Memory neural network prediction module is implemented in the air separation plant of the present invention enabling prediction of total power consumption of the plant. The Long Short Term Memory neural network module (106) further comprises (200) an input layer (201) for handling time series information from different types of gas production through received input parameters; a hidden layer comprising at least three layers for cross entropy learning; and an output layer (210) for predicting probability of plant production power through stochastic means. Figure 1