Cement chimney NOX prediction method based on multivariable time sequence deep network model

Abstract

The invention discloses a cement chimney NOX prediction method based on a multivariable time sequence deep network model. The method comprises the following specific steps: selecting 13 variables as input variables of cement chimney NOX prediction according to a cement NOX generation mechanism, a denitration process and an emission flow process, and uniformly carrying out normalization processing on a multivariable time sequence; then establishing a deep learning LSTM (Long Short Term Memory)-based feature reconstruction model to perform essential feature extraction of the NOX generation process according to features of a multivariable time sequence expressed in a time domain in the cement NOX generation process; establishing a cement chimney NOX prediction model (MT-LSTMs) based on a multivariable time sequence long-short term neural network (LSTM) according to the overall process flow of NOX of the cement chimney; determining the initial parameters of the model and performing forward training on the network model, then training the model by utilizing network dominant cost function error reverse fine tuning, and optimizing model parameters through error correction.

Description

technical field [0001] The invention relates to the field of cement chimney NOX prediction, in particular to a cement chimney NOX prediction method based on a multivariate time-series deep network model. Background technique [0002] my country is the world's largest cement producer, accounting for more than 60% of the world's total cement production. Since the localization of new dry process cement production lines in the 1990s, my country's cement production scale has expanded rapidly. The fuel calcination in cement factories is mainly coal-fired, causing the air to be polluted by gases such as nitrogen oxides or sulfur dioxide. With the continuous increase of coal burning, my country's air pollution is not optimistic. In order to alleviate the air pressure, the concentration of NOX emission in the cement industry has dropped from 800mg / Nm3 to 400mg / Nm3. From this, it can be seen that my country's NOX emission requirements for cement plants are becoming more and more stri...

AI Technical Summary

Problems solved by technology

At the same time, for strongly coupled and multi-dimensional data features, traditional feature extraction methods are difficult to extract the essential features of multivariate time series, which greatly reduces the accuracy of the prediction model

Images