Method and system for product prediction in oil refining process based on variable weighted deep learning

Abstract

The present invention provides a method and system for predicting products in the refining process based on variable weighted deep learning. The method includes: obtaining the process variables in the debutanizer process, and based on the trained deep learning model, using the process variables as The input of the deep learning model obtains the product quality prediction value; wherein, the deep learning model includes at least three variable weighted autoencoders, and when training the deep learning model, every adjacent two variable weighted autoencoders In , the hidden layer feature data of the first variable weighted autoencoder is used as the input variable of the latter variable weighted autoencoder, and the latter variable weighted autoencoder is trained. Using multiple weighted self-encoders to stack into a deep network model can gradually obtain the relevant features of the deep output from the low level to the high level, strengthen the features related to the quality index, and provide accurate prediction values ​​for product quality, with high prediction accuracy and generalization Good sex and other advantages.

Description

technical field [0001] The present invention relates to the field of chemical technology, and more specifically, to a method and system for predicting products in an oil refining process based on variable weighted deep learning. Background technique [0002] Oil refining production is a complex process industrial process with multiple raw materials, multiple devices, multiple processes and multiple products. In oil refining production, after crude oil from different oil fields is evenly mixed, it passes through primary processing equipment such as initial distillation tower, atmospheric and vacuum distillation tower, secondary processing equipment such as catalytic cracking, hydrocracking, delayed coking, and catalytic hydrogenation, Fine production of tertiary processing equipment such as catalytic reforming and hydrofining, and finally obtain petroleum products such as gasoline, diesel, aviation kerosene, fuel oil, etc. A variety of intermediate petrochemical products, as...

AI Technical Summary

Problems solved by technology

[0003] The present invention provides a method and system for predicting oil refining process products based on variable weighted deep learning that overcomes the above problems or at least partially solves the above problems, and solves the problem that the deep learning model in the prior art only focuses on the feature representation of the process data itself and ignores the Feature extraction related to the output quality index, so that the correlation between the extracted features and the quality index cannot be guaranteed

Images