Extreme learning machine-based petrochemical equipment corrosion prediction method

Abstract

The invention relates to an extreme learning machine-based petrochemical equipment corrosion prediction method. According to the method, corrosion data of an oil refinery are monitored; a corrosion database is built; long-term accumulated corrosion data, adopted as samples, are preprocessed; with corrosion influence factors of PH, CL<->, H2S and NH3N selected as input and Fe<2+> and Fe<3+> selected as output, a three-layer neural network is built to train sample data; an extreme learning machine petrochemical equipment corrosion prediction model can be obtained; and the corrosion data monitored on site are inputted into the prediction model, so that the corrosion prediction values of the Fe<2+> and Fe<3+> can be obtained. With the extreme learning machine-based petrochemical equipment corrosion prediction method of the invention adopted, the relationship between the corrosion influence factors and corrosion results can be expressed excellently; and the corrosion situation of petrochemical equipment can be known according to the prediction values; and process parameters can be adjusted, so that the corrosion of the petrochemical equipment can be effectively controlled and prevented.

Description

technical field [0001] The invention relates to petrochemical equipment corrosion protection technology, in particular to a petrochemical equipment corrosion prediction method based on an extreme learning machine. Background technique [0002] Corrosion exists in various industries of the national economy. Corrosion has caused huge losses to the national economy, and the economic losses caused by corrosion account for 3%-5% of the country's total national economic output value this year. Petrochemical refining is in a high temperature and high pressure environment, which is toxic, harmful, flammable and explosive, and corrosion is particularly prominent in the petrochemical industry. In addition, with the high sulfur content of imported crude oil and the development of many oilfields in my country, the quality of crude oil is deteriorating day by day. The water content, sulfur content, salt content, heavy metal content and acid value of crude oil are all increasing, which a...

AI Technical Summary

Problems solved by technology

For example, traditional neural networks (such as BP algorithm) are prone to fall into local optimum, slow learning speed, and difficult to guarantee accuracy.

Images