Nuclear flexible manifold embedding electric smelting magnesium furnace fault monitoring method

Abstract

The invention relates to a nuclear flexible manifold embedding electric smelting magnesium furnace fault monitoring method. The method comprises the steps that various kinds of fault data and normal data are collected; offline modeling is carried out, a KICA monitor model is acquired according to the collected data, and a KFME fault diagnosis model is trained; online monitoring is carried out, the KICA monitor model is utilized for carrying out real-time online monitoring, and process data are collected; if faults are monitored, a KFME monitor model is used for carrying out fault recognition; if faults are recognized, a one-time fault monitoring process is finished; and if no fault is recognized, non-diagnosed fault data are analyzed and marked, and the step of training the KFME fault diagnosis model is executed through jump. By means of the method, a small quantity of marked data and a large quantity of non-marked data are utilized for building a process monitoring model, the model is more accurate, data dimension reduction and original data information retention can be achieved well, the accuracy and antijamming capability of the fault diagnosis model are improved, and unknown faults are monitored.

Description

technical field [0001] The invention relates to a fault monitoring and diagnosis technology for an electric fused magnesium furnace, in particular to a fault monitoring method for a nuclear flexible manifold embedded in an electric fused magnesium furnace. Background technique [0002] In my country, the production process of magnesium oxide is poorly automated, complicated, and has characteristics such as nonlinear, non-Gaussian, and strong coupling, and is prone to failure, resulting in huge economic losses. An effective process monitoring method is urgently needed Predict the production status, reduce the failure rate, and reduce economic losses. In recent years, with the development of computers and automation, the data in the production process can be preserved, and the safety of production has been paid more and more attention. A class of data-driven process monitoring methods using a large amount of historical data has received widespread attention and has become a re...

AI Technical Summary

Problems solved by technology

[0003] In the production process, experienced workers can ensure safe production and obtain high-quality products. However, it is difficult for ordinary workers to achieve such production effects. Therefore, how to use the experience and knowledge of experienced workers to realize industrial production monitoring , has become an important research direction

In recent years, a large amount of high-dimensional data has been obtained through a reliable data acquisition system. However, when using data to detect faults, the large amount of data and the large dimension have become important factors affecting the effect of process monitoring. When monitoring, reducing the data dimension has become a very important issue, which directly affects the effect of actual process monitoring

In the past few years, many process monitoring algorithms have been proposed, but these algorithms do not make full use of the experience and knowledge of old workers to realize the actual industrial process monitoring.

Images