Year 2001
Mikko Vermasvuori

On-line monitoring of process disturbances for the Outokumpu copper smelting process

The aim of this thesis was to develop a fault diagnosis system that detects process disturbances and equipment malfunctions in the Outokumpu copper smelting process.

Fault diagnosis has gained growing interest among researchers during recent years. Early detection of abnormal process conditions increases safety and decreases process downtime and the amount of production that does not meet the quality criteria.

In the literature part, different fault diagnosis methods are described on a general level, and the Kohonen Self-Organizing Maps (SOM) in more detail. Two cases where SOMs are used in industrial fault diagnosis are described.

The experimental part begins with a description of the copper smelting process. After that, the structure of the designed fault diagnosis system is described. Training of the SOMs and their ability to perform classification of different process states is described in detail. The experimental part ends with a description of the rules used in the system and the test results gathered during testing at the Outokumpu Harjavalta copper smelter.

The fault diagnosis system was tested in Outokumpu’‘s smelter in Harjavalta. Testing showed that the system structure worked and the SOMs and rules used gave satisfactory results.

The implemented fault diagnosis system can be used in different kinds of processes. Its most important capability is to detect process disturbances that are difficult to detect directly from measurement data. The prerequisite for using this system is that there is large amount of data including measurements from both normal and abnormal process states.

This info last modified 27 Apr 2024 - contact webmaster