Nuclear power is one of the main players in the Swedish energy system with close to40% of the electricity production. As nuclear power needs to have a well- functioningsafety system for cooling the reactor and prevent contamination in case of anaccident, the components of this system must be maintained in a functioningcondition.The aim of this report is to find a system for predicting malfunction in the valveactuators in one of the safety systems in Forsmark nuclear plant. This is done byanalysing the curve of the consumed power generated during ordinary valve operationand testing to extract key values. Key values include time of the operation, middlepower, total power, peak value, standard deviation, frequency density between 0 and10 Hz and frequency density between 20 and 30 Hz. These values, in turn, are used intwo methods to identify faults. The first method is classification via machine learningand the second is a calculation of the Mahalaniobis distance.None of these methods proved useful in analysing the most commonly reported faultsin the valve actuators but some evidence suggests that an analysis of the Mahalaniobisdistance can be used to detect faults in the power converter and faults relating towear of the mechanical components.
| Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:uu-418070 |
| Date | January 2020 |
| Creators | Schultz, Egil |
| Source Sets | DiVA Archive at Upsalla University |
| Language | Swedish |
| Detected Language | English |
| Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess |
| Relation | UPTEC ES, 1650-8300 ; 20014 |
Page generated in 0.0017 seconds