Application of Machine Learning to Classify Wear Level of Multi-Piston Displacement Pump
2019
artykuł naukowy
angielski
- machine learning
- diagnostics
- signal analysis
- multi-piston pump
- vibrations
EN This article specifiesapplication of machine learning for the purpose of classifying wear level of multi-piston displacement pump. A diagnostic experiment that was carried out in order to acquire vibration signal matrices from selected locations within the pump body is describedherein. Measured signals were subject to time and frequency analysis. Signal attributesrelated to time and frequency were grouped in a table in accordance with pump wear level. Subsequently, classification models for the pump wear level were developed through application of Matlabpackage. Assessment of their accuracy was carried out. A selected model was subject to confirmation. The article includes its summary.
2019222-1 - 2019222-14
40
70