OMICRON Magazine

Increase the relevance of your test data. Machine learning supports accuracy and exposes potential discrepancies in test results at an early stage. Have you ever questioned the quality of your test data? Each of us knows that data, especially reliable data, is an essential tool for our daily work. Particularly, a test performed correctly, the resulting test data, and precise evaluation of test results are critical for a reliable asset condition assessment. To name one example: Sweep Frequency Response Analysis (SFRA) is a powerful and sensitive method for checking the mechanical and electrical integrity of a power transformer. Slight deviations between the fingerprint and the measurement can lead to an incorrect conclusion about an asset’s condition. Therefore, validating the data upfront is crucial. HIGHER DATA QUALITY WITH ARTIFICIAL INTELLIGENCE The graphic shows a process of a reliable and trustworthy assessment. Layer 3 Assessed Data Layer 2 Valid Data Layer 1 Raw Data What causes us to question the quality of measurement data? There are enough reference values to which test results can be compared, such as Cigre data, fingerprint, or sister devices. However, the ever-accelerating energy transition is having a significant impact on the existing power grid infrastructure. Electrical energy is becoming decentralized due to renewable energy sources. Thus the number of feed-in points and nodes is increasing. The large inductive or capacitive loads in the power grid are being replaced by new regulated loads. As a result, the risk of dangerous effects such as short circuits, harmonics, and transients in the electrical power grid is increasing. Due to these changing asset conditions, checking carefully for possible measurement data changes is advisable. 40