Start-up vibration analysis for novelty detection on industrial gas turbines

Abstract

This paper focuses on industrial application of start-up vibration signature analysis for novelty detection with experimental trials on industrial gas turbines (IGTs). Firstly, a representative vibration signature is extracted from healthy start-up vibration measurements through the use of an adaptive neuro-fuzzy inference system (ANFIS). Then, the first critical speed and the vibration level at the critical speed are located from the signature. Finally, two (s- and v-) health indices are introduced to detect and identify different novel/fault conditions from the IGT start-ups, in addition to traditional similarity measures, such as Euclidean distance and cross-correlation measures. Through a case study on IGTs, it is shown that the presented approach provides a convenient and efficient tool for IGT condition monitoring using start-up field data.