Learning Vector Quantization Based Vibration Analysis for Steam Power Plant Rotating Equipment Fault Diagnosis

Abstract

Sudden failure of rotating equipment in a steam power plant can cause a sudden electrical power outage and economical losses. Rotating equipment in a base load steam power plant operated in a non stop loading. Vibration data is able to be extracted from rotating equipment, which is used in vibration analysis to predict early symptom of rotating equipment fault. A Learning Vector Quantization (LVQ) based vibration analysis is able to give a quick and objective analysis based on its learning data. A normalized spectral data of rotating equipment vibration is used as input variables for the LVQ algorithm to perform vibration analysis. The normalized spectral data which is used as training and analysis data is based on literature and empirical data. In this study the LVQ model is used to learn and classify five failure symptoms in power plant rotating equipment. The proposed method can achieve 100% accuracy in three cases, unbalance, parallel misalignment, and cocked bearing. Angular misalignment and bent shaft have not been tested because no actual case happened. Additional training of LVQ algorithm will upgrade the knowledge base.