Multi Sources Information Fusion Based on Bayesian Network Method to Improve the Fault Prediction of Centrifugal Compressor

Abstract The centrifugal compressor is an important machine in the oil and gas industry, so fault prediction of these machines widely discussed literature. Several techniques can should be used compressors: vibration analysis, non-destructive testing techniques, operating parameters, other techniques. But particular cases, tools are inefficient for making a decision regarding combined diagnosis prediction. This paper presents contribution to utilizing multi-source information fusion by Bayesian network. data does not come from same source, but rather parameters. In addition, accuracy ability improved compared with use obtained analysis only or analysis. proposed method validated on BCL 406 type compressor. Furthermore, results showed effectiveness network approach gives more decision-making developed has effect predicting faults.