International Journal on Advanced Science, Engineering and Information Technology, Vol. 12 (2022) No. 1, pages: 23-31, DOI:10.18517/ijaseit.12.1.12448

Designing Expert System for Centrifugal using Vibration Signal and Bayesian Networks

Abstract

Centrifugal machines are crucial in the process of making sugar. It requires proper centrifugal maintenance so that the production process runs smoothly. Efficiency and productivity are, therefore, critical factors in producing high-quality sugar. Centrifuge damages may occur, suddenly causing huge losses. Therefore, predictive treatment is essential. To achieve this, vibration analysis is a reliable and easy method to determine the vibration levels of spinning machines such as centrifuges. Simply by attaching the sensor to the outside of the engine, the engine condition will be read. Unfortunately, not all employees understand how to read vibration measurement data. Even experts need time to analyze the vibration signal. Therefore, the purpose of this study was to design an expert system that diagnoses centrifugal vibrations using the Bayesian Network. The vibration analysis process will be employed in the network using a series of complex nodes and trained according to the reference spectrum analysis. The presence or absence of spectral lines is evidence of Bayesian Network input in updating information regarding centrifugal damage. The result shows that the Bayesian network method was successfully applied to diagnose centrifuges based on vibration data. The inputs in the form of 1X and 2X produced an Unbalance probability value of 90%, Misalignment 93.2%, Looseness 57.38%, Bearing 33.79%, Pulley 50.3%, and produce a centrifugal damage probability of 7.86%. Therefore, these values were the actual conditions of the vibration data.

Keywords:

Centrifuge; vibration analysis; vibration spectrum; Bayesian networks.

Viewed: 1220 times (since abstract online)

• Show fulltext