Prediction of the Remaining Useful Life of Rotating Machinery using a Hybrid PSO-ANN Model

Rotating machines are very critical equipment in the manufacturing and industrial sectors. Unexpected failure of these types of equipment can result in huge maintenance costs. To avoid such implications, the Remaining Useful Life (RUL) of rotating machinery should be predicted. This paper proposes a Hybrid PSO-ANN model for achieving more accurate RUL prediction of rotating machinery bearings. The hybrid model is trained and tested using publicly available vibration monitoring data. Furthermore, the model uses time, RMS and kurtosis measurement values fitted with the Weibull distribution failure rate function from a state of bearing conditions as inputs, and life percentage as output. This was done to reduce the influence of the noise factors on the prediction execution of the model. The parametric examination was performed 36 times to evaluate the impact of parameter adjustment on the prediction performance and determine the most accurate model configuration. The results obtained show that the Hybrid PSO-ANN model can predict potentially and accurately the RUL of rotating machinery bearings.