Machinery Equipment Early Fault Detection Using Artificial Neural Network Based Autoencoder

Machinery equipment early fault detection is still in an open challenge. The objective of this paper is to introduce a parametric method Artificial Neural Network based Autoencoder implemented to perform early fault detection of a machinery equipment. The performance of this method is then compared to one of the industry state of the art nonparametric methods called Similarity Based Modeling. The comparison is done by analyzing the implementation result on both artificial and real case dataset. Root Mean Square Error (RMSE) is applied to measure the performance. Based on the result of the research, both of these methods are effective to do pattern recognition and able to identify data anomaly or in this case is fault identification.