Multi-layer Stacking Ensemble for Fault Detection Classification in Hydraulic System
Kyutae Kim, Jongpil Jeong · 2022
In the manufacturing plant, data is being collected in real-time through sensors, and fault and failure detection can be performed with the collected data. Recently, as the amount of data being collected increases and computer performance improves, attempts to detect defects and failures using artificial intelligence are increasing. In this paper, a multi-layer stacking ensemble model was proposed as a predictive model for the detection of manufacturing plant defect data through hydraulic system data collected through sensors in the manufacturing industry. The proposed model consists of five algorithms commonly used in machine learning and has a neural network structure by making several of these layers. Experiments are conducted using a hydraulic system dataset, and the proposed method shows that the classification performance is better than that of the existing stacking ensemble method.