Fault Detection and Health Management for Machines of Industry 4.0 Using SVM and FTASL-LSTM

K. Kandhammal and S. Duraisamy

Diagnosis of fault and prognosis in the Cyber-Physical System (CPS) has been developed and tested at a very rapid rate in the last few decades. However, due to the high difficulty of computer centers, research into optimizing the reliability of fault assessment, and prognosis through data analysis remains a prominent problem in this sector. The research explores fault diagnosis and prognosis in computer centers based on data mining methods such as Support Vector Computer (SVM) and Fine-tune Threshold and Adapted Softmax Layer (FTASL) Long Short-Term Memory (LSTM) to design a systemic method and acquire analytical technology information in the Industry 4.0 era and the proposed system called SVM-FTASL-LSTM. This analysis is composed of two phases. Analysis of the data collected from CPS sensors, SVM is used to classify key factors that affect the data dimension to the less computational effort. The process engineers should be aware of which factors cause the system to malfunction. The second option is to set up a time series process focused on the FTASL-LSTM network for realtime analysis of the system. Before the computer collapses, the Fault Detection and Classification System (FDC) using SVM-FTASL-LSTM warn the engineer to change the computer settings to prevent the failure of the device.

Volume 12 | 05-Special Issue

Pages: 855-863

DOI: 10.5373/JARDCS/V12SP5/20201826