Fault detection and diagnosis for industrial processes based on clustering and autoencoders: a case of gas turbines

Abstract Industrial machinery maintenance constitutes an important part of the manufacturing company’s budget. Fault Detection and Diagnosis (henceforth referenced as FDD) plays a key role on maintenance, since it allows for shorter times and, in long run, to train predictive algorithms. The impact proper is reflected especially costly type industrial machine: gas turbines. These devices are complex, large pieces that cause considerable service disruption when downtime occurs. In effort shorten these disruptions establish basis development we present this paper approach FDD machinery, such Our exploits data generated by machine-learning based architecture, combining several algorithms with autoencoders sliding windows. proposed solution helps achieve early malfunctioning detection has been tested using real from working environments. order build our solution, first, analyze behavior turbine mathematical point view. Then, develop architecture capable detecting presents abnormal behavior. great advantage proposal (i) does not require existing data, which can be difficult obtain, (ii) limited processes specific time windows, (iii) provides crucial information monitoring staff, generating valuable further maintenance. It worth highlighting although exemplify turbines, tailored other problems complex variable duration could benefit aforementioned advantages.