+386 1 47 71 704 ventil@fs.uni-lj.si

Using Eigenresponse Fuzzy Clustering in a minimal-invasive supervisory system

6. July, 2024 | Science and expert articles

Žiga Stržinar, mag. inž., Institut Jožef Stefan, Ljubljana in Univerza v Ljubljani, Fakulteta za elektrotehniko; dr. Boštjan Pregelj, univ. dipl. inž., Institut Jožef Stefan, Ljubljana; prof. dr. Igor Škrjanc, univ. dipl. inž., Univerza v Ljubljani, Fakulteta za elektrotehniko

Abstract:

Effective supervision of production processes is critical to ensuring product quality, maximizing production asset utilization, and improving overall performance metrics. However, the cost of implementing control systems can be prohibitively high, with such systems typically only being available on mass-produced machines. As a result, control systems are often unavailable on unique machines, which represent the majority of production assets. In order to promote wider use of control systems, our research focuses on utilizing minimally invasive or existing measuring equipment to enable control system implementation on unique machines. Specifically, we concentrate on repeatable industrial processes, with a particular emphasis on batch production. We propose a sequence of tasks that enable real-time detection of anomalies during operation. In this paper, we present a method for time series classification called Eigenresponse Fuzzy Clustering, which uses a condensed representation of each class with a limited number of prototypes known as Eigenresponses. We demonstrate the effectiveness of this method in detecting the actuation of various pneumatic actuators.

Keywords:

Time series analysis, Time series classification, Fuzzy systems; Event detection

Science and expert articles

Z nadaljno uporabo te strani se strinjate z uporabo piškotkov. Več informacij

Spletna stran uporablja piškotke z namenom zagotavljanja boljše uporabniške izkušnje. Z nadaljno uporabo te spletne strani oz. s klikom na gumb "Se strinjam", se strinjate z uporabo piškotkov.

Zapri