Wyniki wyszukiwania - BazTech

Ograniczanie wyników

1 2025

Znaleziono wyników: 1

Liczba wyników na stronie

Wyniki wyszukiwania

Wyszukiwano:
w słowach kluczowych: konserwacja oparta na danych

Sortuj według:

Ogranicz wyniki do:

Identification of the causes of production equipment failure using machine learning methods a case study

Rojek Izabela, Jasiulewicz-Kaczmarek Małgorzata, Piechowski Mariusz, Mikołajewski Dariusz

Zeszyty Naukowe. Organizacja i Zarządzanie / Politechnika Śląska

2025

z. 215

543--556

Purpose: This paper aims to present the possibility of using decision tree (DT) to increase the efficiency and effectiveness of maintenance activities by identifying the probable cause of failure based on historical data. Design/methodology/approach: This study used classifiers based on General Chi-square Automatic Interaction Detector (CHAID) and random forests. Using this group of classifiers brings with it faster u performance, the possibility to process symbolic data directly, and the possibility to add a tree as part of interactive tree building. A separate tree was built for each input parameter to aggregate the results from both trees by considering them together. The proposed solution also analyzes the importance of features (input data). Findings: Based on the research conducted, we have shown that using ML techniques can improve the accuracy of decisions regarding the type of maintenance work that should be carried out to efficiently and effectively remove failures and reduce losses caused by machine downtime. Research limitations/implications: The research is worth extending to use other novel artificial intelligence methods to compare the developed models. A limitation was the amount of data. As new data becomes available, the developed models should be trained to respond to the new data and better adapt to it. Practical implications: Relatively simple AI-based solutions such as CHAID and random forests have yielded fairly high accuracy with very short execution times. Within edge processing, this fulfills the complex trade-off between accuracy and speed in predictive maintenance applications. The presented families of simple algorithms should be developed as a transparent source of opinion for industrial decision-making processes. Originality/value: What is new is the automation of maintenance activities by identifying the probable cause of failure using AI methods. The solution is aimed at company employees who diagnose the causes of failure, ultimately improving the accuracy and speed of diagnostics and service response.