Experts from FEEC are working on a project focused on predictive maintenance of machines

The Faculty of Electrical Engineering and Communication (FEEC BUT) has joined an ambitious project to develop an intelligent multisensor data collection system for predictive maintenance of industrial machines. The project, with total costs of approximately 38 million Czech crowns, represents one of the largest projects undertaken at the institution, both in terms of financial volume and interdisciplinary scope.

A total of 9 experts are involved in the project implementation, including 5 scientific research staff from FEEC BUT, 2 PhD students from FEEC BUT, one employee from the Faculty of Mechanical Engineering (FME), and one from the Faculty of Information Technology (FIT). This inter-faculty collaboration ensures comprehensive coverage of all technical aspects of the project, from sensor technologies through machine learning to software solutions.

The project runs for 2.5 years, specifically from July 1, 2024, to December 31, 2026. It is currently nearing the end of the initial implementation phase, with testing measurements already underway at the first two partner companies.

Project Goals and Current Implementation Status

The main goal is industrial research and experimental development of a prototype device for multisensor data collection intended for preventive machine maintenance. The system will evaluate machine behavior and predict maintenance needs or possible failures based on analysis of measured data, particularly vibrations and sound.

A key component of the solution is cloud software utilizing machine learning algorithms for evaluating collected data. Placing the algorithm in the cloud enables high computational capacity and complex data evaluation. Sharing data from similar or identical machine types will gradually allow for providing more accurate predictions for planning maintenance or component replacement.

The project is in an advanced preparation phase - cloud services, API protocols are already prepared, a neural network is pre-configured, and machine models have been created. Testing measurements at the first two companies are successfully underway and the number of participating companies is gradually expanding. The development team is also exploring the possibility of battery-powered device operation for easier installation.

Experience and Partnerships

The research team has long-term experience with vibrodiagnostics, acoustic diagnostics, signal processing, and machine learning for industrial monitoring. Although they have previously worked on several projects focused on predictive machine maintenance, the current project represents their most complex multisensor approach.

A key project partner is Meta IT s.r.o., with which the faculty has established cooperation based on long-term relationships and has already successfully completed 2 previous projects.



The project is funded by the Ministry of Industry and Trade of the Czech Republic through the Operational Programme Technologie a aplikace pro konkurenceschopnost (OP TAK).

 

Responsible person Ing. Zdeňka Koubová
Date of publication