An overview of the objectives and milestones of the REFIMAN project. The video features full contributions from all the experts involved in this research and innovation journey.

In this introductory video, Prof. Giorgio Dalpiaz (Scientific Coordinator / Lead Partner - University of Ferrara) illustrates the origins, scientific value, and strategic vision of the predictive maintenance platform.

In this talk, Dr. Enrico Armelloni (REFIMAN project fellow at the University of Parma) presents research focused on acoustic and vibrational analysis for predictive maintenance.

The study analyzes in detail the lubrication state of the punches on the Prexima 300 tablet press by IMA S.p.A. Using acoustic acquisition techniques in the audible and ultrasonic bands, acoustic signals were recorded and processed in both time and frequency domains.

The comparative analysis of spectrograms and RMS indices allows for precise diagnosis of both under- and over-lubrication of the components during machine operation.

In this video, Dr. Emanuele Voltolini (REFIMAN project fellow at the Universities of Parma and Ferrara) presents the hardware and software architecture of the data acquisition system developed for the monitoring platform.

The system is based on a Raspberry Pi unit connected to modular acquisition boards, capable of simultaneously managing up to 32 channels at a sampling rate of 48 kHz. To ensure maximum signal cleanliness and eliminate electrical grid interference, the hardware is powered by a UPS/battery system and temporarily disconnected from the power grid via relays during measurements.

The management of the entire measurement cycle is handled by Python software that controls the frequency and duration of the acquisitions, collects data from the ADC modules (up to 8 sensors per board connected via Ethernet cable), and automatically uploads it to the Cloud to make it accessible and downloadable for remote analysis.

In this video, Dr. Elisabetta Manconi presents the activities of the SITEIA Laboratory at the University of Parma regarding the development of a prototype hardware and software system for plant monitoring and predictive maintenance.

The platform enables the retrofitting of sensors onto industrial machinery not originally equipped for it, acquiring real-time vibro-acoustic, position, and temperature data. Composed of an acquisition board, a mini PC, power supply/protection systems, and communication modules, the device offers the flexibility to store data locally or transmit it to the Cloud, allowing analysts and Machine Learning algorithms to evaluate machine health remotely. The system was validated in the laboratory under harsh industrial conditions and field-tested at partner company Fava S.p.A., demonstrating high accuracy and robustness compared to commercial alternatives.

In this video, Dr. Claudio Passadore (REFIMAN project fellow) presents research activities dedicated to estimating and diagnosing the lubrication state of a Prexima 300 tablet press (IMA S.p.A.) using vibration signal processing.

The experimental setup features PCB triaxial accelerometers, encoders, and fiber optic sensors connected to a Siemens SCADAS Mobile acquisition system to measure acceleration and angular position. The study is based on the development of two distinct analysis algorithms:

• Global-level monitoring: estimating the average lubrication of the machine through signal filtering and feature extraction.
  
  

• Local-level monitoring: evaluating the lubrication conditions of each individual punch using signal separation algorithms, filtering, and calculation of correlated features.
  
  

Experimental tests demonstrate that both approaches effectively identify clear and precise transitions between active lubrication periods and non-lubricated phases.