MECAPRES 4.0: Advancing Predictive Mechatronic Design in Industry

The MECAPRES4.0 project, funded by the Basque Government through the ELKARTEK initiative, has concluded with outstanding outcomes for all involved parties.

For years, the industry has sought machines with high dynamics, precision, and reliability. Yet, machines evolve over time, facing wear, adjustments, and material aging, posing a technological challenge to maintain high speeds without sacrificing precision.

Over the past two years, MECAPRES4.0 has focused on predictive mechatronic design technologies for smart industry applications. Unlike traditional approaches, this project not only examined the dynamic evolution of machines but also explored transferring these insights into the mechanical design and control phases.

Led by IKERLAN, with contributions from Tekniker, Ideko technology centers, Mondragon Goi Eskola Politeknikoa, UPV-EHU’s Department of Computer Science and Artificial Intelligence, and the Machine Tool Research Foundation (INVEMA), MECAPRES4.0 has achieved remarkable milestones:

• Agile methods to determine the non-ideal behavior of dynamic mechanisms over time.
• Models for replicating the dynamic response of hydraulic actuators.
• Adaptive intelligent control for machines, enhancing efficiency and robustness.
• Virtual models of plants, equipment, and critical components, mimicking operational functionality and degradation.
• Resilience capabilities to address unforeseen changes.
• Algorithms for estimating cutting point displacement in machining.
• Mechatronic drive models for simulating failure modes and generating synthetic data for AI.
• Predictive models for clearance in spindle benches using experimental signals.
• Simulation models for piezoelectric and voice coil actuators.
• Approaches for modeling clearance and friction in drives.
• Compensation strategies for clearance and friction.
• Hybrid model for random discharge detection in drone batteries.
• Methodology for State of Health (SOH) detection in drone batteries.
• Models for anomaly detection and/or characterization in complex industrial data.
• Methodologies for novelty detection in model training.
• Efficient Deep Neural Network (DNN) architecture search methods.

MECAPRES 4.0 signifies a leap forward in predictive design, empowering industries to anticipate and address challenges proactively.

Original source AFM