Special Session 19

Industry 4.0 requires for continuous monitoring of processes variables, parameters and working conditions. Measurement systems are a fundamental pillar of industrial process monitoring. Physical and economic constraints are often in contrast with efficiency and cost reduction. The possibility to adopt models, known as soft sensors (SSs), devoted to the estimation of process variables, represents a significant turning point.

SSs for the industry represent an interesting field of applications for machine learning techniques. Advanced methodologies like long short-term memory, stacked autoencoders, convolutional neural networks, reservoir computing, and bio-inspired learning techniques have recently been proposed to improve SSs behavior.

Different aspects still need to be investigated such as SS design for time-variant systems, feature and data selection, outliers detection, big/small datasets, choice of the model class, model validation and maintenance, model interpretability, and transfer learning.

This Special Session focuses on recent developments and challenges of SS design both from a theoretical perspective and industrial applications.

• Feature extraction
• Outliers detection
• Data selection
• Big and small datasets
• System identification
• Linear and nonlinear models
• Deep learning techniques
• Optimization strategies
• Recurrent neural networks
• Reservoir computing
• Bio-inspired learning techniques
• Model validation
• Soft sensor maintenance
• Transfer learning
• Model interpretability
• Sparse modeling
• Soft sensors for time-varying systems
• Industrial applications of soft sensors

Salvatore Graziani received the M.S. degree in electronic engineering and the Ph.D. degree in electrical engineering from the Università degli Studi di Catania, Italy, in 1990 and 1994, respectively. Since 1990, he has been with the Dipartimento di Ingegneria Elettrica, Elettronica e Informatica, Università di Catania, where he is a Full Professor of Electric and electronic measurement and instrumentation. His primary research interests lie in the field of sensors, polymeric sensors and actuators. He has coauthored several scientific papers and books.

Luca Patanè received the degree in Computer Science Engineering and the Ph.D. degree in Automation and Electronic Engineering from the University of Catania, Italy, in 2001 and 2005, respectively. He is currently a Research Fellow in Automatic Control Systems at the University of Messina, Italy. His research activities focus on nonlinear systems modeling and control; system identification and soft sensor development; neural networks and learning systems; legged locomotion and insect-inspired control systems; modeling and control of bio-inspired robots. He published more than 130 technical papers and several chapters in books in the field of control and bio-robotics.

Francisco Souza received the B.Sc. degree in Electrical Engineering from the University Federal of Ceara, Brazil and the Ph.D degree in Electrical Engineering from the University of Coimbra, Portugal, in 2014 respectively. He actively works on topics related to industry, and how to optimize it from the data perspective. He has large contribution in the soft sensor area, and also on topics related to applied machine learning, and industrial systems, and industry 4.0. Now, he is a senior researcher at the Radboud University, at Analytical Chemistry and Chemometrics group, where he works on topics related to data science, and sustainability applied to process industry. He is also a member of the European Network for Business and Industrial Statistics.

Maria Gabriella Xibilia received the M.S. degree in Electronic Engineering and the Ph.D. degree in Electrical Engineering from the University of Catania, Italy, in 1991 and 1995, respectively. Since 1998, she has been with the Department of Engineering, University of Messina, Italy, where she is currently an Associate Professor of Automatic Control. She co-authored more than 140 scientific papers and 5 books. Her current research interests include system identification, soft sensors, process control, nonlinear systems, fractional order systems and machine learning.