Lines of Research
Monitoring and Real-time Control
Although plant instrumentation is already a reality in several Brazilian factories, Artificial Intelligence will have a disruptive role in monitoring and controlling industrial environments. This line aims to explore the state of the art of AI in equipment with embedded AI, as well as to implement decision-making within industrial processes, monitoring strategies, and real-time data processing to provide multi-parametric approaches to industrial equipment and machinery, and to develop a correlation of data obtained through data-driven collection and analysis. In this aspect, the results tend to reflect the current condition of the machines and indicate changes in operational conditions that may imply calibration problems, failures, and decreased performance, for example. The analyses can take into account intelligent measurements, signals, vibration variations, temperature, current, and images (HU et al., 2020).
Regarding the scope of action of this line, some themes will be the initial basis of the studies, along with techniques used to overcome the challenges proposed by industry representatives. The preliminary scope includes:
-
Metrology 4.0: This theme is important for the reliability of digitally captured data from physical or manufacturing environments through measurements and monitoring by sensors, meters, and equipment. The current metrological process of calibrating these elements becomes an operational challenge as it often requires production interruptions. Calibrating multiple stages of a real-time manufacturing process presents a highly complex challenge to be overcome by applying AI, creating an environment where uncontrolled conditions of processes, such as dynamic phenomena involving synchronization, response time, and processing speed, can lead to incorrect assessments of the obtained results (EURAMET, 2020);
-
Extended Reality (ER): Extended Reality offers a new perspective for real-time visualization and monitoring of machine and equipment information, as well as operational data from the entire production line. It aids in simulation, assistance, and guidance, enhancing the operator's perception by utilizing artificially inserted information into the real environment. It can also be present in industrial process automation and in the development of products and manufacturing processes (ALCÁCER; CRUZ-MACHADO, 2019; ELSTNER, 2020);
-
Industrial Internet of Things (IIoT): With the scenario and advancement of IoT technologies, new ways of collecting information from the production environment pose new challenges and monitoring and control needs, enabling a strategy for using embedded AI in sensor devices or edge equipment that can be integrated into the context of automation technologies or information technologies in industrial environments. This also calls for the need of a specific metrological framework for IIoT;
-
Pattern recognition and fault diagnosis: Automatic fault diagnosis and pattern recognition are some of the key needs in the industry, as an efficient diagnostic system can reduce costs, enhance safety, and improve process performance in industries (GHOLAMINEJAD; POSHTAN, 2017). New methods of Artificial Intelligence, such as machine learning and advanced image processing techniques (such as convolutional networks), as well as signal processing solutions (like cloud-based and predictive modeling), have shown promising results in novel ways of evaluating faults in the production line and recognizing patterns, including visual inspection through computer vision, signal processing, and image analysis (GHOLAMINEJAD; POSHTAN, 2017; LIANG et al., 2019);
-
Machine performance and optimization: The performance and production reconfiguration based on parameters obtained through real-time monitoring rely on the application of AI to achieve better results, whether for capacity control or intelligent and automatic reorganization of the production line's operational actions;
-
Development of specialized sensors for use in AI: To enable real-time monitoring and control in all the aforementioned areas, it is important to have the capability of developing sensors that can cater to specific monitoring demands. It is not always possible to find off-the-shelf sensors that meet the required specifications for the projects that will be developed in this proposal. Leo et al. (2016) and Abot et al. (2018) provide examples of such sensors.
Regarding the research challenges of utilizing AI and the key pillars of Industry 4.0 in real-time monitoring and control, several disruptive and incremental scenarios are identified, bringing new perspectives for innovation. The digitization of real-time data, virtualization, and conversion into cost and time reduction in manufacturing processes contribute to the establishment of industrial AI ecosystems involving different sectors. This involves addressing the needs and attributes of intelligent systems, data quality issues, and transformative system methodologies.
The objectives of this project aim to have impacts in several areas: a) Development of an autonomous factory with real-time AI-based control and monitoring, indicating calibration issues, failures, and operational safety concerns; b) Increase in productivity with reduced idle time and, consequently, lower costs, improved safety, and performance through preventive and/or corrective maintenance; c) Assisting in rapid diagnosis, productivity improvement, market competitiveness, and intelligent automation of industrial processes through computer vision, image processing, extended reality, and AI; d) Application of techniques for real-time product quality control integrated into the manufacturing process (GAMER et al., 2020).
References
