Experiment 20: INTEGRABLE – MonItoring And CoNTrol For REconfiGuRABLe PrEsses

Motivation of the Experiment

Digital technologies can significantly support EU industries facing the competitiveness of the modern globalized market. In this view, INTEGRABLE aims to develop real-time monitoring systems and AI models to detect production anomalies and support operators in corrective action implementation.
More specifically, the experiment focuses on one of the two fully automatic transfer lines installed at MOLLIS ANTONIO S.r.l., controlled by SCAMM. The line, consisting of five 300-ton hydraulic presses in series, produces different components of household appliances taking as inputs the metal sheets, the process parameters, the instructions for the specific parts, and the dedicated dies.

The set-up may require variable time, and periodic adjustments of positions, alignments, lubrification and dies are necessary each time a quality defect in the manufactured parts is identified. Two main criticalities have been recognized in this process. First, quality inspection is currently performed visually by dedicated operators implying high inefficiencies, delays in defects recognition, and, thus, unintended production of defective parts. Second, if quality defects are identified, operators should perform corrective actions currently relying only on their personal experience.
Therefore, for SCAMM, INTEGRABLE is the first step of a development plan aimed at strategically repositioning the company in the current market by improving the reliability and efficiency of its processes. Moreover, for IMECH, INTEGRABLE represents a valuable use case to demonstrate the advantages connected with digital twinning technologies, boosting the advancement of the technological level of the private Italian industries with an increased consciousness of the usage of data.

Purpose of the Experiment

INTEGRABLE aims at improving the manufacturing reliability and efficiency of the described process through image acquisition of the manufactured parts, image analysis for quality evaluation, and anomaly detection.
In this view, the first achievement involves identifying the main bottlenecks and formalizing corporate knowledge. In this specific case, this translates into the analysis of the quality control process and the definition of a list of quality defects, based on dimensional and aesthetic requirements, that should trigger alarms and stop production.
Following, suitable cameras and algorithms can be identified, relying on the preliminary analysis outcomes. Thus, the selected monitoring system will be installed, enabling the assessment of process quality and the objective determination of any deviations. Then, data will be aggregated and historicized in the company’s information system.
Raw images will constitute the input of the first AI model, a computer vision model, typically a Convolutional Neural Network, which will provide, as a result, a list of possible defective areas. Finally, the interaction with the user will provide the system with information on possible false positives/negatives identified by the vision model, enabling the configuration of a continuous learning approach.

Technical Impact

Real-time monitoring systems can provide significant advantages in the manufacturing sector, given the ever-increasing need for continuous control to ensure product quality and production efficiency. In this view, INTEGRABLE will implement a computer vision model supporting the operator in digitalizing the quality defects recognition process, which has been recognized as the main bottleneck in the SCAMM process.
INTEGRABLE will represent a virtuous case study to demonstrate the advantages connected with this technology, which will: enable definition, visualization and monitoring of the trend of the KPIs of interest; exploit built-in intelligence to optimize implant availability and reduce the production of defective parts; improve performance while reducing waste and energy consumption; enable scheduling maintenance work on machines according to actual need (preventive maintenance); improve the working condition of operators reducing stressful situations.
This experiment is of strategic significance also for the technological partner IMECH. Indeed, data valorization for process optimization is one of the main topics of interest for IMECH industrial partners, constituting one of the crucial pillars to maintain competitiveness in the following years. Thus, to foster the application of this technology in different contexts, additional effort will be devoted to the standardization of the proposed solution, ensuring the tool's scalability and flexibility in order to facilitate the subsequent extension.

Expected Economic Impact

For SCAMM, the introduction of the INTEGRABLE technology will provide an efficient and reliable system to timey detect quality defects, thus, reducing the production of defective parts and the associated waste of material and energy. In this regard, it is significant to highlight that the line is currently used on 3 shifts. The foreseen improvement are expected to increase the process efficiency allowing SCAMM to use it only on 2 shifts. Moreover, the support tool can be integrated also on production lines sold to third parties, thus, increasing SCAMM market competitiveness, offering after-sales services and establishing a business associated with innovative control tools.
Concerning the technological partner, IMECH will exploit the acquired knowledge and technical expertise to bring innovation into the companies it is consulting with as well as external to the consortium, supporting the improvement of the technological level of the private Italian industries, with an increased consciousness in the usage of data. IMECH aims at deepening its knowledge in the field, also introducing additional human resources devoted to this topic. In this view, IMECH has been planning to provide funding for doctoral fellowships on this innovative subject.
Finally, participation in a relevant European project allows IMECH and SCAMM to establish a trusted network with research and industrial entities, opening the way to new and valuable collaborations. 

Project Partners: