Descripción del proyecto
Inteligencia artificial para un mejor control de la producción
La digitalización, la electrónica inteligente y los sensores han aportado enormes beneficios a los procesos de producción mediante el control y la supervisión de etapas de toda la producción. Sin embargo, las opciones actuales como los modelos de aprendizaje automático, el diseño de redes neuronales o el tratamiento manual de datos son incapaces de gestionar la cantidad de datos y no ofrecen modelos o técnicas de toma de decisiones eficaces. El proyecto Pythia, financiado con fondos europeos, tiene como objetivo desarrollar y difundir el uso de un sistema de inteligencia artificial que utiliza algoritmos de aprendizaje profundo muy estudiados que pueden organizar y depurar datos para proporcionar modelos y estrategias mejores. Este sistema, Pythia, ya probado en la industria papelera, puede detectar errores en la producción con hasta treinta minutos de antelación, a la vez que ayuda a comprobar la eficacia de las técnicas.
Objetivo
With the digitalization of the industry, smart electronics and sensors that control and supervise every stage of the paper production process are available. However, current solutions to process the data such as manual data processing, neural networks design or other machine learning models cannot keep up with the amount of gathered data, providing inefficient decision-making models that ultimately do not solve the problem completely. PerfectPattern has developed, Pythia, An Artificial Intelligence (AI) solution based on deep learning algorithms, capable of autonomously gathering, cleaning and analysing data within seconds. Pythia is able to forecast error behavior of a highly complex production system around 30 minutes ahead, with a model that takes under a second to compute data. Thanks to this long pre-warning time, it is possible to build smart strategies to prevent the error. The system even allows to check with Pythia if the strategy will effectively avoid the future error. We not only deploy the leading edge research in AI from Mathematics, we understand key influences coming from leading edge science in Physics (string theory, quantum field theory). Therefore Pythia’s AI is a leapfrog in technology. Pythia has already reported economic benefits to the paper production industry. But the potential of Pythia goes beyond the paper industry. With its unsupervised data analytics and anomaly detection capabilities, Pythia can be expanded as a prediction and production control system for many other industrial applications. With the PoC of Pythia giving successful initial results; there are several big decisions to make before we can continue with the project. We are seeking now to conduct a feasibility study in order to direct our development and commercial efforts.
Ámbito científico
- engineering and technologymechanical engineeringmanufacturing engineering
- engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringcontrol systems
- engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringsensors
- engineering and technologymaterials engineeringwoodworking
- natural sciencesmathematicsapplied mathematicsmathematical model
Programa(s)
Convocatoria de propuestas
Consulte otros proyectos de esta convocatoriaConvocatoria de subcontratación
H2020-SMEInst-2018-2020-1
Régimen de financiación
SME-1 - SME instrument phase 1Coordinador
81379 MUNCHEN
Alemania
Organización definida por ella misma como pequeña y mediana empresa (pyme) en el momento de la firma del acuerdo de subvención.