Machine state forecasting technology coupled with product quality control to unleash productivity for the manufacturing sector (EnCORE)

Descripción del proyecto

Un método para predecir los errores en las máquinas

En la fábrica del futuro será posible predecir cualquier evento y actuar en consecuencia, tanto para máquinas individuales como para sistemas de líneas de fabricación. La clave está en pasar de un mantenimiento programado y regular a otro de tipo preventivo; gracias a esto, las fábricas podrán prever los problemas antes de que ocurran y evitar fallos. El proyecto financiado con fondos europeos EnCORE va a desarrollar un método de mantenimiento predictivo revolucionario. Por ejemplo, usará la tecnología de aprendizaje profundo para poder predecir el estado futuro de una máquina empleando datos sobre los estados normales de la máquina. Los investigadores están intentando comercializar esta solución. La validez del «software» se ha comprobado en dos aplicaciones: una máquina de moldeado por compresión que produce cierres para botellas de plástico y una máquina de conformado que produce cuchillas de afeitar.

Objetivo

In the manufacturing sectors, the traditional planned maintenance approach is no longer viable, as it cannot cope with the ever-rising complexity of production systems. This pressing problem hurts industry’s profitability, and unplanned downtime costs industrial manufacturers €43 billion per year. This pressing problem has fuelled the growth of the predictive maintenance market. Currently, predictive maintenance solutions employ typical machine learning approaches based on monolithic rule-based predictions and require from the customer labelled data that correspond to defective machine states. This impedes the penetration of predictive maintenance in the industry. EnCORE is the fruit of 5 years of R&D to develop proprietary deep neural networks fit for predictive maintenance applications. Our solution uses best-in-class deep learning technology removing the overheads related with data preparation and enable the prediction of machine’s future condition using data that correspond to normal machine states. This is a game changing approach in the predictive maintenance industry. EnCORE is at TRL-6, with our software being validated at two different applications, (1) a compression moulding machine that produces plastic bottle enclosures/caps and (2) a cold forming machine that produces razor blades. Our target market will be the Food & Beverage and Consumer Goods industries targeting both OEMs of machinery and End-Users use such machinery. To take our product to the market, we will employ an hybrid business model using both direct sales and sales through industrial IoT platforms. EnCORE’s unique offering unlocks tremendous value for our customers; this will fuel the adoption of our solution by the industry. In the commercialisation period, we forecast cumulative profits of about €15 million with a strong Return on Investment (ROI) of €13 million. This will allow us to grow our workforce by 83 new employees, to meet the expected market demand for our breakthrough product.