Exp. 10 - SHION - Smart Thermoplastic Injection
The proposed approach into use cognitive technologies (specially Deep Learning algorithms) to be able to extract the knowledge to generate a predictive model to detect when a defect in the production is going to happen by considering mainly information of the Thermoplastic Injection process itself, with the intrinsic parameters of the injection machine for each piece, as well as context information like environmental conditions, operators review, quality laboratory inspections and piece weight. Constant evolution of the cognitive models generated is also supported.
For this experiment, two Thermolympic production lines are considered for the use case. This approach will enable:
- Preventive detection of anomalies and reducing internal and external non-comfomities
- Predictive maintenance for tools and machines
- Operation parameters optimisation by assisted lesson learnt
The shop-floor workers (production/quality) will be advised of undesirable states (deviation on machine parameters, fail presence). It will reduce the required parts manual inspections and set-up cost while keeping or improving quality rate.
Real time integration of production data (ERP, machine, parts, environment) for feeding back workers using evolutionary AI models.
The first year, data analytics system linked to current machine running in two selected machines will become a new service. Turnover will be an estimated of 50,000€. Two new positions will be required. Develop sensors, scales and hardware sellers are among the ones which will be counted like new contacts.