Continuous Monitoring and Control of a Production Process using Predictive Analytics
Keeping a running production process in a proper working state requires continuous monitoring, anticipating
potential problems in advance, looking for possible solutions to avoid these problems and selecting the most cost-effective
solution among the choices. This work shows a solution approach, to the mentioned requirements using Machine Learning
techniques, composed of three phases: a learning phase, a knowledge generation phase and a monitoring and control phase.
Our approach, which is based on using a well-trained composite Machine Learning model to generate a sufficiently large
database of pre computed prediction values, provides a method to reduce product defects, unplanned downtimes, energy
consumption, CO2 emissions, and to increase the production speed and precision.
Keywords - Casting, Machine Learning, Metal forming, Predictive Analytics.